TWI397901B - Method for controlling a particular loudness characteristic of an audio signal, and apparatus and computer program associated therewith - Google Patents

Description

Method for controlling audio signal specific loudness characteristics and related devices and computer programs Field of invention

The present invention relates to audio signal processing. In particular, the present invention relates to the measurement and control of perceived acoustics and/or perceived spectral balance of an audio signal. The present invention can be used, for example, in one or more of an audio replay environment: loudness compensated volume control, automatic gain control, dynamic range control (which includes, for example, limiters, compressors, expanders, etc.), Dynamic equalization, and background noise interference compensation. In addition to the method, the present invention also includes corresponding computer programs and devices.

Background of the invention

There have been many attempts to produce a satisfactory objective measure of loudness. In 1933, Fletcher and Munson determined that human hearing was less sensitive to low and high frequencies than to intermediate (or speech) frequencies. They also found that the sensitivity is relatively reduced as the sound level increases. An early loudness metric was constructed with a combination of microphones, amplifiers, gauges, and filters, and was designed to roughly mimic the frequency response of hearing low, medium, and high sound levels.

While this device provides a single loudness, constant level, and isolated tone measurement, more complex sound measurements do not properly match the subjective loudness perception. This type of sound level gauge has been standardized but is only used for specific applications, such as monitoring and control of industrial noise.

In the early 1950s, Zwicker and Stevens, among others, extended the research of Fletcher and Munson and produced a more realistic loudness-aware processor pattern. . Stevens published a "complex noise loudness calculation" method in the journal of the American Acoustics Association in 1956, and Zwicker wrote his "Psychology and Methodological Components" Published in the journal Acoustica in 1958. In 1959, Zwicker published a graphical representation of loudness calculations, followed by many similar articles. The methods of Stevens and Zwicker were standardized as ISO 532, Part A and Part B, respectively. Both methods contain similar steps.

First, the time-distributed energy along the basement membrane of the inner ear, referred to as excitation, is modeled by passing the audio transmission through a bandpass auditory filter clustered with one of the center frequencies evenly distributed at the critical band rate scale. Each of the auditory filters is designed to simulate a frequency response at a particular location along the inner ear basement membrane, where the center frequency of the filter corresponds to that location. The critical band width is defined as the bandwidth of one such filter. Taking the Hertz unit as a measure, the critical band width of these auditory filters increases with the center frequency. Thus a set of distorted frequency scales can be defined such that the critical bandwidth of all auditory filters measured according to this skew scale is constant. Such a tortuous scale is known as the critical band rate scale and is very useful for understanding and simulating a wide range of psychoacoustic phenomena. See, for example, Psycho-Acoustics - Facts and Models, published by E. Zwicker and H. Fastl, Springer-Verlag, Berlin, 1990. Stevens and Zwicker use a method called the critical band rate scale of the Bark scale, where the critical band width is constant below 500 Hz and increases above 500 Hz. Recently, Moore and Glasberg defined a critical band rate scale called the equivalent rectangular bandwidth (ERB) scale (BCJ Moore, B. Glasberg, T. Baer, "predictive mode for threshold, loudness, and partial loudness". , published in the April 1997 issue of the Journal of the Journal of Audio Engineering, Vol. 145, No. 4, pp. 224-240. Using psychoacoustic experiments using dynamic noise masks, Moore and Glasberg show that the critical band width continues to decrease below 500 Hz, while the critical band width in the Bark scale remains relatively unchanged.

The excitation calculation is followed by a non-linear compressive function that produces a quantity called "specific loudness." The specific loudness is a measure of the perceived loudness of the frequency and time functions and can be used as a unit of measurement along the critical band rate scale, depending on the perceived loudness per unit frequency, such as the Bark or ERB scale discussed above. Finally, the time-varying "total loudness" will be calculated by integrating a specific loudness in the frequency domain. When a particular loudness is evaluated based on a set of defined auditory filters that are evenly distributed along a critical band rate scale, the total loudness will simply be calculated using the particular loudness additions from each filter.

The loudness can be measured by the phon. The acoustic level measured at the mouth is a sound pressure level (SPL) with a subjective loudness equal to the 1 kHz tone of the sound. Conventionally, the reference 0 dB of the SPL is the root mean square pressure of 2x10 ^{- 5} (pascal) Pascal (pressure unit), and therefore, this is also the reference 0 port. Using this definition, the comparison of the frequency below 1 kHz and having a pitch loudness of 1 kHz provides a contour for the equal loudness of the given mouth orientation. Figure 11 shows the frequency of the equal loudness contour between 20 Hz and 12.5 kHz, and at the 4.2 square (considered hearing tolerance) and 120 square (ISO 226:1087 (E), "hearing normal equal loudness level The position of the mouth between the contour lines is). The oral measurement takes into account the sensitivity of human hearing as a function of frequency, but the results do not allow for the assessment of the relative subjective loudness of a sound with varying levels, as no attempt is made to correct the loudness as the SPL grows nonlinearly, That is, the interval of the contour lines changes.

Loudness can also be measured in units of "song (- loudness unit)". There is a one-to-one mapping between the oral unit and the Song unit, as indicated in Figure 11. A Song is defined as the loudness of a set of 40dB (SPL) 1kHz pure sinusoidal waveforms and is equal to 40 squares. The Song unit is defined as twice the increase in Song corresponding to a twofold increase in perceived loudness. For example, 4 Song is perceived as twice as loud as 2 Songs. Therefore, the loudness level will be expressed in Song and will provide more information. The specific loudness is defined as the measure of the perceived loudness of the frequency and time functions, and the specific loudness can be measured in units of Song per unit frequency. Therefore, when using the Bark scale, the specific loudness is in units of Song per Bark, and similarly when using the ERB scale, it is in units of Song per ERB.

As noted above, human ear sensitivity varies with frequency and level, as is noted in the psychoacoustic literature. As a result, the perceived spectrum or timbre of a given set of sounds varies with the acoustic level at which the sound is heard. For example, a group of sounds containing low, medium, and high frequencies, the perceived relative proportion of such frequency components changes with the overall loudness of the sound; in the case of smaller sounds, the sounds of the low and high frequency components are larger than the medium frequency. The sound is louder. This phenomenon is conventional and is alleviated in sound reproduction equipment using so-called loudness control. Loudness control is a volume control that applies a low frequency when the volume is turned down and sometimes also applies a high frequency pull. Therefore, the lower sensitivity of the human ear at extreme frequencies is compensated by the artificial pull-up of those frequencies. The completion of such control is completely static; the degree of compensation applied is a set of functions for volume control settings or some user operation control, rather than a function of the content of the audio signal.

In fact, the perceived relative spectral balance change between low, medium and high frequencies depends on the signal, especially depending on the actual spectrum of the signal and whether the signal should be loud or whisper. Consider the recording of the symphony orchestra. The balance across the spectrum may be correct with the same level of information heard by an audience in the concert hall, whether the symphony orchestra is loud or whispered. For example, if the music is reproduced with a 10 dB smaller sound, the perceived balance across the spectrum changes in one way over a larger segment and in another in a smaller segment. The static loudness control that is conventionally seen does not impose different compensations as a function of music. In the international patent application serial number PCT/US 2004/016964, the file was filed on May 27, 2004, and was published on December 23, 2004 in WO 2004/111994 A2, disclosed by Seefeld et al. A system that adjusts the perceived loudness of a set of audio signals. This PCT application, which is assigned to the United States, is hereby incorporated by reference in its entirety. In this application, a psychoacoustic mode calculates a set of audio signal loudness in units of perception. In addition, the application introduces a technique for calculating a broadband multiplicative gain that, when applied to an audio, produces a set of gains that are substantially the same as the standard loudness - the loudness of the modified audio. However, the application of this wideband gain changes the perceived spectral balance of the audio.

Summary of invention

In one aspect of the present invention, the present invention provides an information method for deriving a particular loudness that can be used to control an audio signal by modifying the audio signal to reduce the difference between its particular loudness and target specific loudness. The specific loudness is a measure of the perceived loudness of the frequency and time functions. In actual production, the specific loudness of the modified audio signal can be processed to approximate the target specific loudness. The approximation is not only affected by general signal processing considerations, but also by the time and/or frequency smoothing that can be used for modification, as explained below.

Since the specific loudness is a measure of the perceived loudness of the audio signal as a function of frequency and time, in order to reduce the difference between the specific loudness of the audio signal and the target specific loudness, the audio signal can be modified as a function of frequency. While in some cases the target specific loudness may be time-invariant and the audio signal itself may be a set of steady-state time-invariant signals, in general, the modification may also modify the audio signal as a function of time.

The arguments of the present invention can also be employed to compensate for background noise interference in an audio replay environment. When the audio is heard in the background noise, the noise may partially or completely mask the audio in a manner that depends on the audio level and the spectrum, as well as the noise level and spectrum. This produces a change in the audio-aware spectrum. Based on psychoacoustic research (please refer, for example, Moore, Glasberg, and Baer, “Predicting Modes for Threshold, Loudness, and Partial Loudness”, April 1997, Journal of the Audio Engineering Society, Volume 45, No. 4) The "partial specific loudness" of the audio can be defined as the perceived loudness of the audio in the presence of the second interfering sound signal, for example, noise.

Accordingly, in another aspect of the present invention, the present invention provides for utilizing a modified audio signal to reduce a difference between a portion of its specific loudness and a target specific loudness to derive a portion of the specific loudness information that can be used to control the audio signal. Methods. Therefore, in a perceptually accurate way to mitigate the effects of noise. In this and other considerations of the interference noise signal of the present invention, it will be assumed that the audio signal can be individually touched and the second interference signal can be individually accessed.

In another aspect, the present invention provides a method of controlling the specific loudness of an audio signal by modifying the audio signal to reduce the difference between its particular loudness and a target specific loudness.

In another aspect, the present invention provides a method for modifying a portion of a particular loudness of an audio signal by modifying the audio signal to reduce the difference between a portion of its specific loudness and a target specific loudness.

When the target specific loudness is not a function of the audio signal, it may be the target specific loudness stored or received. When the target specific loudness is not a function of the audio signal, the modification or derivation may explicitly or implicitly calculate a particular loudness or a portion of a particular loudness. An implicitly calculated example includes a look-up table or a "closed form" mathematical expression in which a particular loudness and/or a portion of a particular loudness is inherently determined (a closed form describes a mathematical representation that is available for limited use) The number of standard mathematical operations and functions, such as exponents and cosines, are accurately represented). While the target specific loudness is not a function of the audio signal, the target specific loudness may be time-invariant and non-frequency-variant or it may only be non-frequency varying.

In another aspect, the present invention provides a process for controlling the measurement of an audio signal or an audio signal by a parameter in accordance with one or more processing programs and one or more processing programs to produce a target specific loudness audio signal. While the target specific loudness may be time-invariant ("fixed"), the target specific loudness may advantageously be a function of a particular loudness of the audio signal. Although the target specific loudness may be a static, non-frequency-changing and non-time-varying signal, in general, the audio signal itself is frequency-variant and time-varying, so that when it is a set of functions of the audio signal, the target specific loudness is Frequency change and time change.

The representation of the audio and target specific loudness or target specific loudness may be from a copy of the transmission or storage medium.

The representation of the target specific loudness may be one or more scale adjustments that adjust the measurement of the audio signal or the audio signal.

The target specific loudness of any of the above-discussed aspects of the present invention may be the measurement function of the audio signal or the audio signal. The specific loudness of an audio signal is an appropriate measure of the audio signal. The function of the audio signal or the measurement of the audio signal may be a scale adjustment of the audio signal or the measurement of the audio signal. For example, the scale adjustment can be a combination of scale adjustments or multiple scale adjustments: (a) a time and frequency-change scale factor Ξ[b,t] achieves a specific loudness scale adjustment with the following relationship (b) A time-varying, non-frequency-varying scale factor Φ[t] achieves a specific loudness scale adjustment with the following relationship (c) A non-time-varying, frequency-varying scale factor achieves a certain degree of loudness adjustment in the following relationship; And (d) a non-time-varying, non-frequency-changing, scale factor Θ[b] achieves a scale adjustment of the specific loudness of the audio signal in the following relationship among them [ b , t ] is the target specific loudness, N [ b , t ] is the specific loudness of the audio signal, b is the frequency measurement, and t is the time measurement.

In case (a) of the time and frequency-change scale factor, the scale adjustment can be determined, at least in part, by the ratio of the desired multi-band loudness and the multi-band loudness of the audio signal. Scale adjustments of this type can be used as a dynamic range control. Further details of the use of the present invention as a dynamic range control argument are set forth below.

At the same time, in the case of time and frequency-variation scale coefficients (a), the specific loudness can be scaled according to the ratio of the measurement of the desired spectral shape to the measurement of the spectral shape of the audio signal. Such scale adjustments can be employed to convert the perceived spectrum of the audio signal from a time varying perceptual spectrum to a substantially non-time varying perceptual spectrum. Such scale adjustments can be used as a set of dynamic equalizers when the specific loudness is scaled by the ratio of the measured spectral shape and the measured spectral shape of the audio signal. The arguments for using the present invention as a dynamic equalizer will be further described in detail below.

In the case of time-varying, non-frequency-varying scale coefficients (b), the scale adjustment can be determined at least in part according to the ratio of the desired wide-band loudness and the wide-band loudness of the audio signal. Such scale adjustments can be used as an automatic gain control or dynamic range control. The argument that the invention is used as an automatic gain control or a dynamic range control will be further described below.

In case (a) (time frequency - variation scale factor) or case (b) (time varying, non-frequency variable scale factor), the scale factor may be a function of the measurement of the audio signal or the audio signal.

In the case of non-time varying, frequency varying scale coefficients or in the case of non-time varying, non-frequency varying, scale coefficients (d), the modification or derivation may include storing scale coefficients or the scale coefficients may be received from an external source.

In either case of case (c) or case (d), the scale factor may not be a function of the measurement of the audio signal or the audio signal.

Any of the arguments of the present invention, as well as variations, modifications, derivations, or generations thereof, may be individually, explicitly, or implicitly calculated as (1) a particular loudness, and/or (2) a portion of a particular loudness, and/or (3) Target specific loudness. Implicit calculations may include, for example, a look-up table or a closed form mathematical representation.

Modifying parameters can be smoothed in time. The modified parameters may be, for example, (1) a majority of amplitude scale adjustments with respect to the frequency band of the audio signal or (2) control of one or more filters, such as a plurality of filters of a multi-branch FIR filter or a multi-pole IIR filter. coefficient. The scale adjustments or filter coefficients (and the filters to which they are applied) can be time-varying.

In calculating a function of a particular loudness of an audio signal defining a target specific loudness or an inverse function of the function, the procedure for performing such calculations is operated in the field of perception (perceptual psychoacoustic) loudness, where the input and output of the calculation For a specific loudness. In contrast, when applying amplitude scale adjustment to the audio signal band or applying filter coefficients to the controllable filtering of the audio signal, the modified parameter operates outside of the perceptual (psychoacoustic) loudness domain and can be classified The audio signal is modified in the field of electrical signals. Although the audio signal modification of the audio signal can be performed in the field of electrical signals, this change in the field of electrical signals is derived from calculations in the field of perceptual (psychoacoustic) loudness so that the modified audio signal has an approximate desired The specific loudness of the target specific loudness.

Deriving the modified parameters from the calculations in the loudness field provides more control over the perceived loudness and perceived spectral balance than the derivation of these modified parameters from the electrical signal domain. Furthermore, the use of a basement membrane to simulate a psychoacoustic filterbank or its equivalent to perform loudness domain calculations provides more detailed control of the perceptual spectrum than a configuration that derives modified parameters from the electrical signal domain.

Various modifications, derivations, and generations of approximating the specific loudness of an unmodified audio signal derived from one or more interfering audio signals, a target specific loudness, a particular loudness from a modified audio signal, or a portion of a particular loudness Degree, the specific loudness of the unmodified audio signal, and an approximation of the target specific loudness derived from the specific loudness or partial specific loudness of the modified audio signal.

The modification or derivation may derive modification parameters at least in part from one or more of: interfering audio signal measurements, target specific loudness, unrecognized audio from a particular loudness or partial specific loudness of the modified audio signal The signal specific loudness evaluation value, the specific loudness of the unmodified audio signal, and the target specific loudness approximation derived from the specific loudness or part of the specific loudness of the modified audio signal.

In particular, the modification or derivation may derive the modification parameters at least in part from: (1) one of the following: a target specific loudness, and a particular one of the unmodified audio signal received from the particular loudness of the modified audio signal. An approximation of the loudness, and (2) a specific loudness of the unmodified audio signal, and a target specific loudness approximation derived from the specific loudness of the modified audio signal, or when the interfering audio signal is to be considered The modification or derivation may derive modified parameters at least in part from (1) one of the interfering audio signals, (2) one of the following, a target specific loudness, and a portion from the modified audio signal. An approximation of a particular loudness of an unmodified audio signal derived from a particular loudness, and (3) a specific loudness of one of the unmodified audio signals, and a set of target specificities derived from a particular loudness of the modified audio signal The loudness is approximate.

A set of preamble configurations can be employed, wherein the particular loudness is derived from an audio signal and wherein the target specific loudness is from an external source of the method or from a store when the modification or derivation includes storing a set of target specific loudness. Additionally, a set of hybrid pre-receiving/responsible configurations can be employed in which an approximation of the target specific loudness is derived from the modified audio signal, and wherein when the modification or derivation includes storing a set of target specific loudness, the target specific loudness is Sources from outside the method or from a set of storage.

The modification or derivation may include one or more processing procedures to obtain the target specific loudness, either explicitly or implicitly, and one or more of the functions of the audio signal or the measurement of the audio signal are explicitly or implicitly calculated. . In addition, a set of preamble configurations can be employed, wherein the specific loudness and the target specific loudness are derived from the audio signal, and the derivation of the target specific loudness is a function of the measurement of the audio signal or the audio signal. In another option, a set of hybrid pre-receiving/responsible configurations can be employed in which the approximation of the target specific loudness is derived from the modified audio signal, and the target specific loudness is derived from the audio signal, and the derivation of the target specific loudness is employed. The function of the audio signal or the measurement of the audio signal.

The modification or derivation may include one or more processing procedures to obtain, explicitly or implicitly, an evaluation value of a particular loudness of the unmodified audio signal that is reflected in the modified audio signal, one or more of which are explicitly or implicitly Inversely calculating an inverse function of the function of the audio signal or the measurement of the audio signal. In addition, a set of feedback configurations is adopted, wherein the evaluation value of the specific loudness of the unmodified audio signal and the approximation of the target specific loudness are derived from the modified audio signal, and the evaluation value of the specific loudness uses the audio signal or the audio signal. The inverse function of the measured function is calculated. In another option, a set of hybrid pre-receiving/responsible configurations are employed in which a particular loudness is derived from the audio signal and an evaluation of the particular loudness of the unmodified audio signal is derived from the modified audio signal, the audio signal or The inverse of the function of the measurement of the audio signal is used to calculate the derivation of the evaluation value.

Modification parameters can be applied to the audio signal to produce a modified set of audio signals.

Another argument of the invention is that the processing program or device may have temporal and/or spatial separation and, therefore, actually have an encoder or encoding program and also have a decoder or decoding program. For example, there may be an encoding/decoding system in which the modification or derivation may transmit and receive or store and at the same time also copy the audio signal and either (1) the modified parameter or (2) the target specific loudness or the target specific loudness representation. Additionally, in practice, there may be only one set of encoders or encoding programs with a set of audio signals and (1) modified parameters or (2) transmission or storage of representations of target specific loudness or target specific loudness. Additionally, in practice, there may be only one set of decoders or decoding programs having a set of audio signals and (1) modified parameters or (2) transmission or storage of representations of target specific loudness or target specific loudness.

Simple illustration

Figure 1 shows a set of functional block diagrams of a pre-production paradigm in accordance with one aspect of the present invention.

Figure 2 shows a set of functional block diagrams of a feedback production paradigm in accordance with one aspect of the present invention.

Figure 3 is a block diagram showing a functional group of a hybrid pre-commissioning/responsibility production example in accordance with one aspect of the present invention.

Figure 4 shows a set of functional block diagrams of another hybrid pre-/re-sales production paradigm in accordance with one aspect of the present invention.

Figure 5 is a set of functional block diagrams showing the manner in which unmodified audio signals and modified parameters determined by any set of pre-grant, feedback, and hybrid pre-request configurations can be stored or transmitted, such as For use in devices or handlers that are separated in time or space.

Figure 6 is a set of functional block diagrams showing that unmodified audio signals and target specific loudness or their representations determined by any set of pre-grant, feedback, and hybrid pre-requested configurations can be stored or transmitted. The manner, for example, is for use in a device or process that is separated in time or space.

Figure 7 is a block diagram showing an exploded view or an exploded flow chart of an overview of one of the arguments of the present invention.

Figure 8 is an ideal characteristic response of a linear filter P(z) suitable as a set of transmission filters in an embodiment of the present invention, wherein the vertical axis is decibel attenuation (dB) and the horizontal axis is in Hertz (Hz) The logarithmic base of the unit is 10 frequencies.

Figure 9 shows the relationship between the ERB frequency scale (vertical axis) and the frequency in Hertz (horizontal axis).

Figure 10 shows a set of idealized auditory filter characteristic responses that approximate a critical band on the ERB scale. The horizontal scale is the frequency in Hertz and the vertical scale is the decibel level.

Figure 11 shows the equal loudness contour of ISO 226. The horizontal scale is the frequency in Hertz (the logarithmic base 10 scale) and the vertical scale is the sound pressure level in decibels.

Figure 12 shows the equal loudness contour of ISO 226 normalized by the transmission filter P(z). The horizontal scale is the frequency in Hertz (the logarithmic base 10 scale) and the vertical scale is the sound pressure level in decibels.

Figure 13a shows an ideal plot of wideband and multiband gain with a loudness scale adjusted to 0.25 for a segment of female speech. The horizontal scale is the ERB band and the vertical scale is the relative gain in decibels (dB).

Figure 13b shows an ideal plot of the original signal, the wideband gain modified signal, and the specific loudness of the multiband gain modified signal, respectively. The horizontal scale is the ERB band and the vertical scale is the specific loudness (Song/ERB).

FIG 14a is a set of ideal view showing: Usually when AGC, L _o [t] to L _i [t] is a set of functions. The horizontal scale is log( L _i [ t ]) and the vertical scale is log ( L _o [ t ]).

FIG 14b is a set of idealized view showing: Usually when DRC, L _o [t] to L _i [t] is a set of functions. The horizontal scale is log( L _i [ t ]) and the vertical scale is log ( L _o [ t ]).

Figure 15 shows an ideal plot of the general band smoothing function for multi-band DRC. The horizontal scale is the number of bands and the vertical scale is the gain output of band b.

Figure 16 is a block diagram showing an exploded view or an exploded flow chart of an overview of one of the arguments of the present invention.

Figure 17 is similar to Figure 1 which also contains a decomposition function block diagram or decomposition flow diagram of the noise compensation in the replay environment.

Best mode for carrying out the invention

Figures 1 through 4 show functional block diagrams of possible pre-delivery, feedback, and two-group hybrid pre-review/recall modes in accordance with the teachings of the present invention.

Referring to the topology example prior to Figure 1, an audio signal is applied to two sets of paths: (1) a processor or device 2 ("Modify Audio Signal") that has the ability to modify the parameters to modify the audio. The set of signal routes, and (2) a set of control routes having a handler or device 4 ("generating modified parameters") capable of generating such modified parameters. The modified audio signal 2 in the first embodiment of the present invention and the modified audio signal 2 in each of the examples in FIGS. 2-4 may be devices or processing programs for modifying the audio signal, for example, based on the generated modified parameter 4 (or corresponding program or separate The modified parameters M of the devices 4', 4" and 4"') from the respective examples of Figures 2-4 are modified in frequency and/or time varying manner to produce modified parameters 4 and The counterparts in Figures 2-4 operate, at least in part, in the field of perceived loudness. As in the various examples of Figures 1-4, the modified audio signal 2 operates in the electrical signal domain and produces a set of modified audio signals. As also in the various examples of Figures 1-4, the modified audio signal 2 and the resulting modified parameter 4 (or their counterparts) modify a set of audio signals to reduce the difference between their particular loudness and a set of target specific loudnesses.

In the first example of Figure 1, the processing program or device 4 may include a number of processing programs and/or devices: a set of "calculation target specific loudness" processing programs or devices 6 that calculate a target specific loudness or measurement that is reflected in the audio signal. An audio signal, for example, a specific loudness of an audio signal, a set of "calculated specific loudness" processing programs or means 8 that compute a particular loudness of an audio signal that is responsive to an audio signal or a measure of an audio signal such as its excitation, and a set of " The Calculate Modification Parameters procedure or device 10 calculates a modified parameter that is responsive to a particular loudness and target specific loudness. The calculation target specific loudness 6 may perform one or more sets of function "F" each having a function parameter. For example, it can calculate a particular loudness of the audio signal and then apply one or more sets of functions F to the audio signal to provide a target specific loudness. This is decomposed in the first figure as the input of the "select function F and function parameters" handler or device 6. Instead of being calculated by the device or process 6, the target specific loudness may be generated by a stored processing program or device included in or associated with the modified parameter 4 (decomposed as a "storage" input to the processing program or device 10), Or a source external to all of the handlers or devices (deprecated as an "external" input to the handler or device 10). Thus, the modified parameters are calculated, at least in part, in the field of perceptual (psychophonic) loudness (ie, at least a particular loudness and, in some cases, a target specific loudness calculation).

Processed programs or devices 6, 8 and 10 (and the processing programs or devices 12, 14, 10' in the example of Fig. 2, 6, 14, 10 in the example of Fig. 3, and 8 in the example in Fig. 4) The calculations performed by 12, 10''') may be performed explicitly and/or implicitly. Examples implicitly include (1) a set of comparison tables whose items are based in whole or in part on a particular loudness and/or Or target specific loudness and/or modified parameter calculations, and (2) a set of closed form mathematical expressions that are inherently wholly or partially based on a particular loudness and/or target specific loudness and/or modified parameters.

Although the calculation processing program or devices 6, 8 and 10 in the example of Fig. 1 (and the processing programs or devices 12, 14, 10' in the example of Fig. 2, 6, 14, 10 in the example of Fig. 3, and 8, 12, 10'' in the example of Figure 4 are shown exploded and separately, for illustrative purposes only. It is understood that one or all of these processes or devices can be combined a single processing program or device is combined in a plurality of processing programs or devices, for example, as in the configuration of FIG. 9 below, like a group of pre-represented topology structures of the example of FIG. 1, the processing program or device The modified parameters are calculated in response to the smoothed excitation derived from the audio signal and the target specific loudness. In the example of Fig. 9, the means or process for calculating the modified parameters implicitly calculates the specific loudness of the audio signal.

As an argument of the present invention, an example of FIG. 1 and an example of other embodiments of the present invention, target specific loudness ( [ b , t ]) can be calculated using a scale adjustment of one or more scales to adjust the specific loudness ( N [ b , t ]). The scale adjustment can be a set of time-varying and frequency-varying scale coefficients Ξ [b, t], which scale the specific loudness according to the following relationship: A set of time-varying, non-frequency-varying scale coefficients Φ[t] scales the specific loudness according to the following relationship A set of non-time-varying, frequency-varying scale coefficients Θ[ b ] scales the specific loudness according to the following relationship A set of scale factors α of the specific loudness of the audio signal are scaled to a specific loudness according to the following relationship Where b is the frequency measurement (eg, the number of bands) and t is the time measurement (eg, the number of blocks). Multiple scale adjustments can also be employed using a combination of a number of specific scale adjustments and/or specific scale adjustments. An example of multiple scale adjustments is given below. In some cases, as further described below, the scale adjustment can be a function of the measurement of the audio signal or the audio signal. In other cases, it will be further explained below. When the scale adjustment is not a function of the audio signal measurement, the scale adjustment can be determined or supplied in different ways. For example, the user may select or apply a set of time-invariant and non-frequency-varying scale coefficients α or a set of non-time-varying, frequency-scaled scale coefficients Θ[ b ].

Thus, the target specific loudness can be expressed as one or more functions F of the audio signal or the measurement of the audio signal (the specific loudness is one of the audio signals may be measured):

If the function F is reversible, the specific loudness ( N [ b , t ]) of the unmodified audio signal can be calculated as the target specific loudness ( The inverse function F ^{- 1 of} [ b , t ]) :

As will be explained later, the inverse function F ^{- 1} is calculated in the feedback and hybrid pre-review/reward examples in Figures 2 and 4.

A set of "selection function and function parameters" inputs that calculate the target specific loudness 6 are shown to indicate a device or process 6 that can calculate a target specific loudness by applying one or more functions in accordance with one or more function parameters. For example, calculating a target specific loudness 8 may calculate a function "F" of a particular loudness of the audio signal to define a target specific loudness. For example, the "Select Function and Function Parameters" input may select one or more sets of specific functions belonging to one or more of the above categories, and one or more function parameters, such as constants for functions (eg, scaling) .

As described above, the scale factor associated with the scale adjustment can be used as a representation of the target specific loudness, as with the target specific loudness, which can be calculated based on the scale adjustment of the particular loudness. Thus, in the example of Figure 9, illustrated below and as described above, the look-up table may be indexed by the scale factor and the stimulus such that the calculation of the specific loudness and the target specific loudness is inherent within the table.

Whether using a set of look-up tables, a set of closed-form mathematical expressions, or some other technique, the modified parameters 4 (and their corresponding processing programs or the various devices 4', 4" and 4' of the examples of Figures 2-4 are generated. The operation of '') causes the calculation to be achieved in the field of perceptual (psychophone) loudness, even if the specific loudness and the target specific loudness cannot be explicitly calculated. A clear specific loudness or a set of hypothetical, implicitly specific loudness will also Similarly, a set of explicit target-specific loudnesses or a set of hypothetical, implicit target-specific loudnesses will also exist. In any case, the modified parameters are calculated to produce modified audio signals to reduce the specific loudness and Modification parameters for differences between target specific loudnesses.

In a replay environment having a second interfering audio signal (e.g., noise), the modified parameter 10 (and its corresponding processing program or devices 10', 10" and 10' of the respective examples in Figures 2-4, respectively, are calculated. ''), which can also selectively receive the measurement of such second interfering audio signal as an input or a second interfering signal itself as a group input. This selective input is displayed in a dashed line. Figure 1 (and Figures 2-4). The measurement of the second interfering signal may be an excitation thereof, such as the example of Figure 17, which is illustrated below. The measurement of the interfering signal or the signal itself (assuming interference) The signals may be separately processed) applied to the device for modifying the parameter processing program or the device 10 of FIG. 1 (and its corresponding processing program or devices 10', 10" and 10' of the respective examples of Figures 2-4, respectively. '') allows such a program or device that is suitably configured to calculate a set of modified parameters that take into account the interfering signal, as will be further illustrated in the "noise compensation" portion. In the example of Figures 2-4, the calculation of a portion of the specific loudness assumes that the appropriate measurement of the interfering signal is not only applied to the respective computational modification parameters 10', 10", or 10"', but is also applied to one. The set "calculated approximation of the specific loudness of the unmodified audio" handler or device 12 and/or a set of "calculated approximations of the target specific loudness" process or device 14 is used to assist the function or device in calculating a portion of the specific loudness. As in the example given in Figure 1, some of the specific loudness is not explicitly calculated - the calculation modification parameter 10 of Figure 1 calculates the appropriate modification parameters so that the specific loudness of the modified audio approximates the target specific loudness. This is further illustrated in the "Noise Compensation" section.

As described above, in each of the examples of Figures 1-4, the parameter M is modified to be applied to the audio signal by the audio signal modifier 2 to reduce the specific loudness or part of the specific loudness and target of the generated modified audio. The difference between specific loudnesses. Ideally, the specific loudness of the modified audio signal is approximately or the same as the target specific loudness. Modifying the parameter M may, for example, be of a type having a time varying gain coefficient applied to a band derived from a filter bank or a coefficient of a time varying filter. Thus, in all of the examples of Figures 1-4, the modified audio signal 2 can be made, for example, as a plurality of amplitude scalers each operating in a set of frequency bands, or a set of time varying filters (eg, a set Multi-branch FIR filter or a set of multi-pole IIR filters).

Herein, the same reference numerals indicate that the device or processing program can be substantially similar to another or other device or processing program having the same reference number. A reference number with an apostrophe (e.g., "10'") indicates that the device or processing program has a similar structure or function, but may be another or other modification having the same basic reference number or apostrophe.

Under certain limitations, an almost equivalent feedback configuration as exemplified in Figure 1 can be implemented. Figure 2 shows this example where the audio signal is also in the signal path. Applied to the modified audio signal processing program or device 2 processing program or device 2 also receives modified parameters M from a set of control routes, one of which in a set of feedback configurations generates a modified parameter handler or device 4' from the modified audio The output of signal 2 receives the modified audio signal as its input. Thus, in the example of Figure 2, the modified audio, rather than the unmodified audio, is applied to a set of control paths. Modifying the audio signal processing program or device 2 and generating a modified parameter handler or device 4' modifies the audio signal to reduce the difference between its particular loudness and target specific loudness. The processing program or device 4' may include a number of functions and or devices: a set of "calculated approximations of a particular loudness of unmodified audio" processing program or device 12, a "calculated approximation of target specific loudness" processing program or device 14 And a set of "calculate modified parameters" handlers or devices 10' that modify the parameters.

Depending on whether the function F is a reversible limit, the handler or device 12 evaluates the particular loudness of the unmodified audio signal by applying the inverse function F ^{- 1} to the particular loudness or partial specific loudness of the modified audio signal. As described above, the device or handler 12 can calculate a set of inverse functions F ^{- 1} . This is decomposed in Fig. 2 as "select inverse function F ^{- 1} and function parameters" input to the processing program or device 12. The "approximate degree of calculation of the target specific loudness" is operated by calculating the specific loudness or part of the specific loudness of the modified audio signal. This particular loudness or partial specific loudness is a set of approximations of the target specific loudness. The approximation of the specific loudness of the unmodified audio signal and the approximation of the target specific loudness are used by the calculated modified parameter 10' to derive a modified parameter M that can be reduced if the modified audio signal 2 is applied to the audio signal. The difference between the specific loudness or part of the specific loudness and the target specific loudness. As mentioned above, these modified parameters M may, for example, be of a time varying gain having a frequency band applied to the filter bank or a time varying filter. The actual embodiment of the feedback loop in calculating the modified parameter 10" may introduce a delay between the calculation and the application of the modified parameter M.

As described above, in a replay environment having a second interfering audio signal, such as a noise, the modified parameter 10', the specific loudness of the unmodified audio 12, the approximate degree of approximation, and the calculated approximation of the target specific loudness 14 are each Optionally, the measurement of the second interfering audio signal is received as an input or a second interfering signal as a group input thereof and the processing program or device 12 and the processing program or device 14 can each calculate the modified audio signal. Part of the specific loudness. This selection input is shown in Figure 2 as a long fold line.

As described above, a hybrid pre-/return production paradigm of the present invention is possible. Figures 3 and 4 show two sets of such production examples. In the examples of Figures 3 and 4, as in the examples of Figures 1 and 2, the audio signals are simultaneously applied to a set of modified audio signal processing programs or devices 2 in a set of signal paths, but in separate control paths. The generated modified parameters (4 in Fig. 3 and 4'' in Fig. 4) each receive the unmodified audio signal and the modified audio signal. In the examples of Figs. 3 and 4, the audio signal is modified. 2 and generating modified parameters (4" and 4"' respectively) to modify the audio signal to reduce the specific loudness at it, which may be implied, and a set of target specific loudness, which may also be implicit, difference between.

As with the example of FIG. 3, the generated modified parameter handler or apparatus 4' may include a number of functions and/or means: a set of target specific loudness 6 as in the example of FIG. 1, a target specificity of the feedback example of FIG. 2 One set of loudness 14 calculates the approximation, and a "calculate modified parameter" handler or device 10". As in the example of Figure 1, the target specific loudness is calculated in the pre-committed portion of the pre-mixed/received paradigm. One or more functions "F" can be executed each having a function parameter. In Figure 3 this is decomposed to indicate the input of the "Select Function F and Function Parameters" handler or device 6. Before this hybrid is given / The feedback portion of the feedback example, the modified audio signal is applied to a set of computational approximations of the target specific loudness 14, as in the feedback example of Figure 2. In the example of Figure 3, the processing program or device 14 is calculated. The operation of modifying the specific loudness or part of the specific loudness of an audio signal is as its operation in the example of Figure 2. This particular loudness or partial specific loudness is a set of approximations of the target specific loudness. Target specific loudness (from the handler Or device 6) The approximation to the target specific loudness (from the processing program or device 14) is applied to the computational modification parameter 10" to derive the modified parameter M, which if applied to the audio signal by the modified audio signal 2, will be reduced in the unmodified audio signal The difference between a specific loudness and a target specific loudness. As mentioned above, these modified parameters M may, for example, be of a time varying gain having a frequency band applied to the filter bank or a time varying filter. In a practical embodiment, the feedback loop may introduce a delay between the calculation and application of the modified parameter M. As described above, in a replay environment having a second interfering audio signal, such as a noise, the calculated modified parameter 10" and the calculated approximation of the target specific loudness 14 each selectively receive the amount of such second interfering audio signal. The input or second interfering signal is itself input as a group and the processing program or device 14 can calculate a portion of the specific loudness of the modified audio signal. Such selective input is shown in Figure 3 as a long polyline.

Calculating the modified parameter 10" may employ a set of error detecting means or functions such that the difference between the target specific loudness and the target specific loudness approximation input adjusts the modified parameter to reduce the approximation and "actual" at the target specific loudness. The difference between the target specific loudness.

This adjustment reduces the difference between the specific loudness of the unmodified audio signal, and the target specific loudness, which may be implied. Therefore, the modification parameter M can be based on the target specific loudness calculated from the specific loudness of the original audio using the function F in the pre-route, and the target calculated from the specific loudness or part of the specific loudness of the modified audio in the feedback route. The error between the specific loudness approximations is adjusted.

Additional pre-grant/reward examples are shown in the example of Figure 4. The difference between this additional example and the example of Fig. 3 is that the inverse function F ^{- 1} is calculated in the feedback route instead of the function F being calculated in the preamble path. As in the example of FIG. 4, the generation of the modified parameter handler or apparatus 4' may include a number of functions and/or means: as in the first example of the first example of FIG. 1, a particular loudness is calculated, as in the feedback example of FIG. The calculated approximation of the specific loudness of the modified audio 12, and a set of computational modification parameters 10'''. As in the pre-example of Figure 1, a set of calculated specific loudnesses 8 provides the specific loudness of the unmodified audio signal as an input to calculate the modified parameter 10'''. As in the feedback example of FIG. 2, depending on whether the function F is a reversible limit, the processing program or device 12 evaluates the unmodified by applying the inverse function F ^{- 1} to the specific loudness or partial specific loudness of the modified audio signal. The specific loudness of the audio signal. The "select inverse and inverse function parameters" as inputs to the calculated approximation of the particular loudness of the unmodified audio 12 are shown to indicate that the device or handler 12 can calculate a set of inverse functions F ^{- 1} , as described above. This is decomposed in Fig. 4 as the input of the "select inverse function F ^{- 1} and function parameters" handler or device 12. Thus, the handler or device 12 provides an approximation of the particular loudness of the unmodified audio signal as another input to calculate the modified parameter 10"'.

As in the example of Figures 1-3, calculating the modified parameter 10''' derives a modified parameter M that, if applied to the audio signal by the modified audio signal 2, will reduce the specific loudness of the unmodified audio signal and in this example It is the difference between the implied target-specific loudness. As mentioned above, these modified parameters M may, for example, be of a time varying gain having a frequency band applied to the filter bank or a time varying filter. In a practical embodiment, the feedback loop may introduce a delay between the calculation and application of the modified parameter M. As described above, in a replay environment having a second interfering audio signal, such as a noise, the calculated modification parameter 10''' and the calculated approximation of the specific loudness of the unmodified audio 12 are each selectively receivable. The measurement of the two interfering audio signals as an input or a second interfering signal is itself input as a group and the processing program or device 12 can calculate a portion of the specific loudness of the modified audio signal. This selective input is shown in Figure 4 as a long fold line.

Calculating the modified parameter 10" may employ a set of error detecting means or functions such that the difference between its particular loudness and the specific loudness approximation input produces an output of the adjusted modified parameter to reduce the approximation and "actual" at a particular loudness. The difference between specific loudnesses. Since the approximation of a particular loudness is derived from the specific loudness or part of the specific loudness of the modified audio, it can be used as the approximation of the target specific loudness, which reduces the specific loudness of the modified audio signal and The difference between the target specific loudness inherent in function F ^{- 1.} Therefore, the modified parameter M can be based on the specific loudness calculated from the original audio in the pre-route, and the inverse function F ^{- 1} is used in the feedback route. The error between the particular loudness of the audio or the specific loudness approximation of the particular loudness calculation is modified. Due to the feedback path, the actual fabrication can introduce a delay between the modification of the modified parameter and the application.

Although the modified parameter M in the examples of Figures 1-4 reduces the difference between the specific loudness of the audio signal and the target specific loudness when applied to a set of modified audio signal processing programs or devices 2, in a practical embodiment The corresponding modified parameters generated for the same audio signal may not be identical to each other.

Although the arguments of the present invention are not primary or necessary, the calculation of the specific loudness of the audio signal or the modified audio signal may advantageously be carried out using the serial number PCT/US2004/016964 previously filed in the International Patent Application, which is incorporated herein by reference. The technique of WO 2004/111964 A2, wherein calculating a combination of one or two or more specific loudness mode functions from a group of two or more sets of specific loudness mode functions, the selection of which is input to an audio signal The measurement of the characteristics is controlled. The description of the specific loudness 104 of Fig. 1 will be explained below.

According to a further aspect of the invention, the unmodified audio signal and either (1) modified parameter or (2) the target specific loudness or the target specific loudness representation (eg, may be used explicitly or implicitly in the calculation The scale adjustments, such as target specific loudness, can be stored or sent, for example, for use in devices and processes that are separated in time and/or space. The modification of the parameter, the target specific loudness, or the representation of the target specific loudness may be determined in any suitable manner, for example, as described in Figures 1-4, prior to the grant, feedback, and hybrid pre-request configuration examples, as described above. In fact, a set of pre-administration configurations, such as the example of Figure 1, is the least complicated and fastest because it avoids calculations based on the modified audio signal. Figure 5 shows an example of transmitting or storing unmodified audio and modifying parameters. Figure 6 shows an example of transmitting or storing an unmodified audio and a target specific loudness or target specific loudness representation.

A set of configurations, as exemplified in FIG. 5, can be used to separate the application of modified parameters to audio signals temporally and/or spatially from the generation of such modified parameters. A set of configurations, as exemplified in FIG. 6, can be used to separate the generation and application of modified parameters from the target specific loudness or its representation in time and/or space. Both configurations result in a simple, low cost replay or receive configuration that avoids the complexity of generating modified parameters or producing target specific loudness. Although the configuration of the fifth pattern is simpler than the configuration of the sixth pattern, the configuration of FIG. 6 has the advantage of less information that needs to be stored or transmitted, especially in the representation of the target specific loudness, such as one or more scales. When adjustments are stored or sent. The reduction in stored or transmitted information is particularly useful in low bit rate audio environments.

Accordingly, a further object of the present invention is to supply a set of devices or processing programs (1) that receive or play from a set of storage or transmission devices or processing programs, modify parameters M, and apply them to simultaneously received audio. a signal or (2) that receives or plays from a set of storage or transmission devices or processing programs, a target specific loudness or a target specific loudness representation, and utilizes the target specific loudness or its representation to simultaneously received audio. The signal (or to the measurement of the audio signal, for example, the particular loudness from which the audio signal is derived) produces a modified parameter M and applies the modified parameter M to the received audio signal. This apparatus or processing program may be categorized as a decoding processing program or decoder; and a device or processing program required to generate stored or transmitted information may be classified as an encoding processing program or encoder. This encoding process or encoder is part of the configuration example of Figures 1-4 that can be used to generate the information needed for the respective decoding process or decoder. This decoding processor or decoder can be combined or operated with virtually any processing or device that processes and/or replicates sound.

In an aspect of the present invention, as in the example of FIG. 5, the unmodified audio signal and the modified parameter M are available, for example, a set of modified parameter generation processing programs or generators, for example, the modified parameter 4 of FIG. 4' of Figure 2, 4' of Figure 3 or 4'' of Figure 4 is applied to any suitable storage or transmission device or function ("storage or transmission") 16 and is used. In the case of the first example of the encoding process or the encoder, if the modified audio is not required to be provided at the time or space of the encoder or the encoding process, the modified audio signal 2 will not need to generate the modified audio, and The storage or transmission 16 may comprise, for example, any suitable magnetic, optical or solid state storage and playback device or any suitable wired or wireless transmission and reception device, the choice of which is not essential to the invention. Play or receive The modified parameters can then be applied to the modified audio signal 2 having the pattern used in the examples of Figures 1-4 to modify the played or received audio signal to approximate its specific loudness. The specific loudness or modification parameters are inherently derived from the configuration. The modified parameters can be stored or transmitted in any of a variety of ways. For example, they can be stored or sent as metadata accompanying the audio signal, which can be routed by separate paths or channels. Transmit, they can be implicitly encoded in the audio, they can be multiplexed, etc. The use of modified parameters to modify the audio signal can be selective and, if selective, their use can be, for example The user selects. For example, modifying the parameters if applied to the audio signal may reduce the dynamic range of the audio signal. The user may choose whether to use this dynamic range reduction.

In another aspect of the invention, as exemplified in Figure 6, the representation of the unmodified audio signal and the target specific loudness or target specific loudness may be applied to any suitable storage or transmission device or function ("storage or transmission" ") 16.

In the case of using the configuration example as shown in Figure 1 as an encoding process or encoder, if it is not necessary to provide any modified parameters or modified audio at the time or spatial position of the encoder or encoding processor, then the modified parameters are calculated. A type 10 processing program or apparatus or a processing program or apparatus for modifying the audio signal type 2 is not required and may be omitted. As in the case of the example of Figure 5, the storage or transmission 16 may comprise, for example, any suitable magnetic, optical or solid state storage and replay device or any suitable wired or wireless transmission and reception device, the choice of which is for the present invention Not the main one. The representation of the target specific loudness or target specific loudness being played or received may then be applied, along with the unmodified audio, to a set of computational modification parameters 10, such as the pattern employed in the example of FIG. 1, or applied to a group Calculating the modified parameter 10", as used in the example of Figure 3, to provide a modified parameter M which can then be applied to the modified audio signal 2, as in the example of Figures 1-4, to modify the play or Receiving an audio signal such that its particular loudness approximates the target specific loudness or the configuration in which the modified parameter is derived. Although the target specific loudness or its representation is immediately available from an encoding process or encoder of the type illustrated in the first example of FIG. 1, The approximation of the target specific loudness or its representation or target specific loudness or its representation may be obtained from an encoding process or encoder of the type of the example of Figures 2 to 4 (approximation of the processing procedures of Figures 2 and 3) Or the device 14 and the program or device 12 of Figure 4 are calculated. The target specific loudness or its representation can be stored or transmitted in any of a variety of ways. For example, they can be stored Stored or transmitted as metadata accompanying audio signals, which may be transmitted by separate paths or channels, which may be implicitly encoded in the audio, they may be multiplexed, etc. Used from being stored or transmitted The target specific loudness or the derived modified parameters to modify the audio signal may be selective, for example, their use may be selected by the user. For example, modifying the parameter may reduce the dynamic range of the audio signal if applied to the audio signal. You can choose whether to use this dynamic range reduction.

While making the disclosure of the present invention a set of digital systems, a set of pre-configured configurations is the most practical, and examples of such configurations will be described in detail, and it is to be understood that the scope of the present invention is not limited thereto.

In this file, some names, such as "filters" or "filter banks" are used here to contain virtually any recursive and non-recursive filtering patterns, such as IIR filters or transformations, and "filtering" information is used. The result of a class filter. The embodiment described below employs a filter bank made by conversion.

Figure 7 shows more detailed details of an embodiment of the invention implemented in a set of pre-configured configurations. The audio first passes through a set of analysis filter bank functions or means ("analysis filter bank") 100, which divides the audio signal into a plurality of frequency bands (thus, Figure 5 shows a plurality of outputs from the analysis filter bank 100, each The output represents a set of frequency bands whose output is transmitted via various functions or means to a synthesis filter bank that combines the frequency bands into a set of combined wideband signals, as further illustrated below. The response of the filter in the analysis filter bank 100 with respect to each frequency band is designed to simulate the reaction of the basement membrane at a particular location in the inner ear. The output of each filter in the analysis filter bank 100 then enters a set of transmission filters or transmission filter functions ("transmission filters") 101 that simulate the filtering effects of audio transmission via the outer and middle ear. If only the audio loudness needs to be measured, the transmission filter can be applied before the analysis filter bank, but since the analysis filter bank output is used to synthesize the modified audio, it is advantageous to apply the transmission filter after the filter bank. . The output of transmission filter 101 then enters a set of excitation functions or devices ("excitation") 102 whose output simulates the energy distribution along the basement membrane. The excitation energy value can be smoothed across time using a smoothing function or device ("smoothing") 103. The function constants of the smoothing function are set according to the needs of the desired application. The smoothed excitation signal is sequentially converted to a particular loudness in a particular loudness function or device ("specific loudness (SL)") 104. The specific loudness is expressed in units of Song per unit frequency. The specific loudness component associated with each frequency band is passed to a particular loudness modification function or device ("SL Modification") 105. The input of the SL modification 105 is the original specific loudness and then outputs the desired or "target" specific loudness, which is preferably a function of the original specific loudness according to the argument of the present invention (see the next title, entitled "Target Specific Loudness"). Depending on the desired effect, the SL modification 105 can operate independently on each frequency band, or a correlation may exist between the frequency bands (as suggested by the intersection of the intersection lines of Figure 7 for frequency smoothing). A set of gain solver functions or means ("gain solver") 106 takes as input the smoothed excitation band component from the excitation 102 and the target specific loudness from the SL modification 105 and determines that it needs to be applied to the analysis filter bank. Each frequency band of the output of 100 is used to convert the measured specific loudness to the gain of the target specific loudness. The gain solver can be made in a variety of ways. For example, the gain solver can include a set of repetitive processes, for example, in the manner of PCT/US2004/016964, which is incorporated by reference in the International Patent Application Serial No. WO 2004/111964 A2, or otherwise. Although the gain per band produced by gain solver 106 can be further smoothed with time using a selective smoothing function or device ("smoothing") 107 to minimize the perception of artifacts, in a comprehensive processing or Time smoothing in the device is preferably applied elsewhere, as explained elsewhere. Finally, the gain is applied to the respective frequency bands of the analysis filter bank 100 via a set of separate multiplication combining functions or combiners 108, and processed or "modified" in the synthesis filter bank function or device ("synthesis filter" The band->modified band is synthesized from the gain-modified band. Moreover, the output from the analysis filter bank can be delayed by a delay function or device ("delay") 109 prior to applying the gain to compensate for any delay associated with the gain calculation. Additionally, in addition to calculating the gain used in applying the gain modification to the band, the gain solver 106 can calculate filter coefficients that control a set of time varying filters, such as a set of multi-branch FIR filters or a set of multi-pole IIRs. filter. To simplify the excitation, the arguments of the present invention are generally illustrated as employing gain coefficients applied to the frequency bands, it being understood that the filter coefficients and time varying filters can also be employed in practical embodiments.

In a practical embodiment, audio processing can be performed in the digital domain. Thus, audio input signals to discrete time sequence x [n] Representative, from which audio source is sampled at a sampling frequency f _s. Suppose the sequence x[n] has been properly scaled so that the average output power of x[n] is in decibels as follows

It is equal to the sound pressure level in dB, which is heard by human listeners. Furthermore, to simplify the excitation, the audio signal is assumed to be mono.

The analysis filter bank 100, the transmission filter 101, the excitation 102, the specific loudness 104, the specific loudness modification unit 105, the gain solver 106, and the synthesis filter bank 110 can be described in more detail below.

The analysis filter bank 100 audio input signal is applied to an analysis filter bank or filter bank function ("analysis filter bank") 100. The individual filters of the analysis filter bank 100 are designed to simulate the frequency response along a particular location in the inner ear of the basement membrane. Filter bank 100 can include a set of linear filters whose bandwidth and spacing adjustments are constant over the equivalent rectangular bandwidth (ERB) frequency scale, such as Moore, Glasberg, and Baer (BCJ Moore, B. Glasberg, T. Baer, supra). , defined by "predictive mode of threshold, loudness, and partial loudness").

While the ERB frequency scale more closely matches human perception and exhibits better performance in producing objective loudness measurements that match subjective loudness results, Bark frequency scales with lower performance can be employed.

At a center frequency f in Hertz, a set of ERB bandwidths in Hertz can be approximated as follows: ERB ( f ) = 24.7 (4.37 f / 1000 + 1) (1)

Based on this relationship, a set of distorted frequency scales is defined such that at any point along the scale being distorted, the corresponding ERB in units of the distorted scale is equal to one. The function of converting from a linear frequency in Hertz to this ERB frequency scale can be obtained by integrating the reciprocal of Equation 1:

At the same time, it is also useful to use Equation 2a to find the representation of f to convert the ERB scale to a linear frequency scale: Where e is in units of ERB. Figure 9 shows the relationship between the ERB scale and the frequency in Hertz.

The analysis filterbank 100 can include a set B of auditory filters, referred to as frequency bands, and are evenly spaced along the ERB scale at the center frequency. More specifically, f _c [1] = f _{m i n} (3a) f _c [ b ] = f _c [b-1] + ERBToHz ( HzToERB ( f _c [ b -1]) + Δ) b = 2 ... B (3b) f _c [ B ]< f _{m a x} , (3c) where Δ is the ERB interval adjustment required for analyzing the filter bank 100, and wherein f _{m i n} and f _{m a x} are respectively The minimum and maximum center frequencies required. Δ=1 can be selected, and f _{m i n} =50 Hz and f _{m a x} =20,000 Hz can be set in consideration of the frequency range sensitive to the human ear. With these parameters, for example, the application of equations 3a-c will produce a B=40 audible filter.

The amplitude frequency response of each auditory filter can be described using a set of circular pass exponential functions, as suggested by Moore and Glasberg. Specifically, the amplitude response of a filter having a center frequency f _c [ b ] can be calculated as: H _b ( f )=(1+ pg ) e ^{- pg} (4a)

The amplitude response of these B auditory filters is approximately the critical band on the ERB scale, as shown in Figure 10.

The filtering operation of the analysis filter bank 100 can be approximated using a finite-length discrete Fourier transform as appropriate, commonly referred to as short-time discrete Fourier transform (STDFT), because the filter is performed at the sampling rate of the audio signal, which is called For full rate production, it is believed to provide a higher time resolution than is required for accurate loudness measurements. Using STDFT instead of full rate production results in improved efficiency and reduced computational complexity.

The STDFT of the input audio signal x[n] is defined as: Where k is the frequency index, t is the time block indicator, N is the DFT size, T is the interval size, and w[n] is the normalized length N window

Note that the variable t in Equation 5a is a discrete indicator representing the time block of the STDFT rather than a time measurement in seconds. Each increment t represents the T-sampling interval along the signal x[n]. The reference to the indicator t is also assumed to be this definition. Different parameter settings and window shapes can be used depending on the details of the production. Select n = 2048, T = 1024 at f _s = 44100 Hz , and provide enough time and frequency for w[n] as a group of Hanning windows. The balance of resolution. The above STDFT can be made more efficient using Fast Fourier Transform (FFT).

Instead of using STDFT, modified discrete cosine transform (MDCT) can be employed to create an analysis filter bank. MDCT is a conversion commonly used in perceptual audio encoders, for example, Dolby AC-3. If the disclosed system is produced with such perceptually encoded audio, the revealed metrics and modifications can be more efficiently produced by processing the existing MDCT coefficients of the encoded audio, thereby eliminating the need for analysis filter bank conversion. . The MDCT of the input audio signal x[n] is given as: ,among them (6)

In general, the spacing dimension T is chosen to be exactly half of the conversion length N and thus it is possible to completely reconstruct the signal x[n].

The output of the transmission filter 101 analysis filter bank 100 is applied to a set of transmission filters or transmission filter functions ("transmission filters") 101 which transmit the respective bands of the filter bank in accordance with the audio transmission via the outer and middle ears. Filtering. Figure 8 shows the transmission filter P(f), an appropriate amplitude frequency response across the audible frequency range. The response is one unit below 1 kHz and, above 1 kHz, will be normalized to equal one unit depending on the reciprocal of the hearing threshold, as specified by the ISO 226 standard.

In order to calculate the loudness of the input audio signal, the excitation 102 must measure the short-term energy of the audio signal in each of the filters of the transmission filter 101 after applying the analysis filter bank 100. This time and frequency change measurement is called excitation. The short-term energy output of each of the filters in the analysis filter bank 100 can be approximated in the excitation function 102 by multiplying the filter response in the frequency domain by the power spectrum of the input signal: Where b is the number of bands, t is the number of blocks, and H _b [ k ] and P [ k ] are the frequency responses of the auditory filters and transmission filters sampled at frequencies corresponding to the index k of the STDFT or MDCT, respectively. . It should be noted that the amplitude response pattern of the auditory filter other than the one specified in Equations 4a-c can also be used in Equation 7 to achieve similar results. For example, the international patent application serial number PCT/US2004/016964, which is hereby incorporated by reference in its entirety, is hereby incorporated by reference in its entirety in the the the the the the the the the The cost of the "brick wall" bandpass approximation.

In summary, the output of the excitation function 102 is represented by the frequency domain in the ERB band b of each time period t.

Time averaging ("smoothing") 103 In some applications of the disclosed invention, as explained below, it may be desirable to smooth the stimulus before it transitions to a particular loudness. For example, smoothing can be performed recursively with the smoothing function 103 according to the following equation: The time constant λ _{b of} each frequency band b is selected depending on the desired application. In most cases the time constant can advantageously be chosen to be proportional to the integration time of human loudness perception within frequency band b. Watson and Gengel experimented to show that this integration time is in the range of 150-175 ms at low frequencies (125-200 Hz) and in the range of 40-60 ms at high frequencies (Charles S. Watson and Roy W. Gengel, " Signal persistence and signal frequency versus relative auditory sensitivity, taken from the American Journal of Auditory Association, 1969, Volume 46, number 4 (Part 2), pages 989-997).

The specific loudness 104 is in a particular loudness converter or transfer function ("specific loudness") 104, and each frequency band of the excitation is converted to a component score of a particular loudness, which is measured in units of ERB units.

When starting to calculate a specific loudness, The excitation level of each band of [ b , t ] can be converted to an equivalent excitation level at 1 kHz, as specified by the equivalent loudness contour of ISO 226 (Fig. 11), with the transmission filter P(z) ( Figure 12) Normalization: Where T _{1 kHz} ( E , f ) is a function produced at a level of _{1 kHz} , which corresponds to the loudness of the level E of the frequency f. In fact, T _{1 kHz} ( E , f ) is interpolated as a comparison table of equal loudness contours and normalized by a transmission filter. Convert to the specific loudness calculation below the equivalent level at 1 kHz.

Then, the specific loudness of each frequency band can be calculated as: N [ b , t ]=α[ b , t ] N _NB [ b , t ]+(1 -α[ b , t ]) N _WB [ b , t ] (10) where N _NB [ b , t ] and N _WB [ b , t ] are specific loudness values depending on the narrowband and wideband signal patterns, respectively. The value α[ b , t ] is an interpolation coefficient calculated from the audio signal between 0 and 1. It is disclosed in International Patent Application Serial No. PCT/US2004/016964 to WO 2004/111964 A2, which describes the technique for calculating α[ b , t ] from the spectral smoothness of excitation: it also specifies "narrowband" and "details" in detail. Broadband" signal mode.

The narrowband and wideband specific loudness values N _NB [ b , t ] and N _wB [ b , t ] can be evaluated from the transformed excitation using an exponential function: Where TQ _{1 kHz} is the excitation level of the _{1 kHz} tone at the lower sound limit. According to the equal loudness contour (Figures 11 and 12) is equal to 4.2dB. Note that these specific loudness functions are equal to zero when the excitation is equal to the smaller sound threshold. For excitations that are greater than the smaller sound threshold, the two functions monotonically increase with the power law according to the Stevens intensity perception law. The exponent of the narrow band function is chosen to be larger than the exponent of the wide band function such that the narrow band function increases more rapidly than the wide band function. The specific choice of the exponent β and the gain G for the narrow and wide band cases was chosen to match the experimental data for the loudness growth of tones and noise.

Moore and Glasberg suggest that the specific loudness should be equal to some small values rather than zero when the stimulus is in the hearing threshold. The specific loudness should then be monotonically reduced to zero as the stimulus is reduced to zero. Because hearing threshold is a possible threshold (tone perception rate is 50%), and some tones, each in their limits, are presented together and can sum up sounds that are more audible than any separate tone. In the disclosed application, adding this property to a particular loudness function will add to the benefit of the gain solver's more appropriate performance when the stimulus approaches the threshold, as will be discussed below. If the specific loudness is defined as zero when the stimulus is below the threshold or threshold, the gain solver will not have a single solution when the stimulus is below the threshold or threshold. If, on the other hand, a particular loudness is defined as a monotonous increase in exciting all values greater than or equal to zero, a single solution would exist as suggested by Moore and Glasberg. A loudness scale factor greater than one unit will result in a gain greater than one unit and vice versa. The specific loudness function of equations 11a and 11b can be modified to have the required property basis: Where the λ constant is greater than one, the η exponent is less than one, and the constants K and C are selected such that the specific loudness function and its first derivative are =λ TQ _{1 kHz} point is continuous.

Depending on the specific loudness, the full or "total" loudness L [ t ] is given by the sum of the specific loudness across all frequency bands b:

The specific loudness modification section 105 in the specific loudness modification function ("specific loudness modification") 105, the target specific loudness, is called [ b , t ] can be calculated from the specific loudness of the SL 104 (Fig. 7) in various ways depending on the desired full device or processing application. As will be explained in more detail below, a set of target specific loudnesses can be calculated using a set of scale factors a, for example, in the case of volume control. See equation 16 and its associated description. In the case of automatic gain control (AGC) and dynamic range control (DRC), a set of target specific loudnesses can be calculated using the ratio of the desired output loudness to the input loudness. See equations 17 and 18 and their associated descriptions. In the case of dynamic equalization, a set of target specific loudnesses can be calculated using the relationship provided in Equation 23 and its associated description.

Benefit solver 106 In this example, gain solver 106 will be smoothed for each frequency band b and each interval t [ b , t ] and target specific loudness [ b , t ] acts as its input and produces a gain G [ b , t ] that is then used to modify the audio. Let the function Ψ{. } represents a nonlinear transformation from excitation to specific loudness.

The gain solver solves G [ b , t ] so that

The gain solver 106 determines the frequency and time varying gains, when applied to the original excitation, produces a particular loudness that is ideally equal to the desired target specific loudness.

In effect, the gain solver determines the frequency and time varying gains that, when applied to the frequency domain version of the audio signal, produce a modified audio signal to reduce the difference between its particular loudness and target specific loudness. Ideally, the modification is such that the modified audio signal has a specific loudness that approximates an approximation of the target specific loudness. The solution of equation 14a can be made in a variety of ways. For example, if the closed form mathematical representation of the reciprocal of a particular loudness exists, and Ψ ^{- 1} {. } indicates that the gain can be calculated directly by reconfiguring equation 14a:

In addition, if Ψ ^{- 1} {. The closed form solution does not exist, and an iterative method can be employed in which each iteration equation 14a is evaluated using the evaluation value of the current gain. The specific loudness produced is compared to the desired target and the gain is changed according to the error. If the gains are properly changed, they will converge to the desired solution. Another method involves pre-calculating a function of a range of excitation values in each frequency band Ψ{. } to generate a set of comparison tables. An approximation of the inverse function can be obtained from this look-up table and the gain can then be calculated from Equation 14b. As noted above, the target specific loudness can be represented by a scale adjustment of a particular loudness:

Substituting Equation 13 into 14c and then substituting 14c into 14b produces an additional representation of the gain:

Gain can be fully motivated [ b , t ] and a specific loudness scale adjustment Ξ [ b , t ] is expressed as a function. Thus, the gain can be calculated via an evaluation of 14d or an equivalent comparison table without having to explicitly calculate a particular loudness or target specific loudness as an intermediate value. However, these values are implicitly calculated via the use of Equation 14d. Another equivalent method of calculating a modified parameter using a specific loudness and a target specific loudness, either explicitly or implicitly, can be designed, and the present invention will cover all such methods.

Synthesis Filter Bank 110 As described above, the analysis filter bank 100 can be efficiently fabricated using Short Time Discrete Fourier Transform (STDFT) or Modified Discrete Cosine Transform, and STDFT or MDCT can be similarly used to make a synthesis filter. Group 110. Specifically, let X [ k , t ] represent the STDFT or MDCT of the input audio, as defined previously, the STDFT or MDCT of the processed (modified) audio in the synthesis filter bank 110 can be calculated as Where S _b [ k ] is the composite filter response associated with band b, and d is the delay associated with delay block 109 in FIG. The pattern of the synthesis filter S _b [ k ] can be selected to be the same as that used in the analysis filter bank H _b [ k ], or they can be modified to lack any gain modification (ie, when G A complete reorganization is provided when [ b , t ] = 1). Those skilled in the art should understand that the final processed audio can then be passed through Fourier or The modified cosine transform of [ k , t ] and the overlap-addition synthesis are generated.

Target Specific Loudness The configuration behavior of the arguments implementing the present invention, such as the examples of Figures 1-7, generally depends on the target specific loudness [ b , t ] is calculated. Although the invention is not limited to any particular function or inverse function of calculating a particular loudness of a target, many such functions and their appropriate applications will be described.

Non-Time-varying and Non-Frequency Variable Functions for Volume Control A standard volume control adjusts the loudness of an audio signal by applying a set of broadband gains to the audio. In general, the gain is coupled to a knob or fader that is adjusted by the user until the loudness of the audio reaches the desired level. The arguments of the present invention allow such controls to be more in line with psychoacoustic production. In accordance with this aspect of the invention, instead of coupling a set of broadband gains to volume control to produce the same amount of gain change across all frequency bands, which may result in a change in the perceived spectrum, a set of specific loudness scale adjustment coefficients will be combined to The volume control is adjusted such that the gain of each of the majority of the bands is changed in consideration of the number of human hearing modes, and ideally, does not result in a change in the perceived spectrum. In the context of this aspect of the invention and its application, "constant" or "non-time varying" is the intention to allow the setting of the volume control scale factor, for example, by a user from time to time. This “non-time-varying” is sometimes referred to as “quasi-time-variant”, “quasi-steady state”, “fragment non-time-varying”, “fragment steady-state”, “stage-time non-time-varying”, and “stage”. Steady state". Given the scale factor, α, the target specific loudness can be calculated as the result of multiplying the specific loudness of the measurement by α:

Since the total loudness L [ t ] is the sum of the specific loudnesses N [ b , t ] across all frequency bands b, the previous modification also adjusts the total loudness by the coefficient α, but the manner can be saved for the change of the volume control adjustment at a specific time. The same perceptual spectrum. In other words, at any particular time, the change in volume control adjustment produces a change in perceived loudness but does not produce a change in the perceived spectrum of the modified audio for the perceived spectrum of the unmodified audio. Figure 13a shows the multi-band gain G [ b , t ] produced by a set of audio signals consisting of female speech at a particular time "t" across frequency band "b" when α = 0.25. The wideband gain required to adjust the original total loudness at 0.25 (horizontal line), as in standard volume control, is also plotted as a comparison. Compared to the intermediate frequency band, the multi-band gain G [ b , t ] increases in the low and high frequency bands. This is consistent with an equal loudness contour that indicates that the human ear is less sensitive at low and high frequencies.

Figure 13b shows the original audio signal, the wideband gain modified signal modified according to prior art volume control, and the specific loudness of the multi-band gain-modified signal modified in accordance with this aspect of the invention. The specific loudness of the multi-band gain modified signal is the original specific loudness adjusted by 0.25. The specific loudness of the wideband gain modified signal has a changed spectral shape relative to the original unmodified signal. In this case, the specific loudness, relatively, loses the loudness at low and high frequencies. This is perceived as blurring of the audio because its volume is reduced and the multi-band modified signal does not have this problem because its loudness is controlled by the gain derived from the perceived loudness field.

In addition to the perceived spectral balance distortion associated with traditional volume control, there is a second problem. Presented in the loudness mode provided by equations 11a-11d, one property of loudness perception is that signal loudness at any frequency will decrease more rapidly as the signal level approaches hearing threshold. As a result, the electrical attenuation required to provide the same loudness attenuation to a set of smaller acoustic signals is less than the electrical attenuation required for a larger acoustic signal. Conventional volume control provides a set of constant attenuations regardless of the signal level, and thus the smaller acoustic signal becomes "too small" compared to the larger acoustic signal when the volume is reduced. In many cases this results in a loss of audio detail. Consider the case where the castanets are recorded in the reverberant room. The main castanets "hit" in this recording is significantly louder than the reverberant echo, but the echo echo transmits the size of the room. When the volume is lowered by conventional volume control, the reverberant echo becomes a minor sound relative to the main hit and will eventually disappear below the hearing threshold, resulting in a "dry" sound cast. The loudness-based volume control prevents the disappearance of the smaller portion of the recording by using the smaller acoustic reverberation portion of the recording to be enhanced relative to the larger main hit, so that the relative loudness between these portions remains constant. . To achieve this effect, the multi-band gain G [ b , t ] must vary with time at a rate commensurate with the temporal resolution of human loudness perception. Because the multi-band gain G [ b , t ] is calculated as smoothing excitation The function of [ b , t ], the time constant λ _b in Equation 8 depends on the speed of the change in gain across time in each frequency band b. As previously described, these time constants can be selected to be proportional to the integration time of human loudness perception within band b and thus produce an appropriate change G [ b , t ] over time. It should be noted that if the time constant is improperly selected (too fast or too slow), then a perceptually unpleasant artificial effect may be introduced to process the audio.

Non-time-varying and frequency-varying functions suitable for fixed equalization In some applications, it may be desirable to apply fixed perceptual equalization to the audio, in which case the target specific loudness may utilize a set of non-time-varying but frequency-changing scale coefficients. [ b ] is calculated as follows among them [ b , t ] is the target specific loudness, N [ b , t ] is the specific loudness of the audio signal, b is the frequency measurement, and t is the function measurement. In this case, the scale adjustment can vary depending on the frequency band. This application can be used to emphasize, for example, the portion of the speech frequency that is dominant in the spectrum in order to increase understanding.

Non-frequency-varying and time-varying functions for automatic gain and dynamic range control The techniques of automatic gain and dynamic range control (AGC and DRC) are well known in the field of audio processing. Abstractly, both techniques measure the level of the audio signal in some way and then modify the signal by the number of gains that measure the level. In the case of AGC, the signal is gain-modified such that its measurement level is closer to the user-selected reference level. In the case of DRC, the signal is gain-modified such that the measurement level range of the signal is converted to the desired range. For example, the user may wish to make the smaller portion of the audio louder and the louder portion smaller. Such systems are described by Robinson and Gundry (Charles Robinson and Kenneth Gundry, "Controlling Dynamic Range via Metadata", 107th AES Conference, New York, 24-27 September 1999, pre-printed 5028). Traditional fabrication of AGCs and DRCs typically uses simple measurements of audio signal levels, such as smoothing peak or root mean square (rms) amplitudes to drive gain modification. Such simple measurements have some degree of correlation with the perceived loudness of the audio, but the arguments of the present invention utilize a loud metric based on the psychoacoustic mode to drive gain modification while allowing for more perceptual AGC and DRC. At the same time, many conventional AGC and DRC systems apply gain modifications with wideband gain, resulting in distortion of the quality (spectral) of the above-described processed audio. On the other hand, the argument of the present invention forms a specific loudness in a manner that a set of multi-band gains are used to reduce or minimize such distortion.

Both AGC and DRC applications employing the teachings of the present invention feature a function of converting or mapping the input wideband loudness L _i [ t ] to a desired output wideband loudness L _o [ t ], where the loudness is in units of perceived loudness. Measurement, such as Song. The input wideband loudness L _i [ t ] is a set of functions of the specific loudness N [ b , t ] of the input audio signal. Although it may be the same as the overall loudness of the input audio signal, it may be a time smoothed version of the total loudness of the audio signal.

Figures 14a and 14b show examples of general mapping functions for AGC and aDRC, respectively. Given this a mapping where L _o [ t ] is a function of L _i [ t ], the target specific loudness can be calculated as

The original specific loudness N [ b , t ] of the audio signal is simply adjusted to produce an output specific loudness at a desired ratio of output wideband loudness to input wideband loudness. [ b , t ]. In an AGC system, the input wideband loudness L _i [ t ] is typically a measure of the long-term total loudness of the audio. This can be achieved by smoothing the total loudness L [ t ] across time to produce L _i [ t ].

In contrast to AGC, the DRC system reacts to a shorter-term change in the loudness of the signal, and thus L _i [ t ] can simply be equal to L [ t ]. As a result, the scale adjustment of the specific loudness, expressed as L _o [ t ]/ L _i [ t ], may rapidly change the artifacts that are not required in the processed audio. A general drawback is that some of the other relatively uncorrelated portions of the spectrum cause audible modulation of portions of the frequency spectrum. For example, classical music choices may include high frequencies where dominant string notes dominate, and low frequencies with tonal drums that make loud noises. When the timpani is hit, the overall loudness L _i [ t ] increases and the DRC system applies attenuation to the entire specific loudness. Then you can hear the string loudness being weakened and enhanced with the timpani. Conventional broadband DRC systems have the same problem of mutual adjustment of such spectrum, and the general solution involves applying DRCs independently to different frequency bands. The system disclosed herein is inherently multi-band because it has a filter bank and employs a calculation of the specific loudness of the perceived loudness pattern, and thus modifies the DRC system to operate in accordance with the inventive multi-band format, the system relatively Direct and will be explained next.

Frequency-variable and time-varying function DRC systems suitable for dynamic range control can be extended to operate in multi-band or frequency-variant mode by allowing input and output loudness to vary independently with frequency band b. These multi-band loudness values are represented by L _i [ b , t ] and L _o [ b , t ], and the target specific loudness can then be given as Where L _o [ b , t ] is independently calculated or mapped from L _i [ b , t ] according to each frequency band b, as shown in Figure 14b. The input broadband loudness is a set of functions that are specific to the loudness of the input audio signal. Although it may be the same as the specific loudness of the input audio signal, it may be a time smoothing and/or frequency smoothing version of the specific loudness of the audio signal.

The most straightforward way to calculate L _i [ b , t ] is to set it equal to the specific loudness N [ b , t ]. In this case, the DRC is performed independently on each frequency band in the auditory filter bank that senses the loudness mode, rather than on the same input and output loudness ratios on all frequency bands as previously described in "suitable for automatic gain and dynamic range control. The non-frequency and time-varying functions are described below under the heading. In a practical embodiment employing 40 bands, the spacing of these bands along the frequency axis is relatively small to provide an accurate measure of loudness. However, applying DRC scale coefficients independently to each frequency band may result in a perceived "cleavage" of the processed audio. In order to avoid this problem, it is optional to use a smoothness of the specific loudness N [ b , t ] across the frequency band to calculate L _i [ b , t ] such that the number of DRCs applied from the frequency band to the frequency band is not excessively changed. This can be achieved by defining a band-smoothing filter Q ( b ) and then smoothing the specific loudness across all bands c based on the standard convolution sum: Where N [ c , t ] is the specific loudness of the audio signal and Q(b-c) is the frequency shift response of the smoothing filter. Figure 15 shows an example of such a band smoothing filter.

If the DRC function for calculating L _i [ b , t ] by the function of L _o [ b , t ] is fixed in each frequency band b, the variation of each frequency band applied to the specific loudness N [ b , t ] will depend on the processed The spectrum of the audio, even if the overall loudness of the signal remains the same. For example, the bass of an audio signal with a loud bass and a small pitch may be removed and the treble is enhanced. Signals with smaller bass and louder treble may have the opposite situation. The final effect is a change in the perceived spectrum of sound quality or audio, which may be desirable in some applications.

However, it may be desirable to perform multi-band DRC without having to modify the average perceived spectrum of the audio. The user may wish that the average modification of each frequency band is substantially the same but still allows the modified short-term changes to operate independently between the frequency bands. The desired effect can be achieved by forcing the DRC's average operating state in each frequency band to be the same as some reference operating states. The user may select this reference DRC for the operating state to be L _i [ t ] required for wideband input loudness. Let the function L _o [ t ]= DRC { L _i [ t ]} represent the DRC mapping required for wide-band loudness. Then order Represents the time-average version of the broadband input loudness, and Represents the time-average version of the multi-band input loudness L _i [ b , t ]. The multi-band output loudness can then be calculated as

Note that the multi-band input loudness is first scaled to have the same average range as the wideband input loudness. The DRC function designed for wideband loudness is then applied. Finally, the resulting results are scaled to the average range of multi-band loudness. This type of multi-band DRC maintains the advantages of a reduced spectrum pump while preserving the average perceived spectrum of the audio.

Frequency-Varying and Time-Varying Functions Suitable for Dynamic Equalization Another application of the argument of the present invention is to intentionally convert the time-varying perceptual spectrum of the audio to the target non-time-varying spectrum while still retaining the original dynamic range of the audio. This process can be referred to as dynamic equalization (DEQ). In traditional static equalization, a simple set of fixed filters is applied to the audio to change its spectrum. For example, a fixed bass or treble boost may be applied. This processing does not take into account the current spectrum of the audio and may therefore not be suitable for some signals, i.e., signals that already contain a relatively large amount of bass or treble. With DEQ, the spectrum of the signal is measured and the signal is then dynamically modified to convert the measured spectrum to the form that is actually static. In the argument of the present invention, this desired form is defined across the frequency bands in the filter bank and is referred to as EQ [ b ]. In a practical embodiment, the measurement spectrum should represent the average spectral shape of the audio which can be generated by smoothing the specific loudness N [ b , t ] across time. This process can be called smoothing specific loudness [ b , t ]. As with multi-band DRC, it may not be desirable for DEQ modifications to vary drastically between bands, and thus a set of band-smoothing functions can be applied to produce a set of bands-smoothed spectrum :

In order to maintain the original dynamic range of the audio, the required spectrum EQ [ b ] should be normalized to have Represents the overall loudness of the same measured spectrum shape. This normalized spectrum shape can Indicates:

Finally, the target specific loudness is calculated as Where β is a set of user-specified parameters from zero to 1, indicating the degree of DEQ to be applied. Referring to Equation 23, when β = ₀ , the original specific loudness is not modified, and when β = ₁ , the specific loudness is scaled by the ratio of the desired spectral shape and the measured spectral shape.

A convenient way to generate the desired spectral shape EQ [ b ] is to have the user set it equal to A measure of the spectral balance of some audio segments that are comfortable for the user. In a practical embodiment, such as shown in Figure 16, the user may be provided with a set of buttons or other suitable actuators 507 that, when actuated, will capture the spectral shape of the current audio. The measurements are then stored and stored as a set of presets (in target specific loudness preset capture and storage 506) that can be loaded later upon DEQ priming (as preset selection 508) EQ [ b ]. Figure 16 is a simplified version of Figure 7, in which only a single set of lines is shown to represent the majority of the bands from analysis filter bank 100 to synthesis filter bank 110. The Figure 17 example also provides a set of dynamic EC specific loudness (SL) modifications 505 that provide a modification of the particular loudness measured by the function or device 104 in accordance with the dynamic equalization previously described.

Combining Processes A user may wish to combine all of the previously described processes, including volume control (VC), AGC, DRC, and DEQ, into a single system. Since these handlers can each be represented by a specific loudness scale, they can all be easily combined as follows: Where Ξ _* [ b , t ] represents the scale adjustment associated with the handler " ^* ". A single set of gains G [ b , t ] representing the target specific loudness of the combined processing can then be calculated.

In some cases, the scaling of a single or a combined loudness modification procedure may change rapidly over time and produce an artifact in the resulting processed audio. Therefore, some subsets of these scale adjustments need to be smoothed. In general, scale adjustments from VC and DEQ vary smoothly over time, but smoothing of the combination of AGC and DRC scale adjustments may be required. Let the combination of these scale adjustments be as follows

The basic concept of smoothing is that as the specific loudness increases, the combined scale adjustments should react quickly, and as the specific loudness decreases, the scale adjustments should be smoothed. This concept corresponds to the conventional implementation of fast impact and slow release employed in audio compressor design. The appropriate time constant for smoothing the scale factor can be calculated by smoothing the band smoothing version of the specific loudness over time. First the band-smoothed version of the specific loudness is calculated: Where is the specific loudness of the N [ c , t ] audio signal and Q(b-c) is the frequency shift response of the smoothing filter, as in Equation 19 above.

The time smoothed version of this band smoothing specific loudness is then calculated as The band-dependent smoothing coefficient λ[ b , t ] is as follows

The smoothed scale adjustment combination is then calculated as Where λ _M [ b , t ] is the band smoothing version of λ[ b , t ]:

The band smoothing of the smoothing coefficients prevents excessive changes in the spanning band of the time smoothing scale adjustment. The time and band smoothing of the above scale coefficients results in processing the audio containing less unpleasant perceptual artifacts.

Noise Compensation Background noise that interferes with the audio that the listener wishes to hear may be present in many audio replay environments. For example, a listener in a moving vehicle may play music in an installed stereo system and noise from the engine and the road surface can significantly change the perception of the music. In particular, the perceived loudness of the music will be reduced in the portion of the spectrum where the noise energy is significant relative to the music energy. If the noise level is large enough, the music will be completely masked.

In accordance with the teachings of the present invention, it will be desirable to select the gain G [ b , t ] such that the specific loudness of the processed audio is equal to the target specific loudness when the interfering noise is present. [ b , t ].

To achieve this effect, the concept of partial loudness can be used, as previously defined by Moore and Glasberg. It is assumed that the measurement of the noise itself and the measurement of the audio itself can be obtained. Let E _N [ b , t ] represent the excitation from the noise and let E _A [ b , t ] represent the excitation from the audio. The specific loudness of the audio and noise is combined as follows: N _TOT [ b , t ]=Ψ{ E _A [ b , t ]+ E _N [ b , t ]}, (31) where, again, Ψ{ . } represents a nonlinear transformation from excitation to specific loudness. It is assumed that the listener's hearing will be separated by a specific loudness that is separated by the specific loudness of the combination between the specific loudness of the audio and the specific loudness of the noise: N _TOT [ b , t ]= N _A [ b , t ]+ N _N [ b , t ]. (32)

Part of the specific loudness of the audio, N _A [ b , t ], is the value to be controlled, and therefore this value must be found. Part of the specific loudness of the noise can be approximated as Where E _TN [ b , t ] is the mask threshold when noise is present, E _TQ [ b ] is the smaller acoustic hearing threshold in band b, and K is the index between zero and one. Combining equations 31-33 gives an expression of the specific loudness of the audio:

Note that when the audio excitation is equal to the noise mask threshold ( E _A [ b , t ]= E _TN [ b , t ]), the partial loudness of the audio is equal to the signal loudness of the smaller sound limit, which is required the result of. When the audio excitation is significantly greater than the noise, the second set of values of Equation 34 will disappear and the specific loudness of the audio will be approximately equal to the loudness when the noise is not present. In other words, when the audio is louder than the noise, the noise will be masked by the audio. The K index is empirically selected to provide the appropriate degree of compliance of the tonal loudness data in the noise as a function of the signal to noise ratio. Moore and Glasberg found that a value of K = 0.3 is appropriate. The noise threshold of the noise can be approximated by the function of the noise excitation itself: E _TN [ b , t ]= K [ b ] E _N [ b , t ]+ E _TQ [ b ] (35) where K [ b ] is a set of constants that increase in the lower frequency band. Thus, the partial loudness of the audio given by Equation 34 can be abstractly represented as a function of the audio excitation and noise excitation: N _A [ b , t ]=Φ{ E _A [ b , t ], E _N [ b , t ]}. (36)

The modified gain solver can then be employed to calculate the gain G [ b , t ] such that when noise is present, a portion of the specific loudness of the processed audio is equal to the target specific loudness:

Figure 17 shows a system of Figure 7, in which the original gain solver 106 is replaced by the above-described noise compensation gain solver 206 (note that multiple sets of vertical lines between the blocks represent that the multiple sets of bands of the filter bank have been replaced by single lines To simplify the graphics). In addition, the graph shows the measurement of the noise excitation (using the analysis filter bank 200, the transmission filter 201, the excitation 202, and the smoothing 203 to correspond to the operation of blocks 100, 101, 102, and 103) and the audio excitation (from Smoothing 103) is fed into the new gain solver 206 along with the target specific loudness (from the SL modification 105).

In its basic mode of operation, the SL modification unit 105 in FIG. 17 can simply target the specific loudness. [ b , t ] is set equal to the original specific loudness of the audio N [ b , t ]. In other words, the SL modification unit provides a set of non-frequency-variant, alpha-scale adjustments of a particular loudness of the audio signal, where α=1. As in the configuration of Figure 17, the gain is calculated such that when the noise is present, the perceived loudness spectrum of the processed audio is equal to the audio loudness spectrum when the noise is not present. Additionally, the previously specified target specific loudness is calculated as a function of the original loudness, and any one or combination of VC, AGC, DRC, and DEQ techniques can be employed in conjunction with the noise compensated loudness modification system.

In a practical embodiment, the noise measurement can be obtained from a microphone placed in the environment in which the audio is to be played or in the vicinity. Additionally, a predetermined set of sample noise excitations that approximate the expected noise spectrum in various situations can be employed. For example, noise in the vehicle can be pre-analyzed at various drive rates and then stored as a table of noise excitation and relative rates. The noise excitation that is fed into the gain solver 206 in Fig. 17 can then be approximated from the look-up table as the vehicle speed changes.

Fabrication The invention can be fabricated in hardware or software, or a combination of both (e.g., a programmable logic array). Except as specified, algorithms included as part of the invention are not inherently related to any particular computer or other device. In particular, it may be more convenient for various general purpose machines to be used in conjunction with programs made in accordance with the techniques herein, or to make more specialized devices (e.g., integrated circuits) to perform the desired method steps. Accordingly, the present invention can be made in one or more computer programs executed on one or more programmable computer systems, each computer system comprising at least one processor, at least one data storage system (including electrical and non-compliant) An electrical memory and/or storage element), at least one input device or device, and at least one output device or device. The code is applied to the input data to perform the above functions and to generate output information. The output information is applied to one or more output devices in a conventional form.

Each such group of programs can be made to communicate with a computer system using any desired computer language (including machine, combination, or high level program, logic, or object oriented programming language). In any case, the language can be a compiled or translated language.

Preferably, each such computer program is stored or downloaded to a set of storage media or devices (eg, solid state memory or media, or magnetic or optical media) that can be read by a general or special purpose programmable computer. The computer system reads the storage medium or device and performs the above steps to configure and operate the computer. The system of the present invention also contemplates a computer-readable storage medium that is configured to be programmed in a computer program, wherein the storage medium is configured such that the computer system operates in a specific and predetermined manner.

Some embodiments of the invention have been described. However, it will be appreciated that various modifications may be made without departing from the spirit and scope of the invention. For example, the order of some of the above steps is independent, and thus may be performed in a different order as described above.

2. . . Modify the audio signal

4, 4', 4", 4'''... generate modified parameters

6. . . Calculate target specific loudness

8. . . Calculate specific loudness

10, 10', 10", 10'''... calculate the modified parameters

12. . . Calculate the approximation of the specific loudness of the unmodified audio

14. . . Calculate the approximation of the target specific loudness

16. . . Store or transfer

100. . . Analysis filter bank

101. . . Transmission filter

102. . . excitation

102. . . excitation

103. . . Smoothing

104. . . Specific loudness (SL)

105. . . SL modification

106. . . Gain solver

107. . . Selective smoothing

108. . . Combiner

109. . . delay

110. . . Synthesis filter bank

200. . . Analysis filter bank

201. . . Transmission filter

202. . . excitation

203. . . Smoothing

206. . . Noise compensation gain solver

505. . . Dynamic EQ specific loudness modification

506. . . Target specific loudness preset capture and storage

507. . . User capture selection

508. . . User preset selection

Figure 1 shows a set of functional block diagrams of a pre-production paradigm in accordance with one aspect of the present invention.

Figure 2 shows a set of functional block diagrams of a feedback production paradigm in accordance with one aspect of the present invention.

Figure 3 is a block diagram showing a functional group of a hybrid pre-commissioning/responsibility production example in accordance with one aspect of the present invention.

Figure 4 shows a set of functional block diagrams of another hybrid pre-/re-sales production paradigm in accordance with one aspect of the present invention.

Figure 5 is a set of functional block diagrams showing the manner in which unmodified audio signals and modified parameters determined by any set of pre-grant, feedback, and hybrid pre-request configurations can be stored or transmitted, such as For use in devices or handlers that are separated in time or space.

Figure 6 is a set of functional block diagrams showing that unmodified audio signals and target specific loudness or their representations determined by any set of pre-grant, feedback, and hybrid pre-requested configurations can be stored or transmitted. The manner, for example, is for use in a device or process that is separated in time or space.

Figure 7 is a block diagram showing an exploded view or an exploded flow chart of an overview of one of the arguments of the present invention.

Figure 8 is an ideal characteristic response of a linear filter P(z) suitable as a set of transmission filters in an embodiment of the present invention, wherein the vertical axis is decibel attenuation (dB) and the horizontal axis is in Hertz (Hz) The logarithmic base of the unit is 10 frequencies.

Figure 9 shows the relationship between the ERB frequency scale (vertical axis) and the frequency in Hertz (horizontal axis).

Figure 10 shows a set of idealized auditory filter characteristic responses that approximate a critical band on the ERB scale. The horizontal scale is the frequency in Hertz and the vertical scale is the decibel level.

Figure 11 shows the equal loudness contour of ISO 226. The horizontal scale is the frequency in Hertz (the logarithmic base 10 scale) and the vertical scale is the sound pressure level in decibels.

Figure 12 shows the equal loudness contour of ISO 226 normalized by the transmission filter P(z). The horizontal scale is the frequency in Hertz (the logarithmic base 10 scale) and the vertical scale is the sound pressure level in decibels.

Figure 13a shows an ideal plot of wideband and multiband gain with a loudness scale adjusted to 0.25 for a segment of female speech. The horizontal scale is the ERB band and the vertical scale is the relative gain in decibels (dB).

Figure 13b shows an ideal plot of the original signal, the wideband gain modified signal, and the specific loudness of the multiband gain modified signal, respectively. The horizontal scale is the ERB band and the vertical scale is the specific loudness (Song/ERB).

FIG 14a is a set of ideal view showing: Usually when AGC, L _o [t] to L _i [t] is a set of functions. The horizontal scale is log( L _i [ t ]) and the vertical scale is log ( L _o [ t ]).

FIG 14b is a set of idealized view showing: Usually when DRC, L _o [t] to L _i [t] is a set of functions. The horizontal scale is log( L _i [ t ]) and the vertical scale is log ( L _o [ t ]).

Figure 15 shows an ideal plot of the general band smoothing function for multi-band DRC. The horizontal scale is the number of bands and the vertical scale is the gain output of band b.

Figure 16 is a block diagram showing an exploded view or an exploded flow chart of an overview of one of the arguments of the present invention.

Figure 17 is similar to Figure 1 which also contains a decomposition function block diagram or decomposition flow diagram of the noise compensation in the replay environment.

2. . . Modify the audio signal

4. . . Generate modified parameters

6. . . Calculate target specific loudness

8. . . Calculate specific loudness

10. . . Calculate the modified parameters

Claims (64)

A method for controlling a particular loudness characteristic of an audio signal, wherein the specific loudness characteristic is a specific loudness measured as one of a perceived loudness as a function of frequency and time, or a second interfering signal is present As part of the measurement of the perceived loudness of the signal as a function of frequency and time, the method comprises the steps of: a) calculating a target specific loudness as a function of the audio signal; b) deriving the use to modify the audio Signaling to reduce frequency and time variation modification parameters between the particular loudness characteristic and the target specific loudness, and c) applying the modified parameters to the audio signal to reduce its specific loudness characteristic and the target specific loudness difference. According to the method of claim 1, wherein the step of calculating a target specific loudness as a function of the audio signal or the deriving frequency and time variation modifying parameter comprises explicitly calculating the specific loudness and/or the partial loudness step. The method of claim 1, wherein the step of calculating a target specific loudness as a function of the audio signal or the deriving frequency and time variation modifying parameter comprises implicitly calculating the specific loudness and/or partial loudness Processing steps. The method of claim 3, wherein the processing step employs a look-up table such that the processing step inherently determines the loudness and/or partial loudness. The method of claim 3, wherein the specific loudness and/or the partial loudness is inherently determined by the mathematical expression of the closed form employed in the processing step. The method of any one of claims 1 to 5, wherein the function of the audio signal used to calculate the target specific loudness comprises one or more scale adjustments of the audio signal. According to the method of claim 6, wherein the one or more scale adjustments comprise a time and frequency variation scale factor Ξ[b,t] adjusting the specific loudness according to the following relationship scale: among them( [ b , t ]) is the target specific loudness, ( N [ b , t ]) is the specific loudness of the audio signal, b is one of the frequencies, and t is one of the time measurements. The method of claim 7, wherein the one or more scale adjustments are determined at least in part by a ratio of desired multi-band loudness and multi-band loudness of the audio signal. According to the method of claim 8, wherein the one or more scale adjustments can be expressed as L _o [b, t] / L _i [b, t] in the following relationship: Where N [ b , t ] is the specific loudness of the audio signal, L _o [b, t] is the desired multi-band loudness, L _i [b, t] is the multi-band loudness of the audio signal, and [ b , t ] is the target specific loudness. According to the method of claim 9, wherein L _o [b, t] is a function of L _i [b, t]. A method according to Paragraph 10 of patent application range, where as _{L i [b, t] L} o [b, t] one of the functions may be expressed _{as: L o [b, t]} = DRC {L i [b, t ]}, where DRC{} represents a dynamic range function that maps L _i [b, t] to L _o [b, t]. The method of claim 9, wherein L _i [b, t] is a time-smoothing and/or frequency smoothing version of the specific loudness of the audio signal. A method according to any one of claims 8 to 12, wherein the method is operable as a dynamic range control, wherein the application of the modification or the modified parameter produces or the target specific loudness corresponds to an audio message The signal, in which the perceived audio spectrum is present or the perceived audio spectrum present in the interfering signal, may vary for different values than the loudness scale adjustment. The method of claim 13, wherein the dynamic range function controls loudness in each frequency band such that short-term changes applied to respective frequency bands vary independently between frequency bands, and are applied to the average change of each frequency band for all The frequency bands are substantially the same. According to the method of claim 14, wherein L _o [b, t] as a function of L _i [b, t] can be expressed as: Where L _o [ t ]= DRC { L _i [ t ]} represents a mapping of the total loudness of the audio signal to the desired total loudness, [ t ] represents a time-average version of the wide-band loudness L _i [t] of the audio signal, and [ b , t ] represents a time-averaged version of the multi-band loudness L _i [b, t] of the audio signal. The method of any one of claims 14 or 15, wherein the method is operable as a dynamic range control, wherein the modification or the application of the modified parameter produces or the target specific loudness corresponds to an audio signal, The ratio of the perceived audio spectrum of the perceived audio spectrum or the presence of an interfering signal to different values is substantially the same as the perceived audio spectrum of the audio signal. The method of claim 7, wherein the specific loudness is scaled by a ratio of a desired spectral shape measurement to a spectral shape measurement of the one of the audio signals. The method of claim 17, wherein the method converts the perceptual spectrum of the audio signal from a time-varying perceptual spectrum to a substantially non-time-variant perceptual spectrum. According to the method of claim 17 or 18, wherein the one or more scale adjustments can be expressed in the following relationship item: And where [ b , t ] is a time-smoothed multi-band loudness of the audio signal, [ b , t ] is a required spectrum EQ[b], which is normalized to have, for example, multi-band loudness [ b , t ] the same wide-band loudness, thus [ b , t ] can be expressed as: Where N [ b , t ] is the specific loudness of the audio signal, [ b , t ] is the target specific loudness, and β is one of the parameters having a range containing zero and one and being limited by it, the parameter controlling the scale adjustment level. According to the method of claim 19, wherein the parameter β is selected or controlled by a source external to the method. The method of claim 20, wherein the source is a user of the method. The method according to any one of claims 17 to 21, wherein the method is usable as a dynamic equalizer, wherein the modification or the application of the modified parameter produces or the target specific loudness corresponds to an audio The signal, in which the perceived audio spectrum is present or the perceived audio spectrum present in the interfering signal, may vary for different loudness scale adjustments for different values. The method of any one of claims 7 to 22, wherein the multi-band loudness of the audio signal is obtained by dividing the audio signal into a plurality of primary frequency bands and performing frequency smoothing across a plurality of the primary frequency bands Approximate. The method according to claim 23, wherein the multi-band loudness in one of the specific frequency bands b-smoothed version L [ b , t ] can be expressed as a convolution sum across all frequency bands c: Where N [ c , t ] is the specific loudness of the audio signal, and Q(bc) is the frequency band-shift response of the smoothing filter. According to the method of claim 6, wherein the one or more scale adjustments comprise a time varying non-frequency variable scale coefficient Φ[t] scale adjustment of the specific loudness, such as the following relationship: among them( [ b , t ]) is the target specific loudness, ( N [ b , t ]) is the specific loudness of the audio signal, b is the frequency measurement, and t is the time measurement. The method of claim 25, wherein the one or more scale adjustments are determined at least in part by a desired ratio of broadband loudness to wideband loudness of the audio signal. According to the method of claim 25 or 26, wherein the scale adjustment in the function of the specific loudness of the audio signal can be expressed as L _o [t] / L _i [t] in the following relationship: Where N [ b , t ] is the specific loudness of the audio signal, L _o [t] is the desired wide-band loudness, L _i [t] is the wide-band loudness of the audio signal, and [ b , t ] is the target specific loudness. According to the method of claim 27, wherein L _o [t] is a function of L _i [t]. According to the method of claim 28, wherein L _o [t] as a function of L _i [t] can be expressed as: L _o [ t ]= DRC { L _i [ t ]} where DRC{} represents a mapping Dynamic range function of L _i [t] to L _o [t]. According to the method of claim 27, wherein L _i [t] is a time-smoothed version of the total loudness of the audio signal. According to the method of claim 27, wherein L _i [t] is one of the long-term loudness of the audio signal. According to the method of claim 27, wherein L _i [t] is one of the short-term loudness of the audio signal. The method of any one of claims 25 to 32, wherein the method can be used as an automatic gain control or dynamic range control, wherein the modification or the application of the modified parameters produces or the target corresponds to loudness In the audio signal, the perceptual audio spectrum in which the audio spectrum is perceived or present in an interference signal, the specific loudness scale adjustment or the partial loudness scale adjustment for different values remains substantially the same as the perceived audio spectrum of the audio signal. The method of any one of clauses 7 to 33, wherein the time and frequency variation scale factor is a function of the measurement of the audio signal or the audio signal. According to the method of claim 6, wherein the one or more scale adjustments comprise a non-time variable frequency scaling coefficient Θ[ b ] adjusting the specific loudness by the following relationship scale: among them [ b , t ] is the target specific loudness, N [ b , t ] is the specific loudness of the audio signal, b is the frequency measurement, and t is the time measurement. The method of claim 35, wherein the modifying or the deriving step comprises storing the scale factor Θ[ b ]. According to the method of claim 35, wherein the scale factor Θ[ b ] is a source received from outside the method. According to the method of claim 6, wherein the one or more scale adjustments comprise a non-time varying non-frequency variable scale coefficient a to adjust the specific loudness of the audio signal by the following relationship scale: among them [ b , t ] is the target specific loudness, N [ b , t ] is the specific loudness of the audio signal, b is the frequency measurement, and t is the time measurement. The method of claim 38, wherein the modifying or the deriving step comprises storing the scale factor α . A method according to any one of claims 35 to 38, wherein the method is operable as a volume control, wherein the application of the modification or the modified parameter produces or the target specific loudness corresponds to an audio signal, The perceived audio spectrum in which the perceived audio spectrum or an interfering signal is present remains substantially the same as the perceived audio spectrum of the audio signal for different values of the specific loudness or partial loudness scale adjustment. The method of any one of claims 1 to 40, wherein the modifying step or the application of the modifying parameter, or the deriving step, explicitly calculates (a) specific loudness, and/or (b) partial ratio Loudness, and/or (c) the target's specific loudness. A method according to any one of claims 1 to 40, wherein the modification step or the application of the modification parameter, or the derivation step implicitly calculates (a) specific loudness, and/or (b) portion Specific loudness, and / or (c) the target specific loudness. According to the method of claim 42, wherein the modification step or the application of the modification parameter, or the derivation step of the modification parameter, or the generation step adopts a comparison table, which inherently determines (a) specific loudness, and/or Or (b) partial loudness, and/or (c) the target specific loudness. According to the method of claim 43, wherein the modification step or the application of the modification parameter, or the derivation step of the modification parameter, or the generating step adopts a closed form mathematical expression, which inherently determines (a) specific loudness , (b) and / or part of the specific loudness, and / or (c) the target specific loudness. The method of any one of claims 1 to 44, wherein the modified parameters are temporally smoothed. The method of claim 45, wherein the modified parameters comprise a plurality of amplitude scale adjustment coefficients for a frequency band of the audio signal. According to the method of claim 46, at least some of the plurality of amplitude scale adjustment coefficients are time-variant. The method of claim 45, wherein the modified parameters comprise a plurality of filter coefficients for controlling one or more filters. The method of claim 48, wherein at least some of the one or more filters are time-varying, and at least some of the filter coefficients are time-varying. The method of any one of claims 1 to 49, wherein the modifying step or the application of the modifying parameter, the deriving step of the modifying parameter, or the generating step is one or more of the following : measuring one of the interfering audio signals, a target specific loudness, An estimate of the specific loudness of the unmodified audio signal from the loudness or partial loudness of the modified audio signal, the specific loudness of the unmodified audio signal, and the ratio derived from the modified audio signal An approximation of the loudness or part of the loudness of the target than the loudness. The method of any one of claims 1 to 49, wherein the modifying step or the applying of the modifying parameter or the deriving step of the modifying parameter derives at least in part from one or more of the following: Measuring one of the interfering audio signals, a target specific loudness, an evaluation value derived from the specific loudness of the modified audio signal or the specific loudness of the unmodified audio signal of the partial loudness, the unmodified audio signal The specific loudness, and an approximation of the target specific loudness from the loudness or partial loudness of the modified audio signal. The method of claim 51, wherein the modifying step or the applying of the modifying parameter or the deriving step of the modifying parameter derives at least in part a modified parameter from: (1) one of the following: a target specific loudness And an evaluation value of one of the specific loudness of the unmodified audio signal received from the loudness of the modified audio signal, and (2) one of the following: The specific loudness of the unmodified audio signal and an approximation of the target specific loudness derived from the loudness of the modified audio signal. The method of claim 51, wherein the modifying step or the applying of the modifying parameter or the deriving step of the modifying parameter derives at least in part a modified parameter from: (1) measuring one of the interfering audio signals, (2) one of the following: a target specific loudness, and an estimate of the specific loudness of the unmodified audio signal derived from the partial loudness of the modified audio signal, and (3) one of the following: The specific loudness of the unmodified audio signal and an approximation of the target specific loudness of the partial loudness of the modified audio signal. The method of claim 52, wherein the method employs a pre-administration configuration, wherein the specific loudness is derived from the audio signal, and wherein the target specific loudness is received from one of the external sources of the method or The modification step or the derivation step of the application or modification parameter of the modification parameter includes receiving from a storage location when storing a target specific loudness. The method of claim 52, wherein the method employs a hybrid pre-request/reward configuration, wherein an approximation of the target specific loudness is derived from the modified audio signal, and wherein the target specific loudness is Received from one of the external sources of the method or when the modification step or the repair The application of the modified parameter or the derivation of the modified parameter includes receiving from a storage when storing a target specific loudness. The method of claim 52, wherein the modifying step or the applying of the modifying parameter or the deriving step of the modifying parameter comprises one or more processing programs to obtain the target specific loudness explicitly or implicitly, wherein The one or more processing programs explicitly or implicitly calculate the function of the audio signal or the measurement of the audio signal. According to the method of claim 56, wherein the method adopts a pre-administration configuration, wherein the specific loudness and the target specific loudness are derived from the audio signal, and the target specific loudness is derived by using the audio signal or the audio signal The measurement of this function. According to the method of claim 56, wherein the method employs a hybrid pre-request/reward configuration, wherein an approximation of the target specific loudness is derived from the modified audio signal, and the target specific loudness is from the audio The signal is derived, and the target is compared to the loudness by the function of the audio signal or the measurement of the audio signal. The method of claim 52, wherein the modifying step or the applying of the modifying parameter or the deriving step of the modifying parameter comprises one or more processing programs to explicitly or implicitly respond to the modified audio signal Obtaining an estimate of the loudness of the unmodified audio signal, wherein the one or more processing programs explicitly or implicitly calculate an inverse function of the function of the audio signal or the measurement of the audio signal. According to the method of claim 59, wherein the method adopts a feedback configuration, wherein one of the specific loudness of the unmodified audio signal is evaluated The estimate, and an approximation of the target specific loudness, is derived from the modified audio signal, the estimate of the loudness being calculated using an inverse function of the function of the audio signal or the measurement of the audio signal. The method of claim 59, wherein the method employs a hybrid pre-request/return configuration, wherein the specific loudness is derived from the audio signal, and the evaluation value of the specific loudness of the unmodified audio signal Derived from the modified audio signal, the derivation of the evaluation value is calculated using the inverse of the function of the audio signal or the measurement of the audio signal. The method of any one of claims 1 to 61, wherein the step a) further comprises transmitting or storing the audio signal and the calculated target specific loudness or its expression, and step b) further comprises The frequency and time variation is received prior to modifying the parameter to receive the transmitted or regenerated stored audio signal and the target specific loudness or its expression, and the execution of step a) is temporally and/or spatially separated from steps b) and c). A device for performing all the steps of the method, the method of any one of claims 1 to 61, the device comprising: computing means for calculating a target specific loudness as the audio signal a function; a derivation device whose derivation can be used to modify the audio signal to reduce frequency and time variation modification parameters between the particular loudness characteristic and the target specific loudness, and An application device that applies the modified parameters to the audio signal to reduce the difference between its specific loudness characteristics and the target specific loudness. A computer program stored on a computer readable medium, adapted to cause a computer to perform all of the steps of the method of any one of claims 1 to 61.