CN101185120B - Systems, methods, and apparatus for highband burst suppression - Google Patents

Abstract

In one embodiment, a highband burst suppressor includes a first burst detector configured to detect bursts in a lowband speech signal, and a second burst detector configured to detect bursts in a corresponding highband speech signal. The lowband and highband speech signals may be different (possibly overlapping) frequency regions of a wideband speech signal. The highband burst suppressor also includes an attenuation control signal calculator configured to calculate an attenuation control signal according to a difference between outputs of the first and second burst detectors. A gain control element is configured to apply the attenuation control signal to the highband speech signal. In one example, the attenuation control signal indicates an attenuation when a burst is found in the highband speech signal but is absent from a corresponding region in time of the lowband speech signal.

Description

Systems, methods, and devices for high-band burst suppression

This application claims the benefit of U.S. Provisional Patent Application No. 60/667,901, filed April 1, 2005, entitled "CODING THE HIGH-FREQUENCYBAND OF WIDEBAND SPEECH." This application also claims the benefit of U.S. Provisional Patent Application No. 60/673,965, filed April 22, 2005, entitled "PARAMETER CODING IN A HIGH-BANDSPEECH CODER."

technical field

The present invention relates to signal processing.

Background technique

Voice communication over the Public Switched Telephone Network (PSTN) has traditionally been limited in bandwidth to the frequency range of 300-3400 kHz. Newer networks for voice communications such as cellular telephones and voice over IP (VoIP) may not have the same bandwidth limitations, and voice communications may need to be transmitted and received over such networks including wideband frequency ranges. For example, it may be desirable to support an audio range extending down to 50Hz and/or up to 7 or 8kHz. There may also be a need to support other applications, such as high quality audio or audio/video conferencing, that may have audio voice content outside the limits of traditional PSTNs.

The extension of the range supported by the speech codec to higher frequencies improves intelligibility. For example, the information to distinguish fricatives such as "s" and "f" is mainly in high frequencies. High-band extension may also improve other qualities of speech, such as realism. For example, even vocalized vowels may have spectral energy well above the PSTN limit.

In the course of research on wideband speech signals, the inventors happened to observe high energy pulses or "bursts" in the upper part of the frequency spectrum. These high-band bursts typically last only a few milliseconds (typically 2 milliseconds, with a maximum length of about 3 milliseconds), can span up to several kilohertz (kHz) in frequency, and occur in different types of speech sounds (both vocalized and unvoiced). or) appears to occur randomly. For some speakers, high-band bursts may occur in every sentence, while for other speakers such bursts may not occur at all. Although these events generally do not occur frequently, they do appear to be ubiquitous, as the inventors have found instances of them in wideband speech samples from several different databases and from several other sources.

High-band bursts have a wide frequency range, but typically only occur in the upper frequency bands of the spectrum (eg, the 3.5 to 7 kHz region) and not in the lower frequency bands. For example, FIG. 1 shows a spectrogram of the word "energy". In this wideband speech signal, a high frequency band burst can be observed at 0.1 s, extending over a broad frequency region around 6 kHz (darker areas in this figure indicate higher intensities). It is possible that at least some of the high frequency band bursts result from the interaction between the speaker's mouth and the microphone, and/or are due to clicks made by the speaker's mouth during speech.

Contents of the invention

According to one embodiment, a signal processing method includes: processing a wideband speech signal to obtain a low-band speech signal and a high-band speech signal; determining that bursts exist in an area of the high-band speech signal; and determining the low-band speech signal There is no burst in the corresponding region of . The method also includes attenuating high-band speech signals over the region based on determining the burst is present and based on determining the burst is not present.

According to an embodiment, an apparatus includes: a first burst detector configured to detect bursts in a low-band speech signal; a second burst detector configured to detect bursts in a corresponding high-band speech signal an attenuation control signal calculator configured to calculate the attenuation control signal based on the difference between the output of the first burst detector and the output of the second burst detector; and a gain control element configured to apply the attenuation control signal to the high-band speech signal.

Description of drawings

Figure 1 shows a spectrogram of a signal comprising high-band bursts.

Figure 2 shows a spectrogram of a signal in which high-band bursts have been suppressed.

FIG. 3 shows a block diagram of an arrangement comprising a filter bank A110 and a high-band burst suppressor C200 according to an embodiment.

4 shows a block diagram of an arrangement including filter bank A110, high-band burst suppressor C200, and filter bank B120.

5a shows a block diagram of an implementation A112 of filter bank A110.

5b shows a block diagram of an implementation B122 of filter bank B120.

Figure 6a shows the bandwidth coverage of the low and high bands for one example of filter bank A110.

FIG. 6b shows the bandwidth coverage of the low-band and high-band for another example of filter bank A110.

6c shows a block diagram of an implementation A114 of filter bank A112.

Figure 6d shows a block diagram of an implementation B124 of filter bank B122.

FIG. 7 shows a block diagram of an arrangement comprising a filter bank A110 , a highband burst suppressor C200 and a highband speech encoder A200 .

8 shows a block diagram of an arrangement comprising filter bank A110, high-band burst suppressor C200, filter bank B120, and wideband speech encoder A100.

FIG. 9 shows a block diagram of wideband speech encoder A102 including highband burst suppressor C200.

10 shows a block diagram of an implementation A104 of wideband speech encoder A102.

FIG. 11 shows a block diagram of an arrangement comprising wideband speech encoder A104 and multiplexer A130.

12 shows a block diagram of an implementation C202 of high-band burst suppressor C200.

13 shows a block diagram of an implementation C12 of burst detector C10.

Figures 14a and 14b show block diagrams of implementations C52-1, C52-2 of an initial region indicator C50-1 and an ending region indicator C50-2, respectively.

15 shows a block diagram of an implementation C62 of coincidence detector C60.

16 shows a block diagram of an implementation C22 of attenuation control signal generator C20.

17 shows a block diagram of an implementation C14 of burst detector C12.

18 shows a block diagram of an implementation C16 of burst detector C14.

19 shows a block diagram of an implementation C18 of burst detector C16.

20 shows a block diagram of an implementation C24 of attenuation control signal generator C22.

Detailed ways

Unless clearly limited by the context, the term "calculate" is used herein to indicate any of its ordinary meanings, such as calculating, generating, and selecting from a list of values. When the term "comprising" is used in the present description and claims, it does not exclude other elements or operations.

High-band bursts are perfectly audible in the original speech signal, but they do not contribute to intelligibility, and by suppressing them the signal quality can be improved. High-band bursts can also be detrimental to the encoding of high-band speech signals, so that by suppressing bursts from high-band speech signals the efficiency of encoding the signal and in particular the efficiency of encoding the temporal envelope can be improved.

Highband bursts can negatively impact highband encoding systems in several ways. First, these bursts can make the speech signal energy envelope much less smooth over time by introducing spikes in the burst. Unless the encoder simulates the temporal envelope of the signal at high resolution (which increases the amount of information to be sent to the decoder), the burst energy may smear in the decoded signal over time and cause glitches. Second, high-band bursts tend to dominate in the spectral envelope as modeled by, for example, a set of parameters (eg linear prediction filter coefficients). This simulation is typically performed for each frame (about 20 milliseconds) of the speech signal. Thus, click-containing frames may be synthesized differently from previous and subsequent frames according to the spectral envelope, which may result in perceptually unpleasant discontinuities.

Highband bursting can cause another problem for speech coding systems where the excitation signal for the highband synthesis filter is derived from or otherwise represents a narrowband residual. In such cases, the presence of high-band bursts can complicate the encoding of high-band speech signals, since high-band speech signals include structures not present in narrow-band speech signals.

Embodiments include systems, methods, and devices configured to detect bursts that are present in a high-band speech signal but not in a corresponding low-band speech signal and reduce the level of the high-band speech signal during each burst . Potential advantages of such embodiments include avoiding artifacts in the decoded signal and/or avoiding loss of coding efficiency without significantly degrading the quality of the original signal. Fig. 2 shows a spectrogram of the wideband signal shown in Fig. 1 after suppressing high-band bursts according to this method.

FIG. 3 shows a block diagram of an arrangement comprising a filter bank A110 and a high-band burst suppressor C200 according to an embodiment. Filter bank A110 is configured to filter wideband speech signal S10 to generate lowband speech signal S20 and highband speech signal S30. The high-band burst suppressor C200 is configured to output a processed high-band speech signal S30a based on the high-band speech signal S30, wherein bursts that occur in the high-band speech signal S30 but do not exist in the low-band speech signal S20 Hair has been suppressed.

FIG. 4 shows a block diagram of the arrangement shown in FIG. 3 that also includes filter bank B120. Filter bank B120 is configured to combine low-band speech signal S20 with processed high-band speech signal S30a to generate processed wideband speech signal S10a. The quality of the processed wideband speech signal S10a may be improved compared to the quality of the wideband speech signal S10 due to the suppression of high frequency band bursts.

Filter bank A110 is configured to filter the input signal according to a split-band scheme to produce low frequency sub-bands and high frequency sub-bands. Depending on application-specific design criteria, the output sub-bands may have equal or unequal bandwidths, and may or may not overlap. Configurations of the filter bank A110 that generate more than two sub-bands are also possible. For example, the filter bank may be configured to generate a very low-band signal that includes components in a frequency range lower than that of the narrowband signal S20 (eg, the range of 50-300 Hz). In this case, wideband speech encoder A100 (see description of FIG. 8 below) is implemented to separately encode this very low frequency band signal, and multiplexer A130 (see description of FIG. 11 below) can be configured to The encoded very low frequency band signal is included in the multiplexed signal S70 (eg, as a separable part).

5a shows a block diagram of an implementation A112 of filter bank A110 configured to generate two subband signals with a reduced sampling rate. Filter bank A110 is arranged to receive wideband speech signal S10 having a high frequency (or high band) part and a low frequency (or low band) part. Filter bank A112 includes a low-band processing path configured to receive wideband speech signal S10 and generate low-band speech signal S20, and a high-band processing path configured to receive wideband speech signal S10 and generate high-band speech signal S30. The low pass filter 110 filters the wideband speech signal S10 to pass selected low frequency sub-bands, and the high pass filter 130 filters the wideband speech signal S10 to pass selected high frequency subbands. Since the two sub-band signals have a narrower bandwidth than the wideband speech signal S10, their sampling rate can be reduced to a certain extent without loss of information. Decimator 120 reduces the sampling rate of the low-pass signal by a desired decimation factor (e.g., by removing signal samples and/or replacing samples with an average value), and decimator 140 similarly decimates by another desired decimation factor. Reduce the sampling rate of the high-pass signal.

5b shows a block diagram of a corresponding implementation B122 of filter bank B120. Up-sampler 150 increases the sampling rate of low-band speech signal S20 (e.g., by zero padding and/or by duplicating samples), and low-pass filter 160 filters the up-sampled signal so that only the low-band portion passes (e.g., , to prevent aliasing). Also, the up-sampler 170 increases the sampling rate of the processed high-band signal S30a, and the high-pass filter 180 filters the up-sampled signal to pass only the high-band portion. The two passband signals are then summed to form wideband speech signal S10a. In some embodiments of an apparatus comprising filter bank B 120, filter bank B 120 is configured to generate a weighted sum of two passband signals based on one or more weights received and/or calculated by the apparatus . Configurations of filter bank B 120 that combine more than two passband signals are also contemplated.

Each of the filters 110, 130, 160, 180 may be implemented as a finite impulse response (FIR) filter or as an infinite impulse response (IIR) filter. The frequency responses of filters 110 and 130 may have symmetrical or differently shaped transition regions between stopbands and passbands. Likewise, the frequency responses of filters 160 and 180 may have symmetrical or differently shaped transition regions between stopbands and passbands. It may be desirable (but not strictly necessary) for low-pass filter 110 to have the same response as low-pass filter 160 and for high-pass filter 130 to have the same response as high-pass filter 180 . In one example, the two filter pairs 110 , 130 and 160 , 180 are quadrature mirror filter (QMF) banks, where the filter pair 110 , 130 has the same coefficients as the filter pair 160 , 180 .

In a typical example, the low pass filter 110 has a passband (eg, a frequency band of 0 to 4 kHz) that includes a limited PSTN range of 300-3400 Hz. Figures 6a and 6b show the relative bandwidths of the wideband speech signal S10, the lowband speech signal S20 and the highband speech signal S30 in two different implementation examples. In these two specific examples, the wideband speech signal S10 has a sampling rate of 16 kHz (representing frequency components in the range 0 to 8 kHz), and the low-band signal S20 has a sampling rate of 8 kHz (representing frequency components in the range 0 to 4 kHz). frequency components).

In the example of Fig. 6a, there is no significant overlap between the two sub-bands. The high-band signal S30 shown in this example can be obtained using a high-pass filter 130 with a passband of 4-8 kHz. In this case, it may be necessary to reduce the sampling rate to 8kHz by downsampling the filtered signal by a factor of two. This operation is expected to significantly reduce the computational complexity of further processing operations on the signal that will shift the passband energy down to the 0 to 4 kHz range without loss of information.

In the alternative example of Fig. 6b, the upper sub-band has a significant overlap with the lower sub-band such that the two sub-band signals describe the 3.5 to 4 kHz region. The high-band signal S30 in this example can be obtained using a high-pass filter 130 with a passband of 3.5-7 kHz. In this case, it may be necessary to reduce the sampling rate to 7kHz by downsampling the filtered signal by a factor of 16/7. This operation is expected to significantly reduce the computational complexity of further processing operations on the signal that will shift the passband energy down to the 0 to 3.5 kHz range without loss of information.

In a typical cell phone used for telephone communications, there is no appreciable response in the frequency range of 7-8 kHz for one or more transducers (ie, microphone and earpiece or speaker). In the example of Fig. 6b, the part of the wideband speech signal S10 between 7 and 8 kHz is not included in the encoded signal. Other specific examples of high pass filter 130 have passbands of 3.5-7.5 kHz and 3.5-8 kHz.

In some embodiments, providing an overlap between sub-bands as in the example of Fig. 6b allows the use of low-pass and/or high-pass filters with smooth roll-off over the overlapping region. Such filters are typically less computationally complex and/or introduce less delay than filters with steeper or "brick-wall" responses. Filters with steep transition regions tend to have higher sidelobes (which can lead to aliasing) than similar order filters with smooth rolloffs. Filters with steep transition regions can also have long impulse responses that can cause ringing glitches. For filter bank implementations with one or more IIR filters, allowing smooth rolloff over overlapping regions can enable the use of filter(s) with poles far from the unit circle, which is important for ensuring stable fixed-point Implementation can be important.

The overlapping of the sub-bands allows for a smooth mixing of the low and high frequency bands, which results in less audible artifacts, reduced aliasing and/or less pronounced transitions from one frequency band to another. Furthermore, in applications where the low-band and high-band speech signals S20, S30 are subsequently encoded by different vocoders, the coding efficiency of the low-band vocoder (eg, waveform coder) may decrease due to increasing frequency. For example, the encoding quality of a low-band speech encoder may decrease at low bit rates, especially in the presence of background noise. In such cases, providing an overlap of sub-bands may increase the quality of the reproduced frequency components in the overlapping region.

Furthermore, the overlapping of sub-bands allows for a smooth mixing of low and high frequency bands, which may result in less audible artifacts, reduced aliasing and/or less pronounced transitions from one frequency band to another. This feature may be particularly desirable for implementations in which the low-band vocoder A120 and the high-band vocoder A200 operate according to different encoding methods, as discussed below. For example, different encoding techniques can produce signals that sound completely different. A coder that encodes a spectral envelope in the form of a codebook index can produce a signal that has a different sound than a coder that encodes a magnitude spectrum instead. A time-domain encoder (eg, a pulse code modulation or PCM encoder) can produce a signal that has a different sound than a frequency-domain encoder. An encoder that encodes a signal with a spectral envelope representation and a corresponding residual signal may produce a signal that has a different sound than an encoder that encodes a signal with only a spectral envelope representation. An encoder that encodes a signal as a representation of its waveform can produce an output that has a different sound than that from a sinusoidal encoder. In such cases, using filters with steep transition regions to delimit non-overlapping subbands can produce abrupt, perceptually significant transitions between subbands in the synthesized wideband signal.

While QMF filterbanks with complementary overlapping frequency responses are often used in subband techniques, such filters are not suitable for at least some bandwidth coding implementations described herein. The QMF filterbanks at the encoder are configured to create a significant degree of aliasing, which is canceled out in the corresponding QMF filterbanks at the decoder. Such an arrangement may not be suitable for applications where the signal induces substantial distortion between filter banks, since the distortion may reduce the effectiveness of the alias cancellation properties. For example, applications described herein include encoding implementations configured to operate at very low bit rates. Due to the very low bit rate, the decoded signal is likely to appear significantly distorted compared to the original signal, so that the use of a QMF filter bank may result in non-canceled aliasing. Applications using QMF filter banks typically have higher bit rates (eg, over 12kbps for AMR and over 64kbps for G.722).

Additionally, an encoder may be configured to produce a composite signal that is perceptually similar to the original signal, but is actually significantly different from the original signal. For example, an encoder that derives a high-band excitation from the narrow-band residual described herein may produce such a signal, since the actual high-band residual may be completely absent in the decoded signal. The use of QMF filter banks in such applications can result in a significant degree of distortion caused by non-canceled aliasing.

If the affected subbands are narrower, the amount of distortion caused by QMF aliasing can be reduced because the effect of aliasing is limited to a bandwidth equal to the width of the subband. For example, as described herein, each sub-band includes approximately half of the wideband bandwidth, however, distortion caused by non-anti-aliasing can affect a large portion of the signal. The quality of the signal may also be affected by the position of the frequency band above which the non-anti-aliasing occurs. For example, distortion occurring near the center of a wideband speech signal (eg, between 3kHz and 4kHz) may be more detrimental than distortion occurring near the edge of the signal (eg, around 6kHz).

While the responses of the filters of the QMF filterbank are strictly correlated with each other, the low-band and high-band paths of filterbanks A110 and B120 can be configured to have completely uncorrelated spectra except for the overlap of the two subbands. We define the overlap of two sub-bands as the distance from the point where the frequency response of the high-band filter drops to -20dB to the point where the frequency response of the low-band filter drops to -20dB. In various examples of filter banks A110 and/or B120, this overlap is in the range of around 200 Hz to around 1 kHz. A range of about 400 to about 600 Hz may represent an ideal compromise between coding efficiency and perceived smoothness. In a specific example as described above, the overlap is around 500 Hz.

It may be necessary to implement filter banks A112 and/or B122 to perform operations as illustrated in Figures 6a and 6b in several stages. For example, FIG. 6c shows a block diagram of an implementation A114 of filterbank A112 that uses a series of interpolation, resampling, decimation, and other operations to perform functionally equivalent operations of high-pass filtering and downsampling operations. Such implementations may be easier to design and/or may allow reuse of logical and/or coded functional blocks. For example, as shown in Figure 6c, the same functional block can be used to perform the operations of decimation to 14kHz and decimation to 7kHz. The spectral inversion operation can be implemented by multiplying the signal with a function ^ejnπ or a sequence (-1) ⁿ whose value alternates between +1 and -1. The spectral shaping operation may be implemented as a low pass filter configured to shape the signal to obtain the desired overall filter response.

Note that due to the spectral inversion operation, the spectrum of the high-band signal S30 is inverted. Subsequent operations in the encoder and corresponding decoder can be configured accordingly. For example, it may be desirable to generate a corresponding excitation signal that also has a spectrally inverse form.

Figure 6d shows a block diagram of an implementation B124 of filter bank B122 that uses a series of interpolation, resampling, decimation, and other operations to perform the functional equivalent of upsampling and high-pass filtering operations. Filterbank B 124 includes a spectral inversion operation in the high frequency band that inverts a similar operation performed, eg, in a filterbank of the encoder (eg, filterbank A 114 ). In this particular example, filter bank B 124 also includes notch filters in the low and high bands that attenuate components of the signal at 7100 Hz, but such filters are optional and need not be included. Co-filed patent application "SYSTEMS, METHODS, AND APPARATUS FORSPEECH SIGNAL FILTERING" (US Patent Application Publication No. 2007/0088558) includes additional description regarding the response of elements of specific embodiments of filter banks A110 and B120 and drawings, and this material is hereby incorporated by reference.

As described above, high-band burst suppression can improve the efficiency of encoding the high-band speech signal S30. 7 shows a block diagram of an arrangement for encoding processed highband speech signal S30a (as produced by highband burst suppressor C200) by highband speech encoder A200 to produce encoded highband speech signal S30b.

One wideband speech coding method involves scaling narrowband speech coding techniques (eg, techniques configured to encode the range of 0-4 kHz) to cover the wideband frequency spectrum. For example, a speech signal can be sampled at a higher rate to include components at high frequencies, and narrowband encoding techniques can be reconfigured to use more filter coefficients to represent this wideband signal. 8 shows a block diagram of an example of wideband speech encoder A100 arranged to encode processed wideband speech signal S10a to produce encoded wideband speech signal S10b.

However, narrowband encoding techniques such as CELP (Codebook Excited Linear Prediction) are computationally intensive, and wideband CELP encoders may consume too many processing cycles to be applicable to many mobile and other embedded applications. Encoding the entire spectrum of a wideband signal to the desired quality using this technique may also result in an unacceptably large increase in bandwidth. Furthermore, transcoding of such encoded signals will be required even before the narrowband portion of such encoded signals can be transmitted to and/or decoded by systems that only support narrowband encoding. FIG. 9 shows a block diagram of wideband vocoder A102 including separate low-band and high-band vocoders A120 and A220, respectively.

It may be desirable to implement wideband speech coding such that at least the narrowband portion of the encoded signal can be sent via a narrowband channel (eg, a PSTN channel) without transcoding or other significant modifications. The availability of wideband coding extensions may also be required, for example, to avoid a significant reduction in the number of serviceable users in applications such as wireless cellular telephony and broadcasting on wired and wireless channels.

One wideband speech coding method involves extrapolating a high-band spectral envelope from an encoded narrowband spectral envelope. Although this approach can be implemented without any increase in bandwidth and without the need for transcoding, however, the coarse spectral envelope or formants of the high-band portion of the speech signal cannot generally be accurately predicted from the spectral envelope of the narrowband portion structure.

10 shows a block diagram of a wideband speech encoder A104 that uses another method to encode a high-band speech signal based on information from a low-band speech signal. In this example, the high-band excitation signal is derived from the encoded low-band excitation signal S50. Encoder A104 may be configured to (for example) according to one or more such implementations as described in Patent Application Publication No. WO2006/107837, entitled "METHODS AND APPARATUS FOR ENCODING AND DECODING AN HIGHBANDPORTION OF A SPEECH SIGNAL" example encodes the gain envelope based on a signal based on a high-band excitation signal, the description of which is hereby incorporated by reference. A specific example of the wideband speech encoder A104 is configured to encode the wideband speech signal S10 at a rate of about 8.55 kbps (kilobits per second), with about 7.55 kbps being used for the low-band filter parameters S40 and the encoded low-band excitation signal S50, and about 1 kbps for the encoded high-band speech signal S60.

It may be desirable to combine the encoded low-band and high-band signals into a single bitstream. For example, it may be desirable to multiplex the encoded signals together for transmission (eg, via a wired, optical or wireless transmission channel) or storage as an encoded wideband speech signal. 11 shows a block diagram of an arrangement comprising a wideband speech encoder A104 and a multiplexer A130 configured to combine the low-band filter parameters S40, the encoded low-band excitation signal S50 and the encoded The high frequency band speech signal S30b is combined into a multiplexed signal S70.

The multiplexer A130 may need to be configured to embed the encoded low-band signal (comprising the low-band filter parameters S40 and the encoded low-band excitation signal S50) as separable substreams of the multiplexed signal S70, This allows the encoded low-band signal to be recovered and decoded independently of another part of the multiplexed signal S70, such as the high-band and/or very low-band signal. For example, multiplexed signal S70 may be arranged such that an encoded low-band signal may be recovered by stripping said encoded high-band speech signal S30b. One potential advantage of this feature is to avoid the need to transcode the encoded wideband signal before passing it to a system that supports decoding of the lowband signal but not the highband portion.

An apparatus including a low-band, high-band, and/or wideband speech encoder as described herein may also include circuitry configured to transmit the encoded signal into a transmission channel (eg, a wired, optical, or wireless channel). Such devices may also be configured to perform one or more channel coding operations on the signal, such as error-correcting coding (e.g., rate-compatible convolutional coding) and/or error-detecting coding (e.g., cyclic redundancy coding), and/or One or more layers of network protocol encoding (eg, Ethernet, TCP/IP, cdma2000).

Any or all of the low-band, high-band, and wideband speech encoders described herein may be implemented in terms of a source filter model that encodes an input speech signal as (A) describing a filter A set of parameters and (B) cause the described filter to generate an excitation signal for a synthetic regenerator of the input speech signal. For example, the spectral envelope of a speech signal is characterized by a number of peaks representing resonances of the vocal field and called formants. Most speech coders encode at least this coarse spectral structure as a set of parameters such as filter coefficients.

In one example of a basic source filter arrangement, the analysis module computes a set of parameters characterizing the filter for a period of time (typically 20 milliseconds) of speech sound. A whitening filter (also known as an analysis or prediction error filter) configured according to those filter parameters removes the spectral envelope to flatten the signal spectrally. The resulting whitened signal (also called residual) has less energy and thus less variance than the original speech signal, and is easier to encode. Errors resulting from the encoding of the residual signal may also be more evenly spread over the frequency spectrum. Filter parameters and residuals are typically quantized for efficient transmission over a channel. At the decoder, a synthesis filter configured according to the filter parameters is excited by the residual to produce a synthesized version of the original speech sound. Synthesis filters are typically configured to have a transfer function that is the inverse of that of the whitening filter.

The analysis module may be implemented as a linear predictive coding (LPC) analysis module that encodes the spectral envelope of the speech signal into a set of linear predictive (LP) coefficients (eg, coefficients of the all-pole filter 1/A(z)). Analysis modules typically process the input signal as a series of non-overlapping frames, computing a new set of coefficients for each frame. The frame period is generally the period during which the signal is expected to be locally stationary; a common example is 20 milliseconds (equal to 160 samples at a sampling rate of 8kHz). One instance of the low-band LPC analysis module was configured to calculate a set of ten LP filter coefficients to characterize the formant structure of each 20 millisecond frame of the low-band speech signal S20, and one instance of the high-band LPC analysis module was configured by It is configured to compute a set of six (or, eight) LP filter coefficients to characterize the formant structure of each 20 millisecond frame of the high-band speech signal S30. It is also possible to implement an analysis module to process the input signal as a series of overlapping frames.

The analysis module may be configured to analyze the samples of each frame directly, or may first weight the samples according to a window function (eg, a Hamming window). Analysis may also be performed on windows larger than the frame (eg, 30 millisecond windows). This window may be symmetrical (eg 5-20-5 such that it includes 5 milliseconds immediately before and after a 20 millisecond frame) or asymmetrical (10-20 such that it includes the last 10 milliseconds of the previous frame). The LPC analysis module is typically configured to calculate the LP filter coefficients using the Levinson-Durbin recursion or the Leroux-Gueguen algorithm. In another implementation, the analysis module may be configured to compute a set of cepstral coefficients for each frame instead of a set of LP filter coefficients.

By quantizing the filter parameters, the output rate of the speech coder can be significantly reduced with relatively little effect on the reproduction quality. Linear prediction filter coefficients are difficult to quantize efficiently and are often mapped by a speech coder into another representation, such as line spectral pairs (LSP) or line spectral frequencies (LSF) for quantization and/or entropy coding. Other one-to-one representations of LP filter coefficients include partial autocorrelation coefficients, log-area ratios, immittance-spectrum-pair (ISP), and immittance-spectrum-frequency (ISF), which are used in GSM (Global System for Mobile Communications) AMR- WB (adaptive multi-rate wideband) codec. Typically, the transformation between a set of LP filter coefficients and a corresponding set of LSFs is reversible, but embodiments also include implementations of speech encoders where the transformation is not invertible without error.

Speech encoders are typically configured to quantize the set of narrowband LSFs (or other coefficient representations) and output the results of this quantization as filter parameters. Quantization is typically performed using a vector quantizer that encodes an input vector as an index to a corresponding vector entry in a table or codebook. Such quantizers may also be configured to perform categorical vector quantization. For example, such a quantizer may be configured to select one of a set of codebooks based on information that has been encoded within the same frame (eg, in a low-band channel and/or in a high-band channel). This technique typically provides increased coding efficiency at the cost of additional codebook storage.

The speech encoder may also be configured to generate a residual signal by passing the speech signal through a whitening filter (also called an analysis or prediction error filter) configured according to the set of filter coefficients. Whitening filters are typically implemented as FIR filters, but HR implementations may also be used. This residual signal will usually contain perceptually important information of the speech frame, eg about the long-term structure of pitch, which is not represented in the filter parameters. Furthermore, this residual signal is usually quantized for output. For example, low-band speech encoder A122 may be configured to compute a quantized representation of the residual signal for output as encoded low-band excitation signal S50. This quantization is typically performed using a vector quantizer that encodes input vectors as indices to corresponding vector entries in a table or codebook, and may be configured to perform categorical vector quantization as described above.

Alternatively, such quantizers may be configured to send one or more parameters from which vectors may be dynamically generated at the decoder rather than being retrieved from storage, as in a sparse codebook approach. This approach is used in coding schemes such as Algebraic CELP (Codebook Excited Linear Prediction) and codecs such as 3GPP2 (Third Generation Partnership 2) EVRC (Enhanced Variable Rate Codec).

Some implementations of the low-band speech encoder A120 are configured to compute the encoded low-band excitation signal S50 by identifying one of a set of codebook vectors that best matches the residual signal. Note, however, that the low-band speech encoder A120 may also be implemented to compute a quantized representation of the residual signal without actually generating the residual signal. For example, low-band speech encoder A 120 may be configured to use a number of codebook vectors to generate a corresponding synthesized signal (e.g., according to a current set of filter parameters), and to select an in-perceptual weighting associated with the generated signal The codebook vector that best matches the original low-band speech signal S20 in the domain.

It may be desirable to implement the low-band vocoder A120 or A122 as an analysis-to-synthesis vocoder. Codebook Excited Linear Predictive (CELP) coding is a general family of analytic-synthetic coding, and implementations of such coders can perform waveform coding of residuals, including, for example, selection of entries from fixed and adaptive codebooks, error minimization Manipulation of manipulation and/or sensory weighting manipulation. Other implementations of analysis-by-synthesis coding include Mixed Excitation Linear Prediction (MELP), Algebraic CELP (ACELP), Relaxed CELP (RCELP), Regular Pulse Excitation (RPE), Multi-Pulse CELP (MPE), and Vector Sum Excited Linear Prediction (VSELP) coding. Related coding methods include Multiband Excitation (MBE) and Prototype Waveform Interpolation (PWI) coding. Examples of standard analytically synthesized speech codecs include: ETSI (European Telecommunications Standards Institute) - GSM Full Rate Codec (GSM 06.10), which uses Residual Excited Linear Prediction (RELP); GSM Enhanced Full Rate Codec ( ETSI-GSM 06.60); ITU (International Telecommunication Union) standard 11.8kb/s G.729 Annex E encoder; IS (Interim Standard)-136 (Time Division Multiple Access Scheme) IS-641 codec; GSM adaptive Multi-Rate (GSM-AMR) codec; and 4GV ^(TM) (Fourth-Generation Vocoder ^(TM )) codec (QUALCOMM Incorporated, San Diego, CA). Existing implementations of RCELP coders include the Enhanced Variable Rate Codec (EVRC) as described in the Telecommunications Industry Association (TIA) IS-127, and the 3rd Generation Partnership Project 2 (3GPP2) Alternative Mode Audio Encoder (SMV). The various low-band, high-band, and wideband encoders described herein may be implemented according to any of these techniques, or any other speech coding technique (whether known or yet to be developed), wherein any other Speech coding techniques represent a speech signal as (A) a set of parameters describing a filter and (B) a residual signal providing at least a portion of the excitation for causing the described filter to reproduce the speech signal.

12 shows a block diagram of an implementation C202 of high-band burst suppressor C200 that includes two implementations C10-1, C10-2 of burst detector C10. Burst detector C10-1 is configured to generate a low-band burst indication signal SB10 indicating the presence of a burst in the low-band speech signal S20. Burst detector C10-2 is configured to generate a high-band burst indication signal SB20 indicating the presence of a burst in high-band speech signal S30. Burst detectors C10-1 and C10-2 may be the same or may be instances of different implementations of burst detector C10. The high-band burst suppressor C202 further includes: an attenuation control signal generator C20 configured to generate an attenuation control signal SB70 according to a relationship between the low-band burst indication signal SB10 and the high-band burst indication signal SB20; and A gain control element C150 (eg, a multiplier or an amplifier) configured to apply an attenuation control signal SB70 to the high-band speech signal S30 to generate a processed high-band speech signal S30a.

In the particular example described herein, it may be assumed that the high-band burst suppressor C202 processes the high-band speech signal S30 in 20 millisecond frames, and that both the low-band speech signal S20 and the high-band speech signal S30 are at 8 kHz down is sampled. However, these specific values are examples only, and are not limiting, and other values may also be selected according to a particular design and/or used as described herein.

Burst detector C10 is configured to calculate the forward and backward smooth envelopes of the speech signal, and to indicate the burstiness according to the time relationship between the edges in the forward smooth envelope and the edges in the backward smooth envelope exist. Burst suppressor C202 comprises two instances of burst detector C10, each arranged to receive a respective one of speech signals S20, S30 and output a respective burst indicating signal SB10, SB20.

13 shows a block diagram of an implementation C12 of burst detector C10 that is arranged to receive one of speech signals S20, S30 and output a corresponding burst-indicating signal SB10, SB20. Burst detector C12 is configured to compute each of the forward and backward smoothing envelopes in two stages. In a first stage, calculator C30 is configured to convert the speech signal into a constant polarity signal. In one example, calculator C30 is configured to calculate the constant polarity signal as the square of each sample of the current frame of the corresponding speech signal. This signal can be smoothed to obtain an energy envelope. In another example, calculator C30 is configured to calculate the absolute value of each incoming sample. This signal can be smoothed to obtain an amplitude envelope. Other implementations of calculator C30 may be configured to calculate the constant polarity signal according to another function, such as clipping.

In the second stage, forward smoother C40-1 is configured to smooth the constant polarity signal in the forward time direction to produce a forward smoothed envelope, and backward smoother C40-2 is configured to smooth the constant polarity signal The sexual signal is smoothed in the backward time direction to produce a backward smoothing envelope. The forward smoothing envelope indicates the level difference over time of the corresponding speech signal in the forward direction, and the backward smoothing envelope indicates the level difference over time of the corresponding speech signal in the backward direction.

In one example, forward smoother C40-1 is implemented as a first-order infinite impulse response (IIR) filter configured to smooth constant polarity signals according to, for example, the following expression:

_Sf (n)= _αSf (n-1)+(1-α)P(n),

And backward smoother C40-2 is implemented as a first-order IIR filter configured to smooth a constant polarity signal according to, for example, the following expression:

_Sb (n)= _αSb (n+1)+(1-α)P(n),

where n is the time index, P(n) is the constant polarity signal, S _f (n) is the forward smoothing envelope, S _b (n) is the backward smoothing envelope, and α is Decay factor for values between 1. It may be noted that a delay of at least one frame may be induced in the processed high-band speech signal S30a partly due to operations such as the calculation of the backward smoothing envelope. However, this delay is perceived as relatively insignificant and not uncommon even in real-time speech processing operations.

It may be desirable to choose a value of α such that the decay time of the smoother is similar to the expected duration of the high-band burst (eg, about 5 milliseconds). Typically, forward smoother C40-1 and backward smoother C40-2 are configured to perform complementary versions of the same smoothing operation, and use the same alpha value, but in some implementations, both smoothers may be configured to Do something different and/or use a different value. Other recursive or non-recursive smoothing functions may also be used, including higher order finite impulse response (FIR) or HR filters.

In other implementations of burst detector C12, one or both of forward smoother C40-1 and backward smoother C40-2 are configured to perform an adaptive smoothing operation. For example, forward smoother C40-1 may be configured to perform an adaptive smoothing operation according to an expression such as:

where smoothing is reduced, or in this case disabled, at strong leading edges of constant polarity signals. In this or another implementation of burst detector C12, backward smoother C40-2 may be configured to perform an adaptive smoothing operation according to, for example, the following expression:

where smoothing is reduced, or in this case disabled, at strong trailing edges of constant polarity signals. Such adaptive smoothing may help define the start of bursts in the forward smoothing envelope and the end of bursts in the backward smoothing envelope.

Burst detector C12 includes an instance of region indicator C50 (initial region indicator C50-1) configured to indicate the start of a high level event (eg, a burst) in the forward smoothing envelope. Burst detector C12 also includes an instance of region indicator C50 (termination region indicator C50-2) configured to indicate the end of a high level event (eg, a burst) in the backward smoothing envelope.

Figure 14a shows a block diagram of an implementation C52-1 of an initial region indicator C50-1 that includes a delay element C70-1 and an adder. Delay element C70-1 is configured to apply a delay of positive magnitude such that the forward smoothing envelope reduces its own delayed version. In another example, current samples or delayed samples may be weighted according to a desired weighting factor.

Figure 14b shows a block diagram of an implementation C52-2 of a termination region indicator C50-2 that includes a delay element C70-2 and an adder. Delay element C70-2 is configured to apply a delay with a negative magnitude such that the backward smoothing envelope reduces an advanced version of itself. In another example, the current sample or the previous sample may be weighted according to a desired weighting factor.

Various delay values may be used in different implementations of zone indicator C52, and delay values having different magnitudes may be used in initial zone indicator C52-1 and terminating zone indicator C52-2. The magnitude of the delay can be chosen according to the desired width of the detected area. For example, small delay values can be used to perform detection of narrow edge regions. In order to obtain strong edge detection, it may be necessary to use a delay of a magnitude similar to the expected edge width (eg, about 3 or 5 samples).

Alternatively, area indicator C50 may be configured to indicate a wider area extending beyond the corresponding edge. For example, the initial area indicator C50-1 may need to indicate the initial area of events extending in the forward direction for a period of time after the leading edge. Likewise, the termination region indicator C50-2 may need to indicate the termination region of events extending in the backward direction for a period of time preceding the trailing edge. In this case, it may be desirable to use a delay value with a larger magnitude, for example a magnitude similar to that of the expected length of the burst. In one such example, a delay of about 4 milliseconds is used.

The processing by region indicator C50 may extend beyond the boundaries of the current frame of the speech signal depending on the magnitude and direction of the delay. For example, the processing by the initial region indicator C50-1 may extend to a previous frame, and the processing by the termination region indicator C50-2 may extend to a subsequent frame.

Compared with other high-level events that may occur in the speech signal, the burst consists of an initial region (as indicated in the initial region indication signal SB50) that coincides in time with the termination region (as indicated in the termination region indication signal SB60). Instructions) to identify. For example, a burst may be indicated when the temporal distance between the initial region and the terminating region is not greater (or less) than a predetermined coincidence interval (eg, the expected duration of the burst). The coincidence detector C60 is configured to indicate the detection of a burst based on the temporal coincidence of the initial zone and the ending zone in the zone indication signals SB50 and SB60. For example, for implementations in which initial region indication signal SB50 and end region indication signal SB60 indicate regions extending from respective leading and trailing edges, coincidence detector C60 may be configured to indicate temporal overlap of the extended regions.

15 shows a block diagram of an implementation C62 of a coincidence detector C60 that includes: a first instance C80-1 of a clipper C80 configured to clip an initial zone indication signal SB50; a clipper C80 A second instance C80-2 of C80-2 configured to clip the end zone indication signal SB60; and a mean value calculator C90 configured to output a corresponding burst indication signal according to the mean value of the clipped signal. Clipper C80 is configured to clip the value of the input signal according to, for example, the following expression:

output = max(input, 0).

Alternatively, clipper C80 may be configured to threshold the value of the input signal according to, for example, the following expression:

Here, the threshold _TL has a value greater than zero. Typically, instances C80-1 and C80-2 of clipper C80 will use the same threshold, but it is also possible that two instances C80-1 and C80-2 use different thresholds.

且经布置以处理高频带结果的均值计算器C90的实例使用较保守的几何均值

The mean value calculator C90 is configured to output respective burst indicating signals SB10, SB20 according to the mean value of the clipped signal, said burst indicating signal indicating the temporal position and intensity of a burst in the input signal and having a value equal to or greater than zero value. Especially for distinguishing bursts with well-defined initial and termination regions from other events with only strong initiation or termination regions, the geometric mean may provide better results than the arithmetic mean. For example, the arithmetic mean of events with only one strong edge may still be high, and the geometric mean of events without one of the edges will be low or zero. However, the geometric mean is usually more computationally intensive than the arithmetic mean. In one example, an instance of mean calculator C90 arranged to process low-band results uses the arithmetic mean

and the instance of Mean Calculator C90 arranged to handle high-band results uses a more conservative geometric mean

Other implementations of mean calculator C90 may be configured to use different kinds of means, such as harmonic means. In another embodiment of coincidence detector C62, one or both of initial region indicating signal SB50 and ending region indicating signal SB60 are weighted relative to the other before or after clipping.

Other implementations of coincidence detector C60 are configured to detect bursts by measuring the temporal distance between leading and trailing edges. For example, one such implementation is configured to identify a burst as a region where the separation between a leading edge in initial region indicating signal SB50 and a trailing edge in ending region indicating signal SB60 is not greater than a predetermined width. The predetermined width is based on the expected duration of the high-band burst, and in one example, a width of about 4 milliseconds is used.

Another implementation of coincidence detector C60 is configured to extend each leading edge in initial region indication signal SB50 in the forward direction by a desired period of time (e.g., based on the expected duration of the high-band burst), and to Each trailing edge in termination region indication signal SB60 extends in the backward direction for a desired period of time (eg, based on the expected duration of the high-band burst). Such implementations may be configured to generate respective burst-indicating signals SB10, SB20 as the logical AND (AND) of these two spread signals, or to generate corresponding burst-indicating signals SB10, SB20 to indicate regions that span regions overlapping The relative strength of the burst (eg, by computing the mean of the spread signal). Such implementations may be configured to only expand edges that exceed a threshold. In one example, the edges are extended for a time period of about 4 milliseconds.

The attenuation control signal generator C20 is configured to generate the attenuation control signal SB70 according to the relationship between the low-band burst indication signal SB10 and the high-band burst indication signal SB20 . For example, the attenuation control signal generator C20 may be configured to generate the attenuation control signal SB70 according to an arithmetic relationship (eg, a difference) between the burst indication signals SB10 and SB20 .

FIG. 16 shows a block diagram of an implementation C22 of attenuation control signal generator C20 configured to divide the low-band burst indication by subtracting the low-band burst indication signal SB10 from the high-band burst indication signal SB20. Signal SB10 is combined with high-band burst indication signal SB20. The resulting difference signal indicates where in the high frequency band bursts exist that do not occur (or are weaker) in the low frequency band. In another embodiment, one or both of the low-band burst-indicating signal SB10 and the high-band burst-indicating signal SB20 is weighted relative to the other.

The attenuation control signal calculator C100 outputs the attenuation control signal SB70 according to the value of the difference signal. For example, attenuation control signal calculator C100 may be configured to indicate an attenuation that varies according to how much the difference signal exceeds a threshold.

Attenuation control signal generator C20 may need to be configured to operate on logarithmically scaled values. For example, it may be necessary to attenuate the high-band speech signal S30 according to the ratio between the levels of the burst indication signals (for example, according to a value in decibels or dB), and this ratio can be according to the logarithm The difference between the scaled values is easily calculated. Logarithmic scaling distorts a signal along the magnitude axis without otherwise changing its shape. 17 shows an implementation C14 of burst detector C12 that includes instances C130-1, C130-2 of logarithmic calculators C130 configured to The smoothing envelope in the latter is logarithmically scaled (eg, to base 10).

In one example, attenuation control signal calculator C100 is configured to calculate the value of attenuation control signal SB70 according to the following formula:

where D _dB represents the difference between the high-band burst indicator signal SB20 and the low-band burst indicator signal SB10, T _dB represents the threshold value, and A _dB is the corresponding value of the attenuation control signal SB70. In one particular example, the threshold T _dB has a value of 8dB.

In another implementation, the attenuation control signal calculator C100 is configured to indicate a linear attenuation according to how much the difference signal exceeds a threshold (eg, 3dB or 4dB). In this example, the decay control signal SB70 does not indicate decay until the difference signal exceeds the threshold. When the difference signal exceeds the threshold, the attenuation control signal SB70 indicates an attenuation value that is linearly proportional to the amount by which the threshold is currently exceeded.

The high-band burst suppressor C202 includes a gain control element C150 (e.g., a multiplier or amplifier) configured to attenuate the high-band speech signal S30 to produce a processed high frequency band speech signal S30a. Typically, unless a high-band burst has been detected at the current position of the high-band speech signal S30, the attenuation control signal SB70 indicates a value of no attenuation (for example, a gain of 1.0 or 0 dB), at which point a high-band burst has been detected. A typical attenuation value is 0.3 or about a 10dB gain reduction in the case of band bursting.

Alternative implementations of attenuation control signal generator C22 may be configured to combine low-band burst indication signal SB10 with high-band burst indication signal SB20 according to a logical relationship. In one such example, the burst indication signal is combined by computing the logical AND of the high-band burst indication signal SB20 and the logical inverse of the low-band burst indication signal SB10. In this case, each of the burst indication signals may first be thresholded to obtain a binary valued signal, and the attenuation control signal calculator C100 may be configured to indicate the one of the two attenuation states according to the state of the combined signal A corresponding state (eg, a state indicating no decay).

It may be desirable to spectrally shape one or both of speech signals S20 and S30 to flatten the spectrum and/or emphasize or attenuate one or more specific frequency regions prior to performing envelope calculations. Low-band speech signal S20, for example, may tend to have more energy at low frequencies, and this energy may need to be reduced. It may also be desirable to reduce the high-frequency components of the low-band speech signal S20 such that burst detection is mainly based on intermediate frequencies. Spectral shaping is an optional operation that may improve the performance of burst suppressor C200.

18 shows a block diagram of an implementation C16 of burst detector C14 that includes shaping filter C110. In one example, filter C110 is configured to filter low-band speech signal S20 according to, for example, the following passband transfer function:

${\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; LB</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0.96</span>}{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0.96</span>}{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}}}{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; ,</span>$

It attenuates very low and very high frequencies.

It may be desirable to attenuate the low frequencies of the high-band speech signal S30 and/or boost the higher frequencies. In one example, filter C110 is configured to filter high-band speech signal S30 according to, for example, the following high-pass transfer function:

${\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; HB</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0.3</span>}{\mathrm{<span\; class="notranslate">\; z</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; ,</span>$

It attenuates frequencies around 4kHz.

It may not be necessary in a practical sense to perform at least some of the burst detection operations at the full sampling rate of the respective speech signal S20, S30. 19 shows a block diagram of an implementation C18 of burst detector C16 that includes an instance C120-1 of downsampler C120 configured to downsample the smoothed envelope in the forward processing path. and an instance C120-2 of a downsampler C120 configured to downsample the smoothed envelope in the backward processing path. In one example, each instance of downsampler C 120 is configured to downsample the envelope by a factor of eight. For the specific example of a 20 millisecond frame (160 samples) sampled at 8 kHz, such a downsampler reduces the envelope to a 1 kHz sampling rate, or 20 samples per frame. Downsampling significantly reduces the computational complexity of high-band burst suppression operations without significantly affecting performance.

The attenuation control signal applied by the gain control element C150 may need to have the same sampling rate as the high-band speech signal S30. 20 shows a block diagram of an implementation C24 of attenuation control signal generator C22 that may be used in conjunction with a downsampled version of burst detector C10. The attenuation control signal generator C24 includes an upsampler C140 configured to upsample the attenuation control signal SB70 to obtain a signal SB70a having a sampling rate equal to that of the highband speech signal S30.

In one example, upsampler C140 is configured to perform upsampling by zero-order interpolation of attenuation control signal SB70. In another example, upsampler C140 is configured to additionally obtain less steep transitions by interpolating between values of attenuation control signal SB70 (e.g., by passing attenuation control signal SB70 through a FIR filter). Perform upsampling. In another example, upsampler C140 is configured to perform upsampling using a windowed sine function.

In some cases, such as in a battery powered device such as a cellular telephone, the high band burst suppressor C200 may be configured to be selectively disabled. For example, it may be desirable to disable operations such as high-band burst suppression in the device's power saving mode.

As mentioned above, the embodiments described herein include implementations that can be used to perform embedded encoding, support compatibility with narrowband systems, and avoid the need for transcoding. Support for high-band encoding may also be used to differentiate chips, chipsets, devices, and/or networks with wideband support and backward compatibility from those with narrowband support only, based on cost. The support for high-band coding described herein can also be used in conjunction with techniques for supporting low-band coding, and a system, method, or device according to this embodiment can support, for example, about 50 or 100 Hz up to about 7 or 8 kHz The encoding of the frequency components of .

As mentioned above, adding high-band support to the speech coder can improve intelligibility, especially with respect to the distinction of fricatives. While such distinctions can often be derived by the listener from a particular context, high-band support can be used as a trigger in speech recognition and other machine interpretation applications such as systems for automated voice menu navigation and/or automated call handling feature. High-band burst suppression can increase accuracy in machine interpretation applications, and it is contemplated that an implementation of high-band burst suppressor C200 can be used in one or more such applications with or without speech coding.

An apparatus according to an embodiment may be embedded in a portable device for wireless communication, such as a cellular telephone or a personal digital assistant (PDA). Alternatively, such equipment may be included in another communication device, such as a VoIP handset, a personal computer configured to support VoIP communications, or a network device configured to route telephone or VoIP communications. For example, an apparatus according to an embodiment may be implemented on a chip or chip set for a communication device. Depending on the particular application, such devices may also include features such as analog-to-digital conversion and/or digital-to-analog conversion of speech signals, circuitry for performing amplification and/or other signal processing operations on speech signals, and/or Or a radio frequency circuit for transmitting and/or receiving encoded speech signals.

It is expressly contemplated and disclosed that embodiments may comprise and/or be combined with any one or more of the other features disclosed in the published patent applications referenced herein. Used together, this application claims the benefit of said US Provisional Patent Application. Such features include the generation of high-band excitation signals from low-band excitation signals, which may include other features such as: inverse sparse filtering, harmonic extension using nonlinear functions, mixing of modulated noise signals with spectrally extended signals , and/or adaptive whitening. Such features include time-warping the high-band speech signal according to the regularization performed in the low-band encoder. Such features include encoding gain envelopes according to the relationship between the original speech signal and the synthesized speech signal. Such features include the use of overlapping filter banks to obtain low-band and high-band speech signals from wideband speech signals. Such features include shifting the high-band signal S30 and/or a high-band excitation signal according to a regularization or other shifting of the low-band excitation signal S50. Such features include fixed or adaptive smoothing of coefficient representations such as high-band LSFs. Such features include fixed or adaptive shaping of noise associated with quantization of coefficient representations such as LSFs. Such features also include fixed or adaptive smoothing of the gain envelope, and adaptive decay of the gain envelope.

The above description of the described embodiments is provided to enable any person skilled in the art to make or use the invention. Various modifications to these embodiments are possible, and the generic principles presented herein may be applied to other embodiments as well. For example, an embodiment may be implemented in part or in whole as a hardwired circuit, as a circuit configuration fabricated as an application specific integrated circuit, or as a firmware program loaded into non-volatile storage or as machine-readable code from The data storage medium loads or loads a software program into the data storage medium, the code being instructions executable by an array of logic elements such as a microprocessor or other digital signal processing unit. The data storage medium may be: an array of storage elements such as semiconductor memory (which may include, but is not limited to, dynamic or static RAM (random access memory), ROM (read only memory), and/or flash RAM), or iron electrical, magnetoresistive, bidirectional, polymeric or phase change memory; or disk media such as magnetic or optical disks. The term "software" shall be understood to include source code, assembly language code, machine code, binary code, firmware, macrocode, microcode, any one or more sets or sequences of instructions executable by an array of logic elements, and such Any combination of instances.

Various elements of an implementation of highband vocoder A200, wideband vocoders A100, A102 and A104, and highband burst suppressor C200 and arrangements comprising one or more such devices may be implemented to reside in For example, electronic and/or optical devices on the same chip or between two or more chips in a chip set, but other arrangements not subject to such limitations are also contemplated. One or more elements of such an apparatus may be implemented, in whole or in part, as one or more sets of instructions arranged to execute one or more arrays of fixed or programmable logic elements (e.g., transistors, gates), such as Microprocessors, embedded processors, IP cores, digital signal processors, FPGAs (Field Programmable Gate Arrays), ASSPs (Application Specific Standard Products) and ASICs (Application Specific Integrated Circuits). One or more such elements may also have a common structure (e.g., a processor for executing portions of code corresponding to different elements at different times, a set of instructions for executing tasks corresponding to different elements at different times, or An arrangement of electronic and/or optical devices that performs operations for different elements at different times). Furthermore, it is possible that one or more such elements are used to perform tasks or other sets of instructions not directly related to the operation of the apparatus, such as tasks related to the operation of another apparatus or system in which the apparatus is embedded.

Embodiments also include additional speech processing, speech coding, and high-band burst suppression methods as explicitly disclosed herein, for example, by describing structural embodiments configured to perform such methods. Each of these methods may also be tangibly implemented (e.g., in one or more of the data storage media enumerated above) as one or more devices comprising an array of logic elements (e.g., a processor, microprocessor, microprocessor, A set of instructions that a machine reads and/or executes, such as a controller or other finite state machine. Thus, the present invention is not intended to be limited to the embodiments shown above but is to be accorded the widest scope consistent with the principles and novel features disclosed in any way herein.

Claims (29)

1. signal processing method, said method comprises:

Calculate the first burst indicator signal, whether the said first burst indicator signal indication detects burst in the low frequency part of audio speech signal;

Calculate the second burst indicator signal, whether the said second burst indicator signal indication detects burst in the HFS of said audio speech signal;

Arithmetic relation or logical relation according between said first burst indicator signal and the said second burst indicator signal produce attenuation control signal; And

The said HFS that said attenuation control signal is imposed on said audio speech signal is to produce treated high-frequency signal part.

2. signal processing method according to claim 1, at least one in wherein said calculating first burst indicator signal and the said calculating second burst indicator signal comprises:

Produce the level and smooth envelope of forward direction of said voice signal appropriate section, the level and smooth envelope of said forward direction is level and smooth on positive time direction;

The prime area of the burst of indication in the level and smooth envelope of said forward direction;

Produce the back to level and smooth envelope of said voice signal appropriate section, said back is level and smooth on negative time orientation to level and smooth envelope; And

Indication is in the termination zone of the burst of said back in level and smooth envelope.

3. signal processing method according to claim 2, at least one in wherein said calculating first burst indicator signal and the said calculating second burst indicator signal comprise and detect said prime area and stop regional overlapping in time with said.

4. signal processing method according to claim 2, at least one in wherein said calculating first burst indicator signal and the said calculating second burst indicator signal comprise according to said prime area and stop regional in time overlapping and indicate burst with said.

5. method according to claim 2; In wherein said calculating first burst indicator signal and the said calculating second burst indicator signal at least one comprises according to the corresponding burst indicator signal of the mean value computation of two signals, and said two signals are that (A) is based on the signal of the indication of said prime area with (B) based on the said signal that stops the indication in zone.

6. method according to claim 1, the level of burst on logarithmically calibrated scale that at least one indication in wherein said first burst indicator signal and the said second burst indicator signal is detected.

7. method according to claim 1, wherein said generation attenuation control signal comprise that the difference that happens suddenly between the indicator signal according to the said first burst indicator signal and said second produces said attenuation control signal.

8. method according to claim 1, wherein said generation attenuation control signal comprise that the degree that the level according to the said second burst indicator signal surpasses the level of the said first burst indicator signal produces said attenuation control signal.

9. method according to claim 1; The wherein said said HFS that said attenuation control signal is imposed on said audio speech signal comprises following among both at least one: (A) said HFS and said attenuation control signal are multiplied each other and (B) amplify said HFS according to said attenuation control signal.

10. method according to claim 1, said method comprise handles said audio speech signal to obtain said low frequency part and said HFS.

11. method according to claim 1, said method comprise becoming a plurality of at least coefficient of linear prediction wave filter based on said treated high-frequency signal signal encoding partly.

12. method according to claim 11, said method comprise said low frequency part is encoded into more than at least the second coefficient of linear prediction wave filter and through code-excited signal,

Wherein said coding comprises according to the gain envelope of encoding based on said signal through code-excited signal based on the signal of said treated high-frequency signal part based on the signal of said treated high-frequency signal part.

13. method according to claim 11, said method comprise said low frequency part is encoded into more than at least the second coefficient of linear prediction wave filter and through code-excited signal, and

Produce high band excitation signal based on said through code-excited signal,

Wherein said coding comprises according to the gain envelope of encoding based on the signal of said high band excitation signal based on the signal of said treated high-frequency signal part based on the signal of said treated high-frequency signal part.

14. an equipment that comprises the highband burst rejector, said highband burst rejector comprises:

First burst detector, it is through being configured to export the first burst indicator signal, and whether the said first burst indicator signal indication detects burst in the low frequency part of audio speech signal;

Second burst detector, it is through being configured to export the second burst indicator signal, and whether the said second burst indicator signal indication detects burst in the HFS of said audio speech signal;

The attenuation control signal generator, it is through being configured to produce attenuation control signal according to arithmetic relation or logical relation between said first burst indicator signal and the said second burst indicator signal; And

Gain control element, it is through being configured to said attenuation control signal is imposed on the said HFS of said audio speech signal.

15. equipment according to claim 14, at least one in wherein said first burst detector and said second burst detector comprises:

The forward direction smoother, it is through being configured to produce the level and smooth envelope of forward direction of said voice signal appropriate section, and said forward direction smoothly is included on the positive time direction level and smooth;

The first area indicator, it is through being configured to indicate the prime area of the burst in the level and smooth envelope of said forward direction;

The back is to smoother, its through be configured to produce said voice signal appropriate section after to level and smooth envelope, said back is level and smooth on negative time orientation to level and smooth envelope; And

The second area indicator, it is through being configured to indicate the termination zone of burst in level and smooth envelope after said.

16. equipment according to claim 15; At least one burst detector in wherein said first burst detector and second burst detector comprises coincidence detector, and said coincidence detector stops zone overlapping in time through being configured to detect said prime area with said.

17. equipment according to claim 15; At least one burst detector in wherein said first burst detector and second burst detector comprises coincidence detector, and said coincidence detector is indicated burst through being configured to stop in time overlapping of zone according to said prime area and said.

18. equipment according to claim 15; At least one burst detector in wherein said first burst detector and second burst detector comprises coincidence detector; Said coincidence detector its through being configured to export corresponding burst indicator signal according to the average of two signals, said two signals be (A) based on the signal of the indication of said prime area with (B) based on the said signal that stops the indication in zone.

19. equipment according to claim 14, the level of burst on logarithmically calibrated scale that at least one indication in wherein said first burst indicator signal and the said second burst indicator signal is detected.

20. equipment according to claim 14, wherein said attenuation control signal generator is through being configured to produce said attenuation control signal according to the difference between said first burst indicator signal and the said second burst indicator signal.

21. equipment according to claim 14, wherein said attenuation control signal generator produces said attenuation control signal through the degree that is configured to level according to the said second burst indicator signal and surpasses the level of the said first burst indicator signal.

22. equipment according to claim 14, wherein said gain control element comprises at least one in multiplier and the amplifier.

23. equipment according to claim 14, said equipment comprises bank of filters, and said bank of filters is through being configured to handle said voice signal to obtain said low frequency part and said HFS.

24. equipment according to claim 14, said equipment comprises the high frequency band speech coder, and said high frequency band speech coder is through being configured to that the signal encoding based on the output of said gain control element is become a plurality of at least coefficient of linear prediction wave filter.

25. equipment according to claim 24, said equipment comprises the low-frequency band speech coder, and said low-frequency band speech coder is through being configured to that said low frequency part is encoded into more than at least the second coefficient of linear prediction wave filter and through code-excited signal,

Wherein said high frequency band speech coder is through being configured to according to the gain envelope of encoding based on said signal through code-excited signal based on the signal of the output of said gain control element.

26. equipment according to claim 25, wherein said high band encoder is through being configured to produce high band excitation signal based on said through code-excited signal, and

Wherein said high frequency band speech coder is through being configured to according to the gain envelope of encoding based on the signal of said high band excitation signal based on the signal of the output of said gain control element.

27. equipment according to claim 14, said equipment comprises cellular phone.

28. a signal handling equipment, it comprises:

Be used to calculate the device of the first burst indicator signal, whether the said first burst indicator signal indication detects burst in the low frequency part of audio speech signal;

Be used to calculate the device of the second burst indicator signal, whether the said second burst indicator signal indication detects burst in the HFS of said audio speech signal;

Be used for according to the arithmetic relation between said first burst indicator signal and the said second burst indicator signal or the device of logical relation generation attenuation control signal; And

The said HFS that is used for said attenuation control signal is imposed on said audio speech signal is to produce the device of treated high-frequency signal part.

29. equipment according to claim 28, the wherein said device that is used for calculating the first burst indicator signal and said are used to calculate at least one of device of the second burst indicator signal and comprise:

Be used to produce the device of the level and smooth envelope of forward direction of said voice signal appropriate section, the level and smooth envelope of said forward direction is level and smooth on positive time direction;

Be used in reference to the device of the prime area that is shown in the burst in the level and smooth envelope of said forward direction;

Be used to produce the back device to level and smooth envelope of said voice signal appropriate section, said back is level and smooth on negative time orientation to level and smooth envelope; And

Be used in reference to the device in the termination zone that is shown in the burst of said back in level and smooth envelope.

Images