NL8103124A - NETWORKS FOR SUPPRESSION OF MID-FREQUENCY MODULATION EFFECTS IN COMPRESSION ORGANIZATION, EXPANSION ORGANIZATION AND NOISE REDUCTION SYSTEMS. - Google Patents

Description

' e

Networks for Suppressing Middle Frequency Emodulation Effects in Compressors, Expansions, and Noise Reduction Systems.

The invention generally relates to circuit devices that change the dynamic region of signals, namely compression members that compress the dynamic region and expansion members that expand the dynamic region. The invention is particularly useful for processing audio signals, but can also be applied to other signals.

Compression members and expansion members are normally used together (a compression expansion system) to effect 10 noise reduction; the signal is compressed for transmission or recording and expanded upon receipt or playback from the transmission channel. Compression means, however, can only be used to reduce the dynamic range, for example, to track the capacity of a transmission channel without additional expansion when the compressed signal is sufficient for the end purpose. Additionally, compression means are used only in certain products, especially audio products, the purpose of which is only to transmit or record compressed broadcasting or pre-recorded signals. Expansions are used only with certain products, especially audio products, the purpose of which is only to receive or play back already compressed broadcasting or pre-recorded signals. With certain products, particularly audio recording and playback products, a single device is often constructed to switch as a compression means to record signals and as an expansion means to play compressed broadcast or pre-recorded signals.

Transmission channels are known to have frequency-dependent characteristics. Accordingly, the frequency spectrum of received or played signals is changed, and when compressed signals are applied to a transmission channel, the complementary expansion of the signals is degraded by the transmission channel frequency characteristics.

In compression expansion type reduction systems, complementarity requires not only that the expansion member has essentially the inverse characteristics of the compression member but also that the transmission channel between the compression member and the expansion member has a relative retains signal amplitudes "at all frequencies" within the palm of compressed signals. As received in the expander, changes in the relative levels caused by the transmission channel are indistinguishable from signal processing by the compression member. Thus, errors in the transit mission channel cause the expanded signals to differ from the input signals to the compression member. Such differences can be important and audible depending on the spectral content of the signals. With large compression expansion ratios, errors in the transmission channel become more important.

Typically, the best audible effect is not the direct effect on signals themselves with very high or very low frequencies, but instead the modulation effect on signals between the extreme bandwidths caused by the extremely high and low frequency signals reaching the expander not reach.

For the discussion, this effect will be referred to as the mid-band modulation effect.

With wideband compression expanders, an amplitude error at the control frequency will be expressed to the same extent in all other parts of the spectrum. This may or may not be acceptable. In slider band expanders, as described below, errors at such frequencies are multiplied significantly at mid frequencies when the control frequencies are at the bandwidth extremes (conversely, when the control frequencies are at the mid frequencies as it usually is). In this case, every 8103 124 I-3 error at the frequency extremes is reduced (this is an advantage of the sliding band compression expanders).

By "transmission channel" as used herein is meant all parts of the system between a compression member 5 and an expansion member.

In the case of noise reduction systems that record compressed signals on relatively narrow bandwidth media such as a slow magnetic tape, eg compact cassettes, it is particularly difficult to provide an accurate complementary expansion of the compressed signals.

This is due to the fact that such systems are unable to provide a flat signal amplitude response, especially at low and high frequencies. However, such problems exist even with professional compression / expansion gear 15 seis. The inadequacy occurs after the compression member and can be caused by the recording device, the tape, the display unit or any combination thereof including bandwidth limiting liters. Likewise, in other systems, such as FM or satellite broadcasting, the errors occur after compression in the transmitter, the medium carrying the broadcast signal, the receiver, or combinations thereof.

Practical compression expansion systems have found it necessary to include sharp cutoff filters for the compression member and for the expansion member.

The filters are at least low-transmissive, but preferably also contain high-transmissive parts. Such bandlimiting filters have angular frequencies at the edges of or outside the conventional bandpass of the system so as not to limit the bandwidth of the system. These filters have several functions: a) attenuation of subcarrier components and the 19 kHz pilot tone used in FM broadcasting, to avoid setting "birdie" beats (whistles) and incorrectly following coding and decoding noise reduction processing chains.

35 b) Attenuation of the tape recorder setting specified in the 8103 124

»V

- h - signal chains can Cover to avoid coding / decoding error tracking.

c) Attenuation of rf or supersonic components in the encoder input signal which may otherwise result in intermodulation products and / or setting "birdies".

d) Attenuation of supersonic band noise or other transmission channel noise at the decoding input, to avoid encoding / decoding error tracking.

e) A signal bandwidth definition to promote complementarity in response from the encoder / decoder.

In professional applications it is desirable to use a high frequency bandwidth limiting filter (for example at 20 - 25 kHz) and preferably also a low frequency bandwidth limiting filter (for example at 20 Hz).

When an ideal channel exists between the decoder and the decoder, strictly speaking the input filter must be decoupled with the decoder since its recording may result in some signal situations with a low level of implementation (the encoding signal is subject to one stage of filtering; the decoder is subject to two stages of filtering). However, removal of the decoding input filter is not practical due to considerations a) to e) above. Even when there is a good channel 25 between the encoder input and the decoder input, the presence of the much needed decoder input filter (for protection purposes) results in an inherent non-complementary system under certain signal conditions. As a result, it may help to expand the expansion input filter. consider as an integral part of the transmission channel.

It is an object of the invention to suppress the adverse effects on complementarity in compression expansion noise reduction systems caused by errors in the transmission channel amplitude response.

In other words, it is an object of the invention to provide for the suppression of the deleterious effects on complementarity in compression expansion type noise reduction systems caused by amplitude response errors that occur between the compression member and the expansion organ.

It is a "particular object of the invention to suppress the non-complementary effects" at very high (low) frequencies that produce audible effects at mid-frequencies (to reduce the mid-band modulation effect).

A further object of the invention is to suppress such adverse effects in systems in which the bandwidth of compressed signals exceeds the bandwidth in which the transmission channel amplitude response is relatively flat.

It is yet another object of the invention to suppress such adverse effects in audio tape systems using compact cassettes or other bandwidth limited media including the audio portion of video cassettes and video records.

It is a further object of the invention to reduce the non-complementary modulation of low level 20 mid-frequency (e.g. 500 Hz - 1 kHz, or so) signals when present with extremely high frequency signals (e.g. above 10 kHz) at such audio tape systems).

It is a further goal to reduce non-complementary effects introduced by the use of expansion pans input filters.

Another object of the invention is to reduce the above effects in sliding belt systems of the kind described below.

In limited bandwidth systems, the bandwidth frequency of compressed signals approaches or exceeds the usual bandwidth of the recording / playback transmission channel, and thus such systems are particularly prone to errors in the recording / playback frequency response, especially in the region of the edges of the high and 35 low frequencies.

8103124

V

- β -

There are two aspects to the issue: 1) the output bandwidth of the compression member can exceed the bandwidth in which the recording / playback response is relatively flat in this complete audio device and 2) the output bandwidth of the compression member. of the device used in the production of a pre-recorded tape or record may exceed the bandwidth in which the playback response of the audio device is flat.

While errors can theoretically occur anywhere in the spectrum as will be explained further below, the invention is directed at suppressing the effect of errors at the bandwidth ends of the system, especially at suppressing the midband modulation effects caused by such errors.

The invention which solves the mid-band modulation problem is quite surprising in its simplicity.

More specifically, the signals processed by the compression member are subject to a sudden high (and / or low) frequency drop with a break point frequency well within the system's usable band-pass slightly below (above) the frequency at which the transmission channel or recording / playback response exhibits significant errors. Preferably, signals processed by the expander are subjected to a complementary hunt so that an overall flat frequency response is maintained. When a lack of overall flat frequency response is acceptable, the invention can be included only in the compression portions of the compression expansion system.

Thus, according to the invention, a part (or parts) of the signal spectrum in which the transmission channel or recording / playback response has damped errors to a level such that the damped part is practically excluded from the control of compression and expansion.

According to the invention, the spectral content of 35th signals processed by the compression member is changed or 8103124-7 - skewed ("spectral skewing") such that the compression distortion is significantly less sensitive to the influence of signals beyond the sudden breakpoint frequency.

A preferred embodiment of the invention is to locate a suitable filter (or network) in the signal path from the input of the expansion device (and preferably a complementary filter in the signal sweep from the output of the expansion device). . This approach is preferred because it requires the least amount of chain components and works at all signal levels. However, a similar device is to use two filters: one in the control circuit wipe of the congresser and the other in the signal path of the output of the ecompression member. Likewise, in the expander one filter is located in the control circuit wipe and the other in the signal path from the input of the expander.

Further alternatives are possible in the case of a compression device or two-way expansion device (for example, as indicated in US-PS 3.8U6.719 and US-PS Re 28. U26): only a suitable filter can be located in the input of the side road of the scaffolding device to affect both the signals passing through it and the control circuit of the side road or the equivalent two-filter device can be used with one filter placed in the side sweep control circuit wipe and the other in the side sweep signal output wipe.

In compression expansion systems it is known to provide a kink at high frequencies in the compression member side of the system and to provide complementary boost in the expansion member side of the system to reduce band saturation for example (US- PS 3.8U6.7199 30 US-PS-U.072.U19, Rundfunktechn. Mitteilungen, Volume 92, (1978) H. 2, pp. 63-7U.) The kink is not sudden and is not sufficient to prevent that high level signals at high (low) frequencies in the uncertain response region of the system affect the eompressor as obtained with the invention.

8103124

♦ Q

- 8 -

From the above publications it is known to place anti-saturation filters in different places in the signal path; before and after the compression member / expansion member and within the compression member / expansion member 5 and at the same time included in its control chain. Thus, there are many possible locations for anti-saturation filters in the compression member and expansion organ signal paths that will serve the purpose of reducing the high frequency signals applied to the recording band. In contrast, the circuit according to the invention must be more precise because it intends to influence the compression member and, consequently, the expansion member by changing the frequency spectrum of the signals that the compression member is treating.

Although the invention is not aimed at alleviating the saturation effects, it will nevertheless help to avoid band saturation to some extent. However, because the breakdown breakdown frequency is simply located at a frequency above (below) the region in which saturation begins to occur, band saturation is preferably treated by other means as described, for example, in Audio, May 1981, pp. 20-26.

As with the known anti-saturation devices, the invention results in a loss of noise reduction effect in the region in which the signals are suddenly damped as a function of the frequency. The sudden drop breakpoint for the upper frequency end of the array is at a relatively high frequency in the range in which the human ear is much less sensitive to noise. With regard to the use of the invention at the low-frequency end of the system, the ear is also less sensitive and again there would be essentially no audible noise at such frequencies with properly designed systems even with reduced noise reduction . The invention contemplates a spectral, bevel only at the high and / or low frequency extremes, because in practical systems these frequency ranges are the only ones in which the transmission channel amplifier has significant errors. A further reason is that noise reduction degradation may be allowed at these extremes due to the response of the human ear.

In another system, a 12 dB / octave bandpass filter with breakpoints of about 20 Hz and 10 kHz is used in the control chain of a wide band consumer audio band compression expansion noise reduction system (sold under the trade name "dbx II"). However, this device does not achieve the results of the invention because there is no filter in the signal path and a high level high (low) frequency signal beyond 10 kHz (or below 20 Hz) is amplified in accordance with the level of what kind of signals also be present within the band pass of the control chain. Thus, such a device provides excessive amplification of high level, high (low) frequency signals (outside the 20 Hz-10 kHz band) when large amplitude signals are not present within the 20 Hz-10 kHz band, resulting in the overdrive of the transmission channel.

It is also known with a video noise reduction system (US-PS 3.8½.719) to provide a notch filter with variable Q responsive to the color level centered on the color television subcarrier frequency to prevent the color signals from compressing and pinch the expansion member and thus eliminate noise reduction at frequencies below the color carrier frequency. Thus, the filter center is within the bandwidth extremes of the signals in the system and within the predictable response of the system transmission channel and, moreover, it does not serve the purpose of overcoming recording / playback response errors underlying the invention.

It is further known to provide 12 dB / octave low pass, band pass and high pass filters in the sideways of compression members and two-way expansion members (US-PS 35 3.8U6.719, Journal of the Audio Engineering; iSociety, Vol 15, No. k, October 1967, pp. 383-388).

8103 124 \ - 10 -

However, these filters serve a completely different purpose, namely, manual splitting of the noise reduction effect into separate sideways and do not affect the upper or lower frequency extremes of the compressor input signals.

Likewise, the sharp cut-off hand limiting filter as previously described has kink point frequencies deliberately outside the useful system bandwidth because they are not intended to affect the upper or lower frequency bandwidth extremes.

Until now, it has been considered undesirable to significantly change the frequency spectrum within the useful system bandwidth. For example, in professional recording, it is inconceivable to provide the break point frequency of some upper band limiting filter below 20 kHz. Likewise, in FM broadcasting, a 15 kHz upper band pass limit is rigorously maintained across all audio signal stages.

Although the invention is not limited to use in some particular type of noise reduction compression expansion system, it will improve the operation of all types of compression expansion members including wide band compression expansion members. the invention is particularly useful in oblique band noise reduction systems. Examples of such systems can be found in US-PS 25 Re 28Λ26 and US-PS 3,757.25¾. In such sliders with sliding tape, high frequency audio compression or expansion is achieved by applying a high frequency boost (for compression) or confining (for expansion) by means of a high pass filter with. a variable lower breakpoint frequency. As the signal level in the high-frequency band increases, the filter breakpoint frequency shifts upward to narrow the chased or contained band and exclude the usable signal from the chasing or limiting. In sliding band devices, one effect of uncertainties in the recording / playback response is the resulting modulation of low level 8103124-11 middle signal signals when high frequency signals in the region of recording / playback uncertainties are present at the compression member input. In two-way sliding band device of the kind described in US-PS Re 28.1 * 26, this effect can be suppressed by providing the sudden high frequency drop in the side path of the device. While such a construction provides the sudden drop only at mid and low signal levels, it is sufficient to practically meet the purpose of the invention in two-way devices for both the sliding band type (US-PS Re 28.1 * 26) and the fixed belt type (US-PS 3.81 * 6.719).

The invention will be elucidated hereinbelow with reference to the drawing.

Figure 1-1 * are block diagrams representing alternative placements of the spectral bevel networks according to the invention.

Figure 5 is a response curve generally showing the characteristics of a high-frequency spectral bevel network and complementary straightening network for use in a cassette tape recorder display.

Figure 6 is the standard CCIR characteristic curve with noise weighting.

Figure 7-10 are representative response curves of typical cassette tape recorder / reproducers.

Figures 11-16 are representative curves useful for understanding the invention.

Figure 1-1 * shows general block diagrams to show the different places where spectral bevel networks according to the invention can be installed.

In Figure 1 which shows the simplest and preferred embodiment, the spectral bevel network (with a low or high frequency part) or networks (with both low and high frequency parts) 2 is located in the input signal path to an eompression device 1 * whose output is applied on '35 is a transmission channel H. On the reproduction side of channel 8103124-12, a complementary expansion member 6 acts on the reproduced signal and applies it to an optional spectral straightening network 8 which has a characteristic complementary to the network in the entrance way of the eompression-5 organ. This location of the network is particularly advantageous where the compression member and the expansion member each contain two or more series devices as described in Audio, May 1981. pp. 20-26.

Figure 2 indicates a similar construction 10 which is not preferred in practice because it is more complicated, requires an additional chain and is more expensive.

In this similar construction, a spectral bevel network 10 is placed in the control circuit of the compression member and a further spectral bevel network 12 is located in the signal output 15 away from the compression member. When facultative straightening is applied on the reproduction side, a complementary straightening network 1 ^ is applied in the input path of the expansion signal and a spectral skewing network 10, which has the same characteristics as the network 10 in the control circuit of the compression member arranged in the control circuit of the expansion member 6. The characteristics of the networks and 12 may differ slightly from those of the network 2 and from each other to obtain the same overall results as the network 2. This remark also applies to networks 1¾ and 16. Where the compression member U and the expansion member 6 each contain series devices as described in. the audio article, only in the control circuit of the first compression device is a spectral oblique network 10 where a network 12 located only in the output path 30 of the first compression series device and (facultatively) a network 1U only in the input path of the last expansion series device with a network 16 only in its control circuit.

Figures 3 and U indicate the location of the spectral bevel networks in the sideways of compression bodies and 35 two-way expansion bodies. Such compression and 810 3 12 4

a

Expansion structures are well known per se and will therefore not be discussed in detail. However, there are two main forms for the further path N (20). An alternative (Figures T and 8 from US-PS 3,846,719) is a filter followed by a controlled limiter which ensures that it limits progressively as the signal level rises with a rectified and smoothed control signal. Another alternative (US-PS Ee 28.4-26) is a high-pass sliding band filter whose passband is progressively narrowed by the control signal 10 to exclude large signal components from the output of the filter. Advantageous breakpoint frequency values for the variable filters are about 375 Hz in the setting state, but progressively narrower high-pass in response to the control signal. Compression members and two-way expansion members (single or in series) may also be used in the construction of Figures 1 and 2.

In Figure 3, the spectral oblique network 18 is placed in the input signal path of the noise reduction side-way circuit 20 of the compression member 22 and (facultatively) 20 the expansion member 24. In Figure 4, a similar construction is shown in connection with a compression member 26 and expansion member 28 of the sliding band type. In this similar construction, a spectral oblique network 28 is placed in the side path control circuit 30 which controls a frequency-25 variable inductance or filter circuit 32 with sliders the band. A further spectral oblique network 24 is located in the side road exit. The same networks are facultatively located in the side road of the expansion organ. The constructions of Figures 1 and 2 are preferred over those of Figures 3 and 4 because the former operate at all signal levels and thus reduce channel overload or band saturation effects at the band extremes.

Although only a single side road is shown in Figures 3 and 4, several side roads can be used as for example in US-PS 3,846,719. Furthermore, the side roads 8103 124 * - 14 - can be constructed in such a way that the compression side road has feedback, and the expansion side road a forward feeding construction such as for example in US-PS 3,903Λ85. Where in two-way serial devices are used in a compression member and an expansion member, such as the type described in Audio, May 19? Ϊ́, pp. 20 - 26, it is sufficient to use a spectral bevel network in the first series arrangement in the compression member and (culturally) in the last series arrangement in the expansion member, although the construction of Figure 1 is preferred.

Preferably, the spectral skew characteristic has: a) a declining break point frequency within the range in which the transmission channel or tape recorder response, including that of the expansion band pass filter, is reasonably dependent - ie slightly below (above) the frequency at which the transmission channel or the tape recorder response is uncertain whether the kink starts.

b) A sudden drop to provide a well-defined boundary for the frequencies controlling the chain.

c) A well defined shape that follows the drop to allow for easy formation of the reciprocal characteristic during reproduction or playback to maintain an overall flat frequency response (if desired).

D) A shape that takes full advantage of the low-level noise sensitivity characteristic of the human ear — that is, a frequency response drop as steep and deep as possible. without introducing a perceptible increase in the noise level when the reciprocal characteristic is applied during playback.

While the spectral bevel characteristic effectively reduces the noise reduction effect above (below) the sharp kink frequency, when the frequency is above about 8 kHz (or below about 50 Hz), the increased noise will not be heard in response to the human ear response .

81 0 3 12 4 - 15 - low level high and low frequency noise, especially when the noise level is extremely low, as is the case when the invention is used in a tape recording ecompression expansion device. This somewhat surprising aspect of the invention has been established experimentally.

Justification for this treatment can also be found in the form of the CCIR noise rejection curve shown in Figure 6. The curve shows the low level noise sensitivity of the human ear. It is noted that the low sensitivity at 10 low frequencies drops very rapidly above a peak of about 6-7 kHz. Thus, there is a reduced psychoacoustic need to maintain a significant noise reduction at frequencies above 8-10 kHz. This is the high-frequency opposite of the perceived ability of noise reduction systems to deliver a subjectively large noise reduction even though the low frequencies have been little or not treated at all. Good design can eliminate hum which is the only low-frequency noise that is subjectively annoying in cassette tape recording.

In professional noise reduction systems where low-frequency noise reduction is provided as an insurance against unexpected buzzing problems, there is usually little need for any noise reduction below the lowest buzzing component likely to be encountered (ie 50 Hz).

In particular at the high-frequency end of the spectrum, the use of a spectral bevel network does not avoid or form a total band limiting filter, which is sometimes known as a "multiplex filter" ("MPX"). The reasons have been given for this. As discussed, a band-30 cutoff filter, commonly used for both recording and playback, has several functions that are only superficially related to those in these discussions. Therefore, even in the case of an ideal signal channel, it is desirable in encoding and decoding to have both: 1) bandwidth limiting filters; 8103 124 * - 16 - 2) spectral beveling and straightening networks.

The spectral networks (2, 10, 12, 18, 28, 3b) provide a sudden self-induction, dive or descend as shown in figure 5 · The intention of the dashed lines is to indicate 5 that the final high frequency ( low frequency response) may not be accurate as indicated by the solid lines. .

A suitable form of the spectral bevel network (10, 2, 12, 18, 28, 3 * 0 »is a sharp low pass (high pass 10 pass filter), with an 18 dB / octave slope and with a break point frequency within the rapidly falling part. of the CCIR noise-weighting curve (Figure 6) and below (above) the top cut-off frequency of the transmission channel A break point or cut-off frequency of about 8-10 kHz (50 Hz) in the low-frequency end) would be suitable for a high quality tape deck with a usable but uncertain response up to a few 15 kHz (or 30 - 60 Hz at the low-frequency end).

The network could also be in the form of a self-induction network with approximately a 10 dB bottom as shown in Figure 5

Another suitable form of the network is a notch filter with a center frequency of about 16 kHz (20 Hz), a Q such that a break point frequency of about 8-10 kHz (40 Hz) is obtained and a depth of some 10 dB.

A double tuned (staggered tuned) notch filter can also be used, especially in professional applications, to provide a wider overall notch where the second notch is placed on some part of the octave above the first notch ( eg 1/3 octave), 30 A depth of some 10 - 15 dB has been found experimentally to eliminate the mid-band modulation effect in the most difficult cases. However, it has been found that a depth of only 6 dB significantly improves the mid-band modulation effect, especially when the drop is very sudden such as 18 dB per octave.

8103124 4 V - 17 -

When a flat overall response is desired, the same network and / or complementary networks 8, 10, 1U, 18, 28, and 3 ^ are used in the display or playback portion of the system.

With reference in particular to the characteristics of the consumer compaet cassette audio tape recorder and playback devices, to more fully appreciate the invention, reference is made to Figures T - 9 which show representative measured high frequency response curves at input levels low enough to detect tape saturation. avoidance for several typical compact cassette recorders. Figure 10 shows representative high frequency response curves at various input levels for another typical compact cassette recorder. It is then noted in Figure 7 that the recorder response drops rapidly beyond 10 kHz. In Figure 8, there is an increase in response starting at about 10 kHz with a pronounced peak at about 17 kHz. The response in Figure 9 has a high frequency peak at 15 kHz with a rapid drop in response beyond this frequency.

The response in figure 10 for a level of -20 dB, which avoids saturation, is almost ideal with being practically flat up to 20 kHz. However, such a good response is rare.

The choice of a high frequency spectral bevel network kink frequency of about 10 kHz is a good choice for such devices because the response curves in Figure 7-10 show that for levels below saturation, the typical recorder has few defects in response below a good setting. 10 kHz. Thus, at most levels, the spectral bias network will ensure that there is essentially no level discrepancy in the playback control signal caused by uncertainties in the response at extremely high frequencies. A breakdown frequency of about 10 kHz is a good choice for such devices because, in addition, this frequency 35 lies on the steeply descending portion of the CCIR noise rejection curve 8103 124 - 18 (Figure 6) and thus may allow some loss of noise reduction by human ear,

In choosing an appropriate break point frequency for a spectral bevel network, the designer can select approximate frequencies different from 10 kHz based on the parameters of his system. For example, in the case of a higher quality transmission channel, a higher break point frequency may be acceptable. With respect to the compact cassette devices of the kind whose characteristics are indicated in Figures 7-10, acceptable high frequency break point frequencies can range from about 8 kHz to perhaps 11 or 12 kHz. While a filter of about 18 dB / octave is desirable to ensure that high level high (low) frequency signals do not control the compression member, a drop of 15 but 12 dB / octave will also practically meet the purposes of the invention for most signals. Attenuations much sharper than 18 dB / octave present difficulties in providing complementary straightening and are more costly.

The required filter drop value depends in part on the sensitivity of the compression member to signals beyond the filter cutoff frequency. For example, consider the sliding bande ompres sie-or-go with two ways as indicated in US-PS Re 28.1 * 26. In this arrangement, a high-frequency pre-emphasis is used in the control circuit of the compression member such that when a signal as indicated in Figure 11 is applied to the compression member (such as a signal may be generated by a broadband compressing means). loud), the pre-emphasis of the control circuit causes the signal to have an energy spectrum as shown in Figure 12.

30 The pre-emphasis signal spectrum has a peak. After rectification, this peak provides the DC control signal that controls the sliding band action of the compression member.

Figure 13 shows the uncertain frequency responses of the tape recorder channel shown for four representative compact cassette tape recorders a, b, c and d. The effect on the 8103 124-19 spectrum of Figure 12 is to ensure that four different spectra are present in the control circuit of the expander (decoder), resulting in four different DC control signals. Clearly, 5 decoding errors will be the result.

In such a case, a desirable spectral oblique network characteristic causes the expander (decoder) to generate the same DC control signal at least as shown in Figure 15- A network-10 characteristic with a break point frequency being about 10 kHz, as shown in Figure 5 fit. Note that the network does not eliminate frequency band shifting; in fact, it can be only slightly reduced. The scrolling (or compression as in the band splitting system of US-PS 15 3.81 * 6.719) now becomes available during playback.

Attention is drawn to the fact that the primary object of the invention can be seen in Figure 15 - namely to ensure that the same peak in the spectrum of the AC current control signal is present at the equivalence point in both the compression member and the expansion organ.

With regard to the sliding band systems of the just mentioned type, the spectral bevel network is particularly useful in suppressing mid-frequency modulation caused by the presence of high level, high frequency signals in the compression means that are not recovered and are attached to the expansion body.

This modulation effect, which occurs very rarely with actual music sources, relates to the basic operation of the sliding band arrangement in imperfect signal channels: a predominant signal effectively controls the sliding band frequency characteristic and can cause audible effects when this signal has not recovered a high frequency during playback. When this predominant signal has a high frequency, importantly above the frequency of a low-level mid-frequency signal, the mid-frequency modulation effect is audible - when the high-frequency signal is intermittent as with thumping sounds, for example, brushed cymbals and is not displayed at the same level by the tape recorder. In that case, the intermediate frequency signals are amplitude modulated even after decoding because the high frequency signal causes the shifting band response to vary without complementary expansion in the playback decoder. It is noted that this effect is not a band saturation effect in soil. It can be caused by inaccurate adjustment and equalization or by slit loss, poor azimuth and the like. However, it is clear that the effect will be worse if there is also saturation in the controlling frequency range.

This question of low-level, mid-frequency-15 modulation effects can be better understood with reference to Figure 16 which shows the results when a series of pilot tone curves using a 15 kHz signal at levels from 0 to -6o dB and below and a sweeping low-level test. tone recorded at - 65 dB and played back on a 20 sliding band system.

Note that when a dominant 15 kHz signal increases in amplitude from -60 dB to -50 dB, for example, the output of the compression member changes by about 2 dB. This 2 dB change must be accurately preserved in the recording process due to the dependence of the center frequency dynamics on this dominant signal (a change of a few 10 dB is provided in the 1 kHz range). Thus, any error in playback in the controlling frequency range is multiplied significantly in the mid-frequency range.

When a noise reduction system treats low frequencies, it will have a corresponding effect at low frequencies. An extremely low frequency source in the input signal to the compression member can drive the circuit of the compression member. If the recorder does not reproduce the source components ~~ 35, signal modulation effects at the 8103124-21 output of the expander will be evident.

8103124

Claims (13)

1. Signal compression member for use in a signal transmission system in which a compressed signal is applied to a transmission channel which in turn is connected to an expansion member, wherein the relative signal amplitude response in the transmission channel between the compression member and the expansion member is uncertain in the region of the high and / or low frequency extremes of the useful bandwidth of signals applied to the compression member 10 and the compression of the compression member being controlled in response to frequency and amplitude level characteristics of the applied signals, the compression member includes a band-pass filter with a break point frequency and filter slope such that the high and / or low frequency extreme ranges are practically excluded from controlling the compression member, characterized by a chain device for damping the high and / or low frequency extreme areas in the compressed signal addition to practically excludes and from the high and / or low frequent extreme regions in controlling the compression member. Signal compression means according to claim 1, characterized by at least one band reject network with at least one steep drop with a break point frequency located to damp the high and / or low frequency extreme areas. Signal compression means according to claim 2. characterized by at least one band rejection network with one or more drops from about 12 cLB / octave to 18 cLB / octave. A signal compression member according to any of claims 2 or 3, characterized by at least one network located in the input signal path to the compression member. A signal compression member according to any of claims 2 or 3 having a control circuit responsive to the frequency and amplitude of applied signals for controlling the compression, characterized by at least one band rejection network located in the control circuit and at least further band whey - 8103124 - 23 - network located in the output path of the compression member. 6. Signal compression member according to any one of claims 2-5, characterized in that at least one band rejection network contains a low-pass filter in the high-frequency extreme range and / or a high-pass filter in the low-frequency extreme range. Signal compression member according to any one of claims 2 or 4-6, characterized in that the at least one band rejection network contains a notch filter in the high and / or low-frequency extreme range. 8. Signal compression means according to any one of claims 2-6, characterized in that the at least one band rejection network comprises a self-induction network in the high and / or low frequency extreme range. Signal compression member according to any one of claims 2-8, characterized in that the band rejection network or networks have a depth of 6 - 10 dB or more in the high and / or low-frequency extremes. 10. A signal compression member as claimed in any one of claims 1 to 9 in combination with an expansion member for expanding the dynamic region of signals compressed by the compression member and received via a transmission channel, the expansion member being a chain device 25 contains regions that are too extreme for boosting the attenuated high and / or low frequencies to achieve a substantially flat frequency response. A signal compression member according to any one of claims 1 to 3, wherein the compression member has a main signal-30 path that is linear with respect to the dynamic region, a combination path in the main path and a further path whose input is connected to the entrance or exit of the main road, the output of which is connected to the combination chain, the further road providing a signal that drives the main road signal through the combination chain but that 81 03 124 - 2k - is limited to a value smaller then the main signal, characterized in that the chain device is located only in the further distance from the compression member. Signal compression means according to claim 11, characterized in that at least one network is located in the input signal path of the further road. 13. A signal compression member according to claim 11, with a control circuit further responsive to the frequency and amplitude of applied signals for controlling the compression, characterized by at least one band reject network located in the control chain and at least one further band reject network in the output signal path of the further road. 1 ^. Signal compression means according to either of claims 12 or 13, characterized in that at least one network 15 contains a low-pass filter in the high-frequency extreme range and / or a high-pass filter in the low-frequency extreme range. * Signal compression member according to either of claims 12 or 13, characterized in that at least one network contains a notch filter in the high and / or low-frequency extreme range. Signal compression means according to either of Claims 12 or 13, characterized in that at least one network contains a self-induction network in the high and / or low-frequency extreme range. 17 17 Device substantially as described in the description and / or shown in the drawing. 81 0 3 1 2.4