WO1998047223A1

WO1998047223A1 - Audio dynamic range processor with adjustable signal observation window

Info

Publication number: WO1998047223A1
Application number: PCT/US1998/007350
Authority: WO
Inventors: G. Berchin; James Mctigue; Robert R. Turnbull
Original assignee: Telex Communications, Inc.
Priority date: 1997-04-15
Filing date: 1998-04-15
Publication date: 1998-10-22
Also published as: AU7111998A

Abstract

The present invention provides an audio dynamic range processor. The processor uses an averaging detector for receiving an input signal and for generating an average signal magnitude (40) that is an average value of the input signal over an observation window of adjustable duration (32). The processor also has a dynamic processor which receives the input signal and adjusts its gain based upon the magnitude of the average signal magnitude.

Description

AUDIO DYNAMIC RANGE PROCESSOR WITH ADJUSTABLE SIGNAL OBSERVATION WINDOW

BACKGROUND OF THE INVENTION

The present invention relates to an audio processing system and more particularly to an audio dynamic range processor with an adjustable signal observation window.

Past audio dynamic range compressors and limiters provide most or all of the following:

1) means for computing the magnitude of an audio signal;

2) means for computing the compressor/limiter attenuation appropriate for the signal magnitude computed in (1);

3) means for delaying application of the attenuation computed in (2), implemented as "attack time" (period of time over which the compressor/limiter attenuation is gradually increased), and/or "hold time" (period of time during which the compressor/limiter attenuation is held constant), and/or "release time" (period of time over which the compressor/limiter attenuation is gradually decreased).

The various "times" indicated in (3), in combination with the "observation window" (effective window-of-time over which the signal magnitude is computed) in (1), are selected in a manner designed to attain sonic characteristics appropriate for different types of audio signals. The observation window in (1) is generally either fixed at a moderately long time interval such that the average magnitude is computed, or fixed at an extremely short time interval such that instantaneous magnitude is computed, or manually selectable to either of the two fixed intervals.

In these devices, the attack time of (3) commences immediately upon the close of the observation window. The attack time is adjusted with the intent that longer- duration/ lower-frequency signals are handled differently than shorter-duration/ higher-frequency signals with regard to the amount of attenuation applied. The hold time and release time parameters are considered separately from the attack time parameter.

SUMMARY OF THE INVENTION The present invention provides an audio dynamic range processor. In one embodiment the processor uses an averaging detector for receiving an input signal and for generating an average signal magnitude that is an average value of the input signal over an observation window of adjustable duration. The processor also has a dynamic processor which receives the input signal and adjusts its gain based upon the magnitude of the average signal magnitude.

In a more specific embodiment, the observation window is implemented using a lowpass filter operating upon the absolute value or squared-magnitude of the instantaneous amplitude of the input signal. The observation window duration is the inverse of the cutoff frequency of the lowpass filter. The observation window duration is adjusted by adjusting the cutoff frequency of the lowpass filter. The dynamic processor has an attack time interval over which the gain of the input signal is gradually adjusted. The attack time interval commences almost immediately upon the close of the observation window. The cutoff frequency associated with the "observation window" and the time interval associated with the "attack time" are separate, independent, and adjustable. In this way the spectral response of the compressor may be adjusted by means of the "observation window" parameter while the temporal response may be independently adjusted by means of the "attack time" parameter. The link between spectral and temporal responses is thus broken, and sonic characteristics previously unattainable are made possible.

In another embodiment, audio dynamic range signal processor uses an averaging detector for receiving an input signal and for generating an average signal magnitude that is an average value of the input signal over an observation window of adjustable duration. The processor also has a compressor that compresses the input signal when the average signal magnitude exceeds a predetermined threshold level. The compressor employs variable gain that is dependent upon the average signal magnitude.

DESCRIPTION OF THE DRAWINGS The present invention will be readily understood with reference to the following specification and drawings wherein;

FIG. 1 is a block diagram of the system; and FIG. 2 is a software flow chart for the present invention. DETAILED DESCRIPTION Figure 1 shows a block diagram of the components of a preferred embodiment of the audio signal processor with adjustable observation window invention. The preferred embodiment uses digital signal processing. However, the invention may also be implemented in the analog domain.

Referring to Figure 1, analog audio input signals designated by the arrow 20 are applied to an analog to digital converter (ADC) 22, for example a Crystal Model No. CS 5390. The input signal may instead be some component of a video signal. The digitized audio signals from the ADC 20 are then applied to a digital signal processing (DSP) computer 24, for example a Motorola Model No. DSP 56004 and processed as set forth in FIG. 2 and described below. The output of DSP 24 is then applied to a digital to analog converter (DAC), for example, a Crystal Model No. CS 4329, and converted back to analog signals, as represented by the arrow 28.

The software for one embodiment of the DSP 24 is illustrated in FIG. 2. A single sample of audio data is received from the A/D converter 22. The duration of the observation window is selected by an operator in step 32, based upon the dynamics and the spectral content of the audio signal and of listening preferences. The observation window is the effective window of time over which the signal magnitude is computed. The absolute value or magnitude of the audio sample received in step 32 is computed by the DSP 24 in step 34. Subsequently, in step 36, the exponential time constant "TC" is computed using the formula indicated and using the user-selected window duration. The sample rate selected for the preferred embodiment is 48,000 samples per second, however, other sample rates could be used. For the purposes of this representative embodiment, variable TC's exponent includes a multiplier set to three (3), thereby establishing time, TC, as spanning three mathematical time constants of exponential decay. The use of three mathematical time constants of exponential decay is a duration constant commonly used in signal processing.

The Current Level "CL" parameter that was calculated for the previous audio sample is stored in computer memory in step 38. The CL parameter is then calculated in step 40 for the current audio sample using the TC parameter computed in step 36, the absolute value of the current audio sample computed in step 34, and the previous value of the CL parameter stored in step 38. This algorithm implements a first-order lowpass filter through which the absolute values of the audio data are passed, thus representing the averaged level, or average signal magnitude, of the audio data. Steps 32 - 42 are therefore a detection of the average signal magnitude. Such first-order lowpass filters algorithms ^"as used by the average detector in steps 32 - 42 are well-known in the art. Many types may be used, for example, chapter 6.4 of the textbook, A. Oppenheim, A. Willsky, and I. Young Signals and

Systems (1983, Prentice-Hall, Inc.), provides different types of lowpass filter algorithms that include a time constant.

The lowpass filter algorithm of the preferred embodiment has an adjustable cutoff frequency. The observation window duration is simply the inverse of the cutoff frequency of the lowpass filter. Therefore, the observation window duration is adjusted by adjusting the cutoff frequency of the lowpass filter. If the observation window is adjusted to a moderately long time interval, such as one that envelopes several digital audio samples or a longer analog audio segment, the resulting CL will reflect a true average magnitude of the signal. Conversely, if the observation window is set to a relatively short time interval, such as one that would envelope a single digital audio sample or a short analog audio segment, the average detector will generate a value corresponding to the instantaneous or peak magnitude of the input signal.

In step 42, the CL parameter for the current audio sample is stored in computer memory to be used as the "Previous Level" parameter for the subsequent audio sample. Thus, even though the observation window may encompass and use several audio samples, the average detector of steps 32 - 42 may regenerate the CL parameter after the reception of each ensuing audio sample.

The average signal magnitude calculated in step 40 controls processing of the input signal performed by a dynamic processor of the DSP 24. The dynamic processor is used for adjusting the amplitude or signal level of the input signal with a variable gain. The dynamic processor's variable gain is dependent upon the average signal magnitude; thus, the average signal magnitude controls the dynamic processor's processing by setting the dynamic processor's gain.

The dynamic processor described above, like typical dynamic processors known in the art, may be comprised of a compressor, a limiter, a gate, an expander, a ducker, a de-esser, or any combination thereof. As is known in the art (see, e.g., B. Hurtig Multi-Track Recording for Musicians at pp. 55-59 (1988, Alfred Publishing Co.)), dynamic processors control the amplitude levels, or dynamic range, of audio signals. Dynamic processors, such as compressors, limiters, de-essers, and duckers, reduce the dynamic range of audio signals. Conversely, dynamic processors, such as expanders and gates, increase the dynamic range of audio signals. For instance, compressors reduce the level of signals which pass a threshold value, while limiters prevent the level of signals from exceeding the threshold level. Expanders, on the other hand, reduce the level of signals which fall below a threshold value, while gates virtually silence signals that fall below the threshold level. Duckers typically compress, or suppress, a second signal under the amplitude of the input signal when the input signal magnitude exceeds a threshold level. De-essers, as their name indicates, are designed to reduce the "hiss" associated with sibilance. De-essers accomplish this by compressing the appropriate frequencies (approximately 2000 - 6000 Hz) of an input signal associated with sibilance ("sibilance frequencies"), when the signal level in the sibilance frequencies of the input signal exceeds a threshold level.

Past dynamic compressors, however, have not used the input signal's average signal magnitude, calculated over an adjustable observation window as described above, as the basis for comparison against the threshold level. The embodiment shown in steps 44 - 52 in Figure 2 employs a compressor as the dynamic processor. In step 44, the average signal magnitude, represented by the CL parameter, is compared with the previously established threshold. If the amplitude is above the threshold then control passes to step 46. If the amplitude is less than or equal to the threshold then control passes to step 48. The amplitude of the audio data is determined in step 44. If the amplitude is above the threshold, the value of the Gain parameter (which is actually an attenuation in this case; "gain" is used here in its generic sense so that it can represent either boost or cut) is determined in step 46 such that it will reduce the amplitude, or signal level, of the audio data to the appropriate level based upon the previously established compression ratio settings. Methods for determining the gain value are well known in the art.

If the amplitude of the audio data is determined to be below the threshold, the gain is set to 1.0 in step 48. In particular, signals with amplitude below the threshold are unaffected by the compressor, therefore the gain parameter is set to 1.0. Sudden changes in the value of the gain parameter are prevented through the use of "Attack Time" (in the case of a reduction in Gain) or "Release Time" (in the case of an increase in Gain) implementations that ramp the Gain value gradually from its previous value to its new value in step 50. Methods for implementing Attack and Release Times are well-known in the art. The adjusted value of the current audio sample is computed as the gain value determined in step 50 and the audio sample obtained in step 30. The adjusted audio sample is applied to the DAC 26 in step 54.

While a preferred embodiment of the present invention has been described, it should be understood that various changes, adaptations and modifications may be made therein without departing from the spirit of the invention and the scope of the appended claims.

Claims

WHAT IS CLAIMED IS:

1. An audio dynamic range signal processor, comprising: an averaging detector for receiving an input signal and for generating an average signal magnitude, the average signal magnitude being an average value of the input signal over an observation window duration, the observation window having adjustable duration; and a compressor for compressing the input signal when the average signal magnitude exceeds a predetermined threshold level, the compressor having variable gain dependent upon the average signal magnitude.

2. The signal processor of claim 1, wherein the average value of the input signal generated by the averaging detector is the root-mean-square average value of the input signal.

3. The signal processor of claim 1, wherein the averaging detector comprises a lowpass filter algorithm receiving the squared-magnitude of the input signal.

4. The signal processor of claim 1, wherein the averaging detector comprises a first-order lowpass filter algorithm receiving the absolute value of the input signal.

5. The signal processor of claim 4, wherein the lowpass filter algorithm has a variable cutoff frequency, the cutoff frequency being the inverse of the observation window duration.

6. The signal processor of claim 1, wherein the average signal magnitude approaches the peak magnitude of the input signal as the observation window decreases.

7. The signal processor of claim 1, wherein the observation window is manually adjustable.

8. The signal processor of claim 1, wherein the input signal comprises a series of discrete samples of an analog waveform.

9. The signal processor of claim 1, wherein the input signal comprises a digital signal.

10. The signal processor of claim 8, wherein the average signal magnitude is regenerated with each discrete sample.

11. The signal processor of claim 10, wherein the averaging detector comprises a lowpass filter algorithm receiving the squared-magnitude of each discrete sample.

12. The signal processor of claim 10, wherein the averaging detector comprises a lowpass filter algorithm receiving the absolute value of each discrete sample.

13. The signal processor of claim 12, wherein^" the lowpass filter algorithm has a variable cutoff frequency, the cutoff frequency being the inverse of the observation window duration.

14. A method of processing an audio dynamic range signal, comprising: receiving an input signal; adjusting the duration of an adjustable observation window; generating an average value of the input signal over the observation window defined as the average signal magnitude; and adjusting the amplitude of the input signal with variable gain, the gain being dependent upon the average signal magnitude.

15. An audio dynamic range signal processor, comprising: an averaging detector for receiving an input signal and for generating an average signal magnitude, the average signal magnitude being an average value of the input signal over an observation window duration, the observation window having adjustable duration; and a dynamic processor for adjusting the amplitude of the input signal with variable gain, the dynamic processor gain being dependent upon the average signal magnitude.

16. The signal processor of claim 15, wherein the dynamic processor has an attack time duration over which the dynamic processor adjusts the amplitude of the input signal, the attack time commencing immediately upon the close of the observation window.

17. The signal processor of claim 15, wherein the dynamic processor comprises a compressor, the compressor for compressing the input signal when the average signal magnitude exceeds a predetermined threshold level, the compressor having variable gain dependent upon the average signal magnitude.

18. The signal processor of claim 15, wherein the dynamic processor comprises a limiter, the limiter for limiting the amplitude of the input signal when the average signal magnitude exceeds a predetermined threshold level.

19. The signal processor of claim 15, wherein the dynamic processor comprises an expander, the expander for expanding the amplitude of the input signal when the average signal magnitude falls below a predetermined threshold level, the expander having variable gain dependent upon the average signal magnitude.

20. The signal processor of claim 15, wherein the dynamic processor comprises a ducker, the ducker for compressing the amplitude of a second signal under the amplitude of the input signal when the average signal magnitude exceeds a predetermined threshold level, the ducker having variable gain dependent upon the average signal magnitude.

21. The signal processor of claim 15, wherein the dynamic processor comprises a gate, the gate for squelching the input signal when the average signal magnitude falls below a predetermined threshold level.

22. A signal processor, comprising: an averaging detector for receiving an input signal and for generating an average signal magnitude, the average signal magnitude being an average value of the input signal over an observation window duration, the observation window having adjustable duration; and a dynamic processor for adjusting the amplitude of the input signal with variable gain, the dynamic processor gain being dependent upon the average signal magnitude.