Data compression method and apparatus

• US 4,546,342 A
• Filed: 12/14/1983
• Issued: 10/08/1985
• Est. Priority Date: 12/14/1983
• Status: Expired due to Fees

• Current Assignee: Digital Recording Research Limited Partnership
• Original Assignee: Digital Recording Research Limited Partnership
• Inventors: Weaver, Charles S., Sweeney, Lawrence E. Jr., LeBlanc, Robert A.
• Primary Examiner(s): Miller, Charles D.
• Application Number:
  US06/561,723
• Time in Patent Office: 664 Days
• Field of Search: 360/32, 375/31, 375/122, 375/32, 381/29-35, 340/347 AD, 340/347 DD, 358/261
• US Class Current: 341/51
• CPC Class Codes: H03M 7/42 using table look-up for the coding or decoding process, e.g. using read-only memory H03M7/4006 takes precedence

Abstract

A music data compression system is disclosed which includes an analog to digital converter for converting the analog music signal to digital sample signal form, a digital compression filter for compression filtering the digital sample signals, and an encoder for truncated Huffman encoding the compression filter output. An entropy setting unit is included in the system for step control of the signal supplied to the digital compression filter in accordance with one or more threshold levels of the envelope of the music signal, and for reducing the entropy of the signal with increases in the energy level of the music signal. Different codes may be implemented in accordance with the threshold(s), and an identifying code word is inserted in the Huffman encoded stream. A decoder, digital reconstruction filter, and digital to analog converter are used to reconstruct the analog music signal. The identifying code word in the encoded signal stream to the decoder is used to select the necessary algorithm for decoding, and an entropy setting unit is included for step control of the gain of the digital reconstruction filter output under control of the identifying code word when step gain control is included prior to compression filtering.

Claims

1. In a data reduction system of the type for preparing varying average energy level analog input signals for storage or transmission, the combination comprising

analog to digital converter means for converting the analog input signal to equal length digital sample signals,

digital compression filter means responsive to digital sample signals from said analog to digital converter means for generating a stream of equal length compressed signals,

digital encoding means implementing a truncated variable word length code for encoding the compressed signals from said digital compression filter means,

entropy setting means for controlling the entropy of signals supplied to said digital compression filter means to control the entropy of signals from the compression filter,

mode control means for obtaining a measure of the average energy level of the input signal prior to encoding by said digital encoding means and for producing an output indicative of an energy level band within which said signal falls, there being a plurality of different energy level bands ranging from low to high energy levels through which said signal may operate, and

means for controlling the entropy setting means in response to the output from the mode control means for step control thereof with energy level band changes, the entropy of signals supplied to the digital compression filter means being reduced by said entropy setting means with changes from a lower to a higher energy level band and being increased with changes from a higher to a lower energy level band, the ratio σ

/q being changed by a factor of 2^x with each change in energy level band, wherein σ

is standard deviation of the compressed signal stream from the digital compression filter means, q is quantization level of the digital sample signals, and x is a non-zero integer.

2. In a data reduction system as defined in claim 1 wherein said mode control means comprises

envelope detector means for obtaining a measure of the average energy level of the input signal, and

threshold means responsive to the output from the envelope detector means for producing a change in the output from the mode control means when the output from the envelope detector means exceeds a threshold value of said threshold means.

3. In a data reduction system as defined in claim 1 wherein said mode control means is responsive to the input to the analog to digital converter means for obtaining a measure of the average energy level of the analog input signal to the digital converter means.

4. In a data reduction system as defined in claim 1 wherein said mode control means is responsive to the output from the analog to digital converter means for obtaining a measure of the average energy level of the digital sample signals.

5. In a data reduction system as defined in claim 1 wherein said mode control mean is responsive to the output from the digital compression filter means for obtaining a measure of the average energy level of the compressed signal stream.

6. In a data reduction system as defined in claim 1 wherein said entropy setting means comprises means for setting at least one least significant bit of the sample signal to a predetermined value with energy level changes from a lower to a higher energy level, q, band to increase the quantization level of the sample signal by a factor of 2^N where N is the number of least significant bits set to predetermined values.

7. In a data reduction system as defined in claim 6 wherein said mode control means comprise

envelope detector means, and

threshold means responsive to the output from the envelope detector means for producing a change in the output from the mode control means when the output from the envelope detector means passes through a threshold value of said threshold means.

8. In a data reduction system as defined in claim 1 wherein said digital encoding means has a plurality of operating modes for implementing a plurality of different truncated variable word length codes, and means responsive to the output from said mode control means for selecting the operating mode of the digital encoder means for operation with a code for optimum reduction in the bit rate of the encoded compressed signal output therefrom, changes in the output from the mode control means simultaneously effecting changes in the operating mode of the digital encoding means and entropy setting means.

9. In a data reduction system as defined in claim 8 including,

means responsive to the mode control signal output from said mode control means for inserting an identifying code word in the encoded compressed stream from the digital encoding means with changes in the output from the mode control means for identifying the code being implemented by said digital encoding means.

10. In a data reduction system as defined in claim 9 including,

means for periodically inserting said identifying code word in the encoded compressed stream after every change in the identifying code word.

11. In a data reduction system as defined in claim 9 including,

code checker and stripper means for stripping the identifying code word from the encoded compressed signal stream from the digital encoder means and for producing a code identification signal corresponding to said identifying code word,

digital decoder means responsive to the encoded compressed signal stream from said code checker and stripper means and having a plurality of different operating modes for decoding the different codes implemented by said digital encoding means,

means responsive to the code identification signal from said code checker and stripper means for selecting the operating mode of the digital decoder means required for decoding the encoded compressed signal stream from said code checker and stripper means, and

digital reconstruction filter means responsive to decoded signals from said digital decoder means for reconstruction filtering thereof.

12. In a data reduction system as defined in claim 1 wherein said entropy setting means for controlling the entropy of signals supplied to said digital compression filter means comprises gain setting means for changing the amplitude of the sample signal whenever the energy level of the input signal changes from one to another energy level band, the standard deviation, σ

, of the compressed signal stream being changed by a factor of 2^x without a change in the quantization level of the sample signal stream by operation of said gain setting means, and

means responsive to the mode control signal output from said mode control means for inserting an identifying code word in the encoded compressed signal stream from the digital encoding means with changes in the mode control signal for identifying the setting of the gain control means.

13. In a data reduction system as defined in claim 12 wherein said digital encoding means has a plurality of operating modes for implementing a plurality of different truncated variable word length codes, and

means responsive to the output from the mode control means for selecting the operating mode of the digital encoder means for operation with a code which provides a reduction in the bit rate of the encoded compressed signal output therefrom.

14. In a data reduction system as defined in claim 12 wherein said digital encoding means implements a truncated Huffman code.

15. In a data reduction system as defined in claim 12 wherein said gain setting means comprises a variable gain amplifier for reducing the amplitude of the analog input signal to the analog to digital converter means with input signal energy level changes from a lower to a higher energy level band.

16. In a data reduction system as defined in claim 12 including,

code checker and stripper means for stripping the identifying code word from the encoded compressed signal stream from the digital encoder means and for producing a code identification signal corresponding to said identifying code word,

digital decoder means responsive to the encoded compressed signal stream from said code checker and stripper means for decoding said encoded compressed signal stream,

digital reconstruction filter means responsive to decoded signals from said digital decoder means for reconstruction filtering thereof,

controllable gain setting means for controlling the amplitude of the output from the digital reconstruction filter means, and having a gain which is controlled by the code identification signal from said code checker and stripper means for increasing the amplitude of the digital reconstruction filter output with input signal energy level changes from a lower to a higher energy level band.

17. In a data reduction system as defined in claim 16 wherein said digital encoding means implements a truncated Huffman code.

18. In a data reduction system as defined in claim 16 including,

digital to analog converter means for converting the output from the digital reconstruction filter means to analog form, and

wherein said gain setting means for controlling the entropy of signals supplied to said digital compression filter means, and said controllable gain setting means for controlling the amplitude of the output from the digital reconstruction filter means comprise variable gain amplifier means for reducing the amplitude of the analog input signal to the analog to digital converter means by a factor of 2^-x and for increasing the amplitude of the analog signal from the digital to analog converter means by a factor of 1/2^-x, respectively, with analog input signal energy level changes from one energy level band to a higher energy level band.

19. In a data reduction system as defined in claim 1 including,

high frequency deemphasis filter means for reducing the amplitude of high frequency components of digital sample signals supplied to said digital compression filter means.

20. Digital decoding and decompression means for producing digital signals f_n (out) from a stream of encoded digital compressed signals, different portions of said stream having been encoded using different codes, said stream including an identifying code word at the start of each portion of the stream encoded using a different code, said digital decoder and decompression means comprising,

code checker and stripper means for stripping the identifying code word from the stream of encoded compressed signals and for producing a code identification signal corresponding to said identifying code word,

digital decoder means responsive to the encoded compressed signal stream from said code checker and stripper means and having a plurality of different operating modes for implementing different algorithms for decoding the different codes,

means responsive to the code identification signal from said code checker and stripper means for selecting the operating mode of the digital decoder means required for decoding the encoded compressed signals from said code checker and stripper means, and

means for reconstruction filtering of decoded signals from said digital decoder means for producing digital signals f_n (out).

21. In a data compression method for preparing analog input signals having a substantially Gaussian distribution for storage or transmission, which method includes converting the analog input signal to a digital sample signal stream of equal word length sample signals, digital compression filtering said sample signal stream for generating a stream of compressed signals of equal word length, and digital encoding the compressed signal stream to generate a stream of variable word length encoded compressed signals, the improvement including,

reducing the ratio of σ

/q by a factor of 2^-x whenever the average energy level of the analog input signal exceeds a predetermined threshold level for entropy reduction of the sample signal stream, wherein σ

is standard deviation of the compressed signal stream, q is quantization level of the sample signal stream, and x is a non-zero positive integer.

22. In a data compression method as defined in claim 21 wherein the step of reducing the ratio of σ

/q comprises restricting at least one least significant bit of the sample signal to a selected value whenever the average energy level of the analog input signal exceeds a predetermined threshold level.

23. In a data compression method as defined in claim 22 which includes restricting increased numbers of least significant bits of the sample signals to selected values as the average energy level of the input signals exceeds higher predetermined threshold levels.

24. In a data compression method as defined in claim 22 which includes implementing a different code with changes in the input signal average energy levels above and below the threshold level.

25. In a data compression method as defined in claim 24 which includes inserting an identifying code word in the encoded compressed signal stream with each change in the average energy level of input signal above and below the threshold level to identify the code implemented for use in subsequent decoding of the encoded compression signals.

26. In a data compression method as defined in claim 22 including,

obtaining a measure of the average energy level of the input signal by envelope detection thereof before compression filtering.

27. In a data compression method as defined in claim 21 wherein the step of reducing the ratio of σ

/q comprises reducing the amplitude of the analog input signal by a factor of 2^-x.

28. In a data compression method as defined in claim 21 which includes obtaining a measure of the average energy level of the analog input signal before digital compression filtering said sample signal stream for control of the reducing step.

29. In a data compression method as defined in claim 21 which includes obtaining a measure of the average energy level of the analog input signal following digital compression filtering before digital encoding of the comprised signal stream for control of the reducing step.

30. In a data compression method as defined in claim 21 wherein the step of digital encoding implements a truncated variable word length code.

31. In a data compression method as defined by claim 30 wherein the code comprises a truncated Huffman code.

32. In a data compression method for preparing a digital sample signal stream of equal word length sample signals including digital sample signal streams having a substantially Gaussian distribution for storage or transmission, which method includes digital compression filtering said sample signal stream for generating a stream of compressed signals, and digital encoding the compressed signal stream to generate a stream of variable word length encoded compressed signals, the improvement including,

restricting one or more least significant bits of the sample signals before compression filtering thereof to selected values whenever the average energy level of the signal stream before digital encoding thereof exceeds a predetermined threshold level for entropy reduction thereof.

33. In a data compression method as defined in claim 32 which includes restricting increased numbers of least significant bits of the sample signals to selected as the average energy level of the signal stream before digital encoding exceeds higher predetermined threshold levels.

34. In a data compression method as defined in claim 32 which includes implementing a different truncated variable word length code with changes in the compressed signal stream average energy levels above and below threshold level.

35. In a data compression method as defined in claim 34 which includes inserting an identifying code word in the encoded compressed signal stream with each change in the average energy level of the compressed signal stream above and below the threshold level to identify the code implemented for use in subsequent decoding of the encoded compression signals.

36. In a data compression method as defined in claim 34 wherein the code comprises a truncated Huffman code.

37. In a data compression method as defined in claim 32 which includes obtaining a measure of the average energy level of the signal stream before digital compression filtering thereof for controlling the restricting step.

38. In a data compression method as defined in claim 32 which includes obtaining a measure of the average energy level of the signal stream after digital compression filtering thereof for controlling the restricting step.

39. In a data compression method as defined in claim 32 wherein the restricting step comprises truncation of sample signals before compression filtering thereof.

40. In a data compression method for preparing analog input signals including those having a substantially Gaussian distribution for storage or transmission, which method includes converting the analog input signal to a digital sample signal stream, digital compression filtering said sample signal stream for generating a stream of compressed signals, and digital encoding the compressed signal stream using a truncated variable word length code to generate a stream of variable word length encoded compressed signals, the improvement comprising,

changing the amplitude of the analog input signal by a factor of 2^x, where x is a non-zero integer whenever the average energy of the analog input signal crosses a predetermined threshold level for changing the entropy thereof, and

inserting an identifying code word in the encoded compressed signal stream with each change in the average energy of the analog input signal above and below the threshold level to identify the magnitude of the amplitude reduction.

41. In a data compression method as defined in claim 40 wherein the step of changing the amplitude of the digital sample signal stream includes controlling the amplitude of the analog input signal by means of a variable gain amplifier.

42. In a data compression method as defined in claim 40 including inserting said identifying code word in the encoded signal stream periodically after each change in the average energy of the analog input signal above and below the threshold level.

43. In a data reduction system of the type for preparing a digital sample signal stream of equal word length sample signals for storage or transmission,

digital compression filter means for compression filtering the digital sample signal stream and generating a compressed signal stream of equal word length compressed signals,

digital encoding means for encoding the compressed signal stream by use of a variable word length code, and

means for controlling the quantization level of the digital sample signals and entropy of the sample signal stream supplied to the digital compression filter means by restricting one or more least significant bits of the digital sample signals to selected values in response to a measure of the average energy level of the digital sample signal stream above a predetermined level.

44. In a data reduction system as defined in claim 43 wherein said digital encoding means has a plurality of different operating modes for implementing a plurality of different codes, and

means for implementing a different code with a change in the quantization level of the digital sample signals.