Compression and expansion of digitized voice signals

• US 4,631,746 A
• Filed: 02/14/1983
• Issued: 12/23/1986
• Est. Priority Date: 02/14/1983
• Status: Expired due to Term

• Current Assignee: Intel Corporation
• Original Assignee: Wang Laboratories Incorporated
• Inventors: Daly, Daniel F., Bergeron, Lawrence E., Grosso, Elaine C.
• Primary Examiner(s): Kemeny, E. S. Matt
• Application Number:
  US06/466,250
• Time in Patent Office: 1,408 Days
• Field of Search: 381/21-40, 375/30, 375/31, 375/122
• US Class Current: 704/217
• CPC Class Codes: G10L 21/04 Time compression or expansion
• CPC Class Codes: H04B 1/66 for reducing bandwidth of signals; for improving efficiency of transmission H04B1/68 takes precedence

Abstract

Speech signal data compression is improved by a first compression using Time Domain Harmonic Scaling (TDHS), wherein two periods of voiced data are averaged together, followed by a second compression using Continuously Variable Slope Delta Modulation (CVSD). Pitch period detection and sampling-rate conversion are also features of the invention.

Claims

1. A method of compressing a digital input signal representative of a continuously-varying input signal into a digital pitch value and a plurality of associated signal values, said digital input signal including a plurality of binary words, each reprsentative of the amplitude of said continuously-varying input signal sampled at successive intervals of time, including the steps of:

processing a selected plurality of the binary words of said digital input signal to determine the pitch period of said continuously-varying input signal and providing a digital pitch value representative thereof;

performing a time domain harmonic compression on a first plurality of said binary words in response to said pitch value to provide a TDHC output signal representative of the digital input signal but compressed in time by a first predetermined factor, so that said TDHC output signal has a second plurality of binary words equal to said first plurality divided by said first factor;

interpolating, in response to the TDHC output signal, a predetermined number of binary words between each pair of binary words in the TDHC output signal to provide an expanded signal representative of the TDHC output signal but having a plurality of binary words equal to said second plurality multiplied by a second factor;

encoding said expanded signal by means of a delta modulator to provide an encoded output signal wherein each word in the expanded signal is converted into one bit in the encoded output signal representative of the slope of the expanded signal, the encoded output signal providing said signal values.

2. The method of claim 1 wherein the step of encoding includes the step of performing a continuously-variable slope delta modulation to provide said encoded output signal.

4. The method of claim 1 wherein said first factor is two and said second factor is four.

5. The method of claim 4 wherein the rate at which the continuously-varying input signal is sampled is approximately 8 kilohertz.

6. The method of claim 1 wherein the step of processing includes the step of autocorrelating the digital input signal with itself to determine said pitch period.

7. The method of claim 1 further including the step of transmitting as an associated set of values said pitch value and the encoded output signal.

8. The method of claim 1 further including the step of storing as an associated set of values said pitch value and the encoded output signal.

9. The method of claim 1 wherein the pitch value and associated plurality of signal values are uncompressed into a substantial recreation of the digital input signal, including the steps of:

decoding the bits of the signal values by means of a delta demodulator to provide a demodulated output signal including a plurality of binary words;

deleting selected binary words from the demodulated output signal to provide a reduced output signal representative of the demodulator output signal but having a plurality of binary words equal to the number of binary words in the demodulated output signal divided by said second factor;

performing a time domain harmonic expansion on the binary words in said reduced output signal as a function of the associated pitch value to provide a TDHE output signal expanded by said first factor and including a plurality of binary words equal in number to the number of words in said demodulated output signal multiplied by said first factor;

whereby the plurality of binary words in the expanded output signal is a substantial recreation of the binary input signal.

10. The method of claim 9 wherein the step of decoding includes the step of performing a continuously-variable slope delta demodulation to provide said encoding output signal.

11. The method of claim 9 wherein the step of deleting includes the step of upsampling the TDHC output signal to delete said selected binary words between each pair of binary words in the TDHC output signal.

12. The method of claim 9 wherein said first factor is two and said second factor is four.

13. The method of claim 12 wherein the rate at which the continuously-varying input signal is sampled is approxmately 8 kilohertz.

14. The method of claim 9 wherein the step of processing includes the step of autocorrelating the digital input signal with itself to determine said pitch period.

15. The method of claim 9 further including, between said steps of encoding and decoding, the step of transmitting as an associated set of values said pitch value and the encoded output signal.

16. The method of claim 9 further including, between said steps of encoding and decoding, the step of storing as an associated set of values said pitch value and the encoded output signal.

34. The method of claim 1 wherein the pitch value and associated plurality of signal values are uncompressed into a recreation of the digital input signal, said recreation having the same pitch as said input signal but selectively speeded up in time in response to a speed control signal, including the steps of:

decoding the bits of the signal values by means of a delta demodulator to provide a demodulated output signal including a plurality of binary words;

deleting selected binary words from the demodulated output signal to provide a reduced output signal representative of the demodulator output signal but having a plurality of binary words equal to the number of binary words in the demodulated output signal divided by said second factor;

selectively performing or not performing, in response to the speed control signal, a time domain harmonic expansion on the binary words in said reduced output signal as a function of the associated pitch value to provide, in response to a first value of the speed control signal, a TDHE output signal expanded by said first factor and including a plurality of binary words equal in number to the number of words in said reduced output signal multiplied by said first factor, and wherein the step of time domain harmonic expansion is bypassed in response to a second value of the speed control signal.

3. The method of claim wherein the step of interpolating includes the step of upsampling the TDHC output signal to provide said predetermined number of binary words between each pair of binary words in the TDHC output signal.

17. Apparatus for compressing a digital input signal, representative of a continuously-varying input signal, into a digital pitch value and a plurality of associated signal values, said digital input signal including a plurality of binary words, each representative of the amplitude of said continuously-varying input signal sampled at successive intervals of time, comprising:

means for processing a selected plurality of the binary words of said digital input signal to determine the pitch period of said continuously-varying input signal and for providing a digital pitch value representative thereof;

means for performing a time domain harmonic compression on a first plurality of said binary words in response to said pitch value to provide a TDHC output signal representative of the digital input signal but compressed in time by a first predetermined factor, so that said TDHC output signal has a second plurality of binary words equal to said first plurality divided by said first factor;

means for interpolating, in response to the TDHC output signal, a predetermined number of binary words between each pair of binary words in the TDHC output signal to provide an expanded signal respresentative of the TDHC output signal but having a plurality of binary words equal to said second plurality multiplied by a second factor;

delta modulator means for encoding said expanded signal to provide an encoded output signal wherein each word in the expanded signal is converted into one bit in the encoded output signal representative of the slope of the expanded signal, the encoded output signal providing said signal values.

18. The apparatus of claim 17 wherein the means for encoding comprises means for performing a continuously-variable slope delta modulation to provide said encoded output signal.

19. The apparatus of claim 17 wherein the means for interpolating comprises means for upsampling the TDHC output signal to provide said predetermined number of binary words between each pair of binary words in the TDHC output signal.

20. The apparatus of claim 17 wherein said first factor is two and said second factor is four.

21. The apparatus of claim 20 wherein the rate at which the continuously-varying input signal is sampled is approximately 8 kilohertz.

22. The apparatus of claim 17 wherein the means for processing comprises means for autocorrelating the digital input signal with itself to determine said pitch period.

23. The apparatus of claim 17 further comprising means for transmitting as an associated set of values said pitch value and the encoded output signal.

24. The apparatus of claim 17 further comprising means for storing as an associated set of values said pitch value and the encoded output signal.

25. The apparatus of claim 17 further including apparatus for uncompressing the pitch value and associated plurality of signal values into a substantial recreation of the digital input signal, and further comprising:

delta demodulator means for decoding the bits of the signal values to provide a demodulated output signal including a plurality of binary words;

means for deleting selected binary words from the demodulated output signal to provide a reduced output signal representative of the demodulator output signal but having a plurality of binary words equal to the number of binary words in the demodulated output signal divided by said second factor; and

means for performing a time domain harmonic expansion on the binary words in said reduced output signal as a function of the associated pitch value to provide a TDHE output signal expanded by said first factor and inlcuding a plurality of binary words equal in number to the number of words in said demodulated output signal multiplied by said first factor;

whereby the plurality of binary words in the expanded output signal is a substantial recreation of the digital input signal.

26. The apparatus of claim 25 wherein the means for decoding comprises means for performing a continuously-variable slope delta demodulation to provide said encoded output signal.

27. The apparatus of claim 25 wherein the means for deleting comprises means for upsampling the TDHC output signal to delete said selected binary words between each pair of binary words in the TDHC output signal.

28. The apparatus of claim 25 wherein said first factor is two and said second factor is four.

29. The apparatus of claim 28 wherein the rate at which the continuosuly-varying input signal is sampled is approximately 8 kilohertz.

30. The apparatus of claim 25 wherein the means for processing comprises means for autocorrelating the digital input signal with itself to determine said pitch period.

31. The apparatus of claim 25 further comprising means for transmitting as an associated set of values said pitch value and the encoded output signal.

32. The apparatus of claim 25 further comprising means for storing as an associated set of values said pitch value and the encoded output signal.

33. The apparatus of claim 17 further including apparatus for uncompressing the pitch value and associated plurality of signal values into a recreation of the digital input signal, said recreation having the same pitch as the continuously-varying input signal but selectively speeded up in time in response to a speed control signal, comprising:

delta demodulator means for decoding the bits of the signal values to provide a demodulated output signal including a plurality of binary words;

means for deleting selected binary words from the demodulated output signal to provide a reduced output signal representative of the demodulator output signal but having a plurality of binary words equal to the number of binary words in the demodulated output signal divided by said second factor;

means for performing a time domain harmonic expansion on the binary words in said reduced output signal as a function of the associated pitch value to provide a TDHE output signal expanded by said first factor and including a plurality of binary words equal in number to the number fo words in said reduced output signal multiplied by said first factor; and

means, responsive to the value of the speed control signal for selecting the TDHE output signal in response to a first value of the speed control signal and for selecting the reduced output signal in response to a second value of the speed control signal, the selected signal providing said recreation of the digital input signal;

whereby the plurality of binary words in said recreation is a substantial recreation of pitch of the continuously-varying input signal but speeded up in time.