System and method for performing voice compression

• US 5,742,930 A
• Filed: 09/28/1995
• Issued: 04/21/1998
• Est. Priority Date: 12/16/1993
• Status: Expired due to Fees

• Current Assignee: Voice Compression Technologies Incorporated
• Original Assignee: Voice Compression Technologies Incorporated
• Inventors: Howitt, Andrew Wilson
• Primary Examiner(s): Hafiz, Tariq R.
• Application Number:
  US08/535,586
• Time in Patent Office: 936 Days
• Field of Search: 364/724.15, 381/42.51, 395/2, 395/2.1, 395/2.21, 395/2.28, 395/2.34-2.39, 395/2.79, 395/425, 704/500-504
• US Class Current: 704/502
• CPC Class Codes: G10L 19/00 Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis in musical instruments G10H
• CPC Class Codes: G10L 19/012 Comfort noise or silence coding
• CPC Class Codes: G10L 19/16 Vocoder architecture

Abstract

Voice compression is performed in multiple stages to increase the overall compression between the incoming analog voice signal and the resulting digitized voice signal over that which would be obtained if only a single stage of compression were to be used. A first type of compression is performed on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal, and a second, different type of compression is performed on the intermediate signal to produce an output signal that is compressed still further. As a result, compression better than 1920 bits per second (and approaching 960 bits per second) are obtained without sacrificing the intelligibility of the subsequently reconstructed analog voice signal. Voice compression is also performed by recognizing redundant portions of said voice signal, such as silence, and replacing such redundant portions with a special code in said compressed signal. Among other advantages, the higher total compression allows speech to be transmitted in far less time than would otherwise be possible, thereby reducing expense.

Claims

1. A method of voice compression comprising the steps of:

performing a first type of compression on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal in accordance with a speech compression procedure;

storing the intermediate signal;

performing a second type of compression different from the first type on said stored intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said first type of compression is of a kind that causes loss of a portion of the information contained in the intermediate signal with respect to the voice signal, and said second type of compression is of a kind that causes no loss of information contained in the output signal with respect to the intermediate signal.

2. A method of voice compression comprising the steps of:

performing a first type of compression on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal;

storing the intermediate signal;

performing a second type of compression different from the first type on said stored intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said output signal is compressed in time with respect to said voice signal.

3. A method of voice compression comprising the steps of:

performing a first type of compression on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal in accordance with a speech compression procedure;

performing a second type of compression different from the first type on said intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

storing said intermediate signal as a data file prior to performing said second type of compression.

5. A method of voice compression comprising the steps of:

performing a first type of compression on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal;

performing a second type of compression different from the first type on said intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said voice signal includes speech interspersed with silence, and said first type of compression produces said intermediate signal as a sequence of frames each of which corresponds in time to a portion of said voice signal and said voice signal includes data representative of said portion of said voice signal, and further comprising detecting at least one of said frames which corresponds to a portion of said voice signal that contains silence, replacing said at least one of said frames in said sequence with a binary code that indicates silence, and thereafter performing said second type of compression on said sequence.

6. The method of claim 5 wherein said frames have a selected minimum size, said code being smaller than said minimum size.

7. A method of voice compression comprising the steps of:

performing a first type of compression on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal;

performing a second type of compression different from the first type on said intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said first type of compression produces said intermediate signal as a sequence of frames each of which corresponds in time to a portion of said voice signal and contains data that represents a plurality of characteristics of said voice signal, said data for at least one of said characteristics being interleaved with said date for at least one other of said characteristics in said frame, and further comprising;

deinterleaving said delta so that said data for each one of said characteristics appears together in said frame, and

thereafter performing said second type of compression on said sequence.

4. The method of claim 7 further comprising storing said output signal as a data file.

8. The method of claim 7 wherein said one characteristic includes amplitude content and said other characteristic includes frequency content.

9. A method of voice compression comprising the steps of:

performing a first type of compression on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal;

performing a second type of compression different from the first type on said intermediate signal to produce an output signal that is compressed with respect to the intermediate signal;

add

wherein said first type of compression produces said intermediate signal as a sequence of frames each of which corresponds in time to a portion of said voice signal and contains data that represents information contained in said portion of said voice signal and data that does not represent said information, and further comprising;

removing said data that does not represent said information from each one of said frames, and

thereafter performing said second type of compression on said sequence.

10. A method of voice compression comprising the steps of:

performing a first type of compression on a voice signal to produce an intermediate signal that is compressed with respect to the voice signal;

performing a second type of compression different from the first type on said intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said first type of compression produces said intermediate signal as a sequence of frames each of which corresponds in time to a portion of said voice signal and includes a plurality of bits of data at least some of which represent information contained in said portion of said voice signal, each said frame being a non-interger number of bytes in length, and further comprising;

adding a selected number of bits to each said frame to increase the length thereof to an integer number of bytes, and

thereafter performing said second type of compression on said sequence.

11. A method of performing compression on a voice signal that includes redundant signal information, comprising the steps of:

performing compression on a voice signal to produce a first compressed signal;

detecting at least one portion of said compressed signal that corresponds to a portion of said voice signal that contains only said redundant signal information;

replacing said at least one portion of said first compressed signal with a binary code that indicates said redundant signal information.

12. The method of claim 11 wherein said compression produces said compressed signal as a sequence of frames each of which corresponds to a portion of said voice signal and includes data representative of said portion of said voice signal, and further comprising the steps of:

detecting at least one of said frames which corresponds to said portion of said voice signal that contains only said redundant signal information, and

replacing said at least one of said frames in said sequence with said binary code.

13. The method of claim 11 further comprising performing a second, different type of compression on said first compressed signal to produce a second compressed signal that is compressed with respect to said first compressed signal.

14. The method of claim 11 wherein said step of detecting includes determining that a magnitude of said first compressed signal that corresponds to a level of said voice signal is less than a threshold.

15. The method of claim 11 further comprising the steps of:

detecting said code in said first compressed signal, and replacing said code with a period of sound or silence represented by said redundant signal information of a selected length, and

thereafter performing decompression of said compressed signal to produce a second voice signal that is expanded with respect to said compressed signal and that is a recognizable reconstruction of the voice signal prior to compression.

16. The method of claim 11 wherein said redundant signal information represents silence.

17. Voice compression apparatus comprising:

a first compressor for performing a first type of compression on a voice signal to produce an intermediate signal that is a signal in accordance with a speech compression procedure;

a memory for storing the intermediate signal;

a second compressor for performing a second type of compression different from the first type on the stored intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said first compressor causes loss of a portion of the information contained in the intermediate signal with respect to the voice signal, and said second compressor causes no loss of information contained in the output signal with respect to the intermediate signal.

18. Voice compression apparatus comprising:

a first compressor for performing a first type of compression on a voice signal to produce an intermediate signal that is a signal in accordance with a speech compression procedure;

a second compressor for performing a second type of compression different from the first type on the intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

a memory for storing said intermediate signal as a data file.

19. The apparatus of claim 18 further comprising a memory for storing said output signal as a data file.

20. Voice compression apparatus comprising:

a first compressor for performing a first type of compression on a voice signal to produce an intermediate signal that is a signal;

a second compressor for performing a second type of compression different from the first type on the intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said voice signal includes speech interspersed with silence, and said first compressor produces said intermediate signal as a sequence of frames each of which corresponds in time to a portion said voice signal and includes data representative of said portion of said voice signal, and further comprising;

a detector for detecting at least one of said frames which corresponds to a portion of said voice signal that contains substantially only silence,

means for replacing said at least one of said frames in said sequence with a binary code that indicates silence, and

means for thereafter applying said sequence to said second compressor.

21. The apparatus of claim 20 wherein said frames have a selected minimum size, said code being smaller than said minimum size.

22. Voice compression apparatus comprising;

a first compressor for performing a first type of compression on a voice signal to produce an intermediate signal that is a signal;

a second compressor for performing a second type of compression on the intermediate signal different from the first type to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said first compressor produces said intermediate signal as a sequence of frames each of which corresponds to a portion of said voice signal and contains data that represents a plurality of characteristics of said voice signal, said data for at least one of said characteristics being interleaved with said data for at least one other of said characteristics in said frame, and further comprising;

means for deinterleaving said data so that said data for each one of said characteristics appears together in said frame, and

means for thereafter applying said sequence to said second compressor.

23. The apparatus of claim 22 wherein said one characteristic includes amplitude content and said other characteristic includes frequency content.

24. Voice compression apparatus comprising;

a first compressor for performing a first type of compression on a voice signal to produce an intermediate signal that is a signal;

a second compressor for performing a second type of compression different from the first type on the intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said first compressor produces said intermediate signal as a sequence of frames each of which corresponds to a portion of said voice signal and contains data that represents information contained in said portion of said voice signal and data that does not represent said information, and further comprising;

means for removing said data that does not represent said information from each one of said frames, and

means for thereafter applying said sequence to said second compressor.

25. Voice compression apparatus comprising:

a first compressor for performing a first type of compression on a voice signal to produce an intermediate signal that is a signal;

a second compressor for performing a second type of compression different from the first type on the intermediate signal to produce an output signal that is compressed with respect to the intermediate signal; and

wherein said first compressor produces said intermediate signal as a sequence of frames each of which corresponds to a portion of said voice signal and includes a plurality of bits of data at least some of which represent information contained in said portion of said voice signal, each said frame being a non-integer number of bytes in length, and further comprising;

circuitry for adding a selected number of bits to each said frame to increase the length thereof to an integer number of bytes, and

means for thereafter applying said sequence to said second compressor.

26. Apparatus for performing compression on a voice signal that includes speech interspersed with redundant signal information, comprising:

a compressor for performing compression on a voice signal to produce a first compressed signal that is compressed with respect to the voice signal,

a detector for detecting at least one portion of said first compressed signal that corresponds to a portion of said voice signal that contains substantially only said redundant signal information,

means for replacing said at least one portion of said first compressed signal with a binary code that indicates said redundant signal information.

27. The apparatus of claim 26 wherein said compressor produces said compressed signal as a sequence of frames each of which corresponds to a portion of said voice signal and includes data representative of said portion of said voice signal, said detector detecting at least one of said frames which corresponds to said portion of said voice signal that contains substantially only said redundant signal information, and said means for replacing substituting said at least one of said frames in said sequence with said binary code.

28. The apparatus of claim 26 further comprising a second compressor for performing a second, different type of compression on said first compressed signal to produce a second compressed signal that is compressed with respect to said first compressed signal.

29. The apparatus of claim 26 wherein said detector includes means for determining that a magnitude of said first compressed signal that corresponds to a level of said voice signal is less than a threshold.

30. The apparatus of claim 26 further comprising:

a second detector for detecting said binary code in said first compressed signal and replacing said code with a period of sound or silence represented by said redundant signal information of a selected length, and a decompressor for performing decompression of said first compressed signal to produce a second voice signal that is expanded with respect to said compressed signal and that is a recognizable reconstruction of the voice signal prior to compression.

31. The apparatus of claim 26 wherein said redundant signal information represents silence.