Speech processing coder, decoder and command recognizer

• US 5,909,662 A
• Filed: 03/11/1997
• Issued: 06/01/1999
• Est. Priority Date: 08/11/1995
• Status: Expired due to Fees

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Taniguchi, Tomohiko, Yamazaki, Yasushi, Sato, Tomonori, Kawai, Chiharu, Matsuzawa, Hitoshi
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Storm, Donald L.
• Application Number:
  US08/793,838
• Time in Patent Office: 812 Days
• Field of Search: 704/222, 704/200, 704/221, 704/230, 704/231, 704/254, 704/255, 704/266
• US Class Current: 704/221
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• 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 19/04 using predictive techniques
• CPC Class Codes: H04M 2201/40 using speech recognition

Abstract

The present invention relates to a speech processing device equipped with both a speech coding/decoding function and a speech recognition function, and is aimed at providing a speech processing device equipped with both a speech coding/decoding function and a speech recognition function by using a small amount of memory. The speech processing device of the present invention includes a speech analysis unit for obtaining analysis results by analyzing input speech, a codebook for storing quantization parameters and quantization codes indicating the quantization parameters, a quantizing unit for selecting the quantization parameters and the quantization codes corresponding to the analysis results from the codebook and for outputting selected quantization parameters and selected quantization codes, a coding unit for outputting encoded codes of the input speech including the selected quantization codes, a speech dictionary for storing registered data which represent speech patterns by using the codebook, and a matching unit for obtaining the speech patterns corresponding to the input speech by using one of the analysis results, the selected quantization parameters, and the selected quantization codes.

Claims

1. A speech processing device comprising:

a speech analysis unit for obtaining analysis results by analyzing input speech;

a codebook for storing quantization parameters and quantization codes indicating said quantization parameters;

a quantizing unit for selecting said quantization parameters and said quantization codes corresponding to said analysis results from said codebook and for outputting selected quantization parameters and selected quantization codes;

a coding unit for outputting encoded codes of said input speech without involving recognition of the input speech in a coding mode during which the input speech is transmitted to an exterior of the device, said encoded codes including said selected quantization codes;

a speech dictionary for storing registered data which represent speech patterns by using said codebook; and

a matching unit for obtaining said speech patterns corresponding to said input speech in a recognition mode by using one of said analysis results, said selected quantization parameters, and said selected quantization codes,

wherein said coding unit operates in the coding mode but not in the recognition mode while said matching unit operates in the recognition mode but not in the coding mode.

2. The speech processing device as claimed in claim 1, wherein said registered data of said speech dictionary comprises series of said quantization codes, said series of said quantization codes indicating series of said quantization parameters in said codebook to represent said speech patterns.

3. The speech processing device as claimed in claim 2, further comprising a parameter conversion unit for converting said selected quantization parameters of said input speech into input-converted parameters and for converting said speech patterns of said speech dictionary into dictionary-converted parameters, wherein said matching unit matches said input-converted parameters with said dictionary-converted parameters to obtain said speech patterns corresponding to said speech input.

4. The speech processing device as claimed in claim 3, wherein said quantization parameters comprise LPC coefficients, and said input-converted parameters and said dictionary-converted parameters comprise LPC cepstrum coefficients.

5. The speech processing device as claimed in claim 1, further comprising a reference-signal selecting unit for selecting one of said analysis results, said selected quantization parameters, and said selected quantization codes to supply said selected one to said matching unit.

6. The speech processing device as claimed in claim 1, further comprising a speech-dictionary generating unit for generating said speech dictionary by using quantization parameters obtained from analysis results of input speech.

7. The speech processing device as claimed in claim 1, further comprising an analysis-condition changing unit for changing analysis conditions between a speech coding operation for coding said input speech and a speech recognition operation for recognizing said input speech, said analysis conditions being used by said speech analysis unit for obtaining said analysis results.

8. A speech processing device comprising:

a codebook for storing quantization parameters and quantization codes indicating said quantization parameters;

an inverse-quantizing unit for selecting said quantization parameters corresponding to received quantization codes from said codebook and for outputting selected quantization parameters;

a speech synthesizing unit for synthesizing speech by using said selected quantization parameters;

a speech dictionary which is a data storage for storing registered data which represents speech patterns by using said codebook, said registered data including series of said quantization codes which in turn indicate a series of said quantization parameters in said codebook, thereby representing said speech patterns; and

a matching unit for obtaining said speech patterns corresponding to said received quantization codes by using one of said selected quantization parameters and said received quantization codes.

9. The speech processing device as claimed in claim 8, wherein said registered data of said speech dictionary comprises series of said quantization codes, said series of said quantization codes indicating series of said quantization parameters in said codebook to represent said speech patterns.

10. The speech processing device as claimed in claim 9, further comprising a parameter conversion unit for converting said selected quantization parameters into reception-converted parameters and for converting said speech patterns of said speech dictionary into dictionary-converted parameters, wherein said matching unit matches said reception-converted parameters with said dictionary-converted parameters to obtain said speech patterns corresponding to said received quantization codes.

11. The speech processing device as claimed in claim 10, wherein said quantization parameters comprise LPC coefficients, and said reception-converted parameters and said dictionary-converted parameters comprise LPC cepstrum coefficients.

12. The speech processing device as claimed in claim 8, further comprising a reference-signal selecting unit for selecting one of said selected quantization parameters and said received quantization codes to supply said selected one to said matching unit.

13. The speech processing device as claimed in claim 8, further comprising a speech-dictionary generating unit for generating said speech dictionary by using quantization parameters obtained from received quantization codes.

14. The speech processing device as claimed in claim 8, further comprising a de coding unit for separating quantization codes of more than one type from received encoded codes.

15. An speech analysis unit for obtaining analysis results by analyzing input speech;

a codebook for storing quantization parameters and quantization codes indicating said quantization parameters;

a quantizing unit for selecting said quantization parameters and said quantization codes corresponding to said analysis results from said codebook and for outputting first selected quantization parameters and selected quantization codes;

a coding unit for outputting encoded codes of said input speech, said encoded codes including said selected quantization codes;

an inverse-quantizing unit for selecting said quantization parameters corresponding to received quantization codes from said codebook and for outputting second selected quantization parameters;

a speech synthesizing unit for synthesizing speech by using said second selected quantization parameters;

a speech dictionary for storing registered data which represent speech patterns by using said codebook; and

a matching unit for obtaining said speech patterns corresponding to said input speech by using one of said analysis results, said first selected quantization parameters, and said selected quantization codes, and for obtaining said speech patterns corresponding to said received quantization codes by using one of said second selected quantization parameters and said received quantization codes.

16. The speech processing device as claimed in claim 15, wherein said registered data of said speech dictionary comprises series of said quantization codes, said series of said quantization codes indicating series of said quantization parameters in said codebook to represent said speech patterns.

17. The speech processing device as claimed in claim 16, further comprising synthesizing control means for controlling said speech synthesizing unit to synthesize speech corresponding to a matched speech pattern, said matched speech pattern being said speech patterns corresponding to said input speech and obtained by said matching unit.

18. The speech processing device as claimed in claim 17, wherein said synthesizing control means comprises:

first means for supplying series of said quantization parameters representing said matched speech pattern and other necessary speech synthesizing parameters to said speech synthesizing unit; and

second means for generating at least some of said other necessary speech synthesizing parameters.

19. The speech processing device as claimed in claim 18, wherein said second means stores at least some of said other necessary speech synthesizing parameters in addition to generating at least some of said other necessary speech synthesizing parameters.

20. The speech processing device as claimed in claim 19, wherein among said at least some of said other necessary speech synthesizing parameters generated by said second means, at least some are predetermined constants.

21. The speech processing device as claimed in claim 18, wherein said series of said quantization parameters comprise series of LPC coefficients, and said other necessary speech synthesizing parameters comprise parameters for emulating a sound source.

22. The speech processing device as claimed in claim 15, further comprising a reference-signal selecting unit for selecting one of said analysis results, said first selected quantization parameters, said selected quantization codes, said second selected quantization parameters, and said received quantization codes.

23. The speech processing device as claimed in claim 15, further comprising a speech-dictionary generating unit for generating said speech dictionary by using one of quantization parameters obtained from analysis results of input speech and quantization parameters obtained from received-quantization codes.

24. The speech processing device as claimed in claim 23, further comprising a dictionary-signal selecting unit for selecting a dictionary-generation signal for generating said speech dictionary from quantization parameters obtained from analysis results of input speech and quantization parameters obtained from received-quantization codes.

25. The speech processing device as claimed in claim 15, further comprising a parameter conversion unit for converting said speech patterns of said speech dictionary into dictionary-converted parameters, for converting said first selected quantization parameters of said input speech into input -converted parameters, and for converting said second selected quantization parameters of said received quantization codes into reception-converted parameters, wherein said matching unit matches said input-converted parameters with said dictionary-converted parameters to obtain said speech patterns corresponding to said speech input, and matches said reception-converted parameters with said dictionary-converted parameters to obtain said speech patterns corresponding to said received quantization codes.

26. The speech processing device as claimed in claim 25, wherein said quantization parameters comprise LPC coefficients, and said input-converted parameters, said reception-converted parameters, and said dictionary-converted parameters comprise LPC cepstrum coefficients.

27. The speech processing device as claimed in claim 15, wherein coding/decoding is provided by a Code Exited Linear Prediction.

28. The speech processing device as claimed in claim 15, further comprising a speech checking unit for checking processing results of said quantizing unit and for controlling said matching unit not to carry out a process of said matching unit when said processing results are found to correspond to a noise portion of said codebook.

29. The speech processing device as claimed in claim 15, further comprising a high-speed RAM and a low-speed RAM, wherein said speech dictionary is stored in said low-speed RAM and said high-speed RAM is used as a work area during a coding/decoding operation, while said speech dictionary in said low-speed RAM is transferred to said high-speed RAM and said speech dictionary transferred to said high-speed RAM is used for speech recognition during a speech recognition operation.