Method and apparatus for recognizing tone languages using pitch information

• US 6,510,410 B1
• Filed: 07/28/2000
• Issued: 01/21/2003
• Est. Priority Date: 07/28/2000
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Chen, Julian Chengjun, Fu, Guo Kang, Shen, Li Qin, Li, Hai Ping
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): NOLAN, DANIEL A
• Application Number:
  US09/627,595
• Time in Patent Office: 907 Days
• Field of Search: 704/251, 704/254, 704/243, 704/207, 704/255, 704/236
• US Class Current: 704/251
• CPC Class Codes: G10L 15/02 Feature extraction for speech recognition; Selection of recognition unit
• CPC Class Codes: G10L 25/15 the extracted parameters being formant information
• CPC Class Codes: G10L 25/90 Pitch determination of speech signals

Abstract

A method and an apparatus for automatic recognition of tone languages, employing the steps of converting the words of speech into an electrical signal, generating spectral features from the electrical signal, extracting pitch values from the electrical signal, combining said spectral features and the pitch values into acoustic feature vectors, comparing the acoustic feature vectors with prototypes of phonemes in an acoustic prototype database including prototypes of toned vowels to produce labels, and matching the labels to text using a decoder comprising a phonetic vocabulary and a language model database.

Claims

1. A method for identifying toned vowels in words of speech comprising:

converting the words of speech into an electrical signal;

generating spectral features from said electrical signal;

extracting pitch values from said electrical signal;

combining said spectral features and said pitch values into acoustic feature vectors;

comparing said acoustic feature vectors with prototypes of phonemes in an acoustic prototype database including prototypes of toned vowels to produce labels; and

matching said labels to text using a decoder comprising a phonetic vocabulary and a language model database.

2. The method of claim 1 further comprising the step of constructing the acoustic prototypes, wherein the step of constructing comprises the steps of:

preparing a training text from said words of speech;

transcribing said training text into sequences of phonemes including vowels with tones;

converting spoken utterances of said training text into an electrical signal;

generating spectral features from said electrical signal;

extracting pitch values from said electrical signal;

combining said spectral features and said pitch values into acoustic feature vectors;

comparing said acoustic feature vectors with said sequences of phonemes including vowels with tone to produce acoustic prototypes for each phoneme.

3. The method of claim 2, wherein said acoustic prototypes are stored in a database.

4. The method of claim 1, wherein said phonetic vocabulary comprises a database of words of speech including tone information.

5. The method of claim 1, wherein said language model database is used to determine a probability of a word.

6. The method of claim 1, wherein said words of speech comprise at least one syllable having tonal content.

7. The method of claim 6, wherein said toned vowel determines a tone of said syllable.

8. A program storage device readable by machine, tangibly embodying a program of instructions executable by machine to perform the method steps for identifying toned vowels in words of speech, the method comprising the steps of:

converting the words of speech into an electrical signal;

generating spectral features from said electrical signal;

extracting pitch values from said electrical signal;

combining said spectral features and said pitch values into acoustic feature vectors;

comparing said acoustic feature vectors with prototypes of phonemes in an acoustic prototype database including prototypes of toned vowels to produce labels; and

matching said labels to text using a decoder comprising a phonetic vocabulary and a language model database.

9. The program storage device of claim 8, further comprising instructions for performing the step of constructing the acoustic prototypes, wherein the instructions for constructing the acoustic prototypes comprise instructions for performing the steps of:

receiving as input a training text from said words of speech;

transcribing said training text into sequences of phonemes including vowels with tones;

converting spoken utterances of said training text into an electrical signal;

generating spectral features from said electrical signal;

extracting pitch values from said electrical signal;

combining said spectral features and said pitch values into acoustic feature vectors;

comparing said acoustic feature vectors with said sequences of phonemes including vowels with tone to produce acoustic prototypes for each phoneme.

10. The program storage device of claim 9, wherein said acoustic prototypes are stored in a database.

11. The program storage device of claim 8, wherein said phonetic vocabulary comprises a database of words of speech including tone information.

12. The program storage device of claim 8, wherein said language model database is used to determine a probability of a word.

13. The program storage device of claim 8, wherein said words of speech comprise at least one syllable having tonal content.

14. The program storage device of claim 8, wherein said toned vowel determines a tone of said syllable.

15. A system for identifying toned vowels in words of speech, comprising:

means for converting the words of speech into an electrical signal;

means for generating spectral features from said electrical signal;

means for extracting pitch values from said electrical signal;

means for combining said spectral features and said pitch values into acoustic feature vectors;

means for comparing said acoustic feature vectors with prototypes of phonemes in an acoustic prototype database including prototypes of toned vowels to produce labels; and

means for matching said labels to text using a decoder comprising a phoneic vocabulary and a language model database.

16. The system of claim 15, wherein said phonetic vocabulary comprises a database of word of speech including tone information.

17. The system of claim 15, wherein said language model database is used to determine a probability of a word.

18. The system of claim 15, wherein said words of speech comprise at least one syllable having tonal content.

19. The system of claim 18, wherein said toned vowel determines a tone of said syllable.