Methods and apparatus for generating and using speaker independent garbage models for speaker dependent speech recognition purpose

• US 5,895,448 A
• Filed: 04/30/1997
• Issued: 04/20/1999
• Est. Priority Date: 02/29/1996
• Status: Expired due to Term

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Nynex Science And Technology, Inc.
• Inventors: Raman, Vijay R., Vysotsky, George J.
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Chawan, Vijay B.
• Application Number:
  US08/846,613
• Time in Patent Office: 720 Days
• Field of Search: 379/88, 704/251, 704/238, 704/243, 704/247, 704/240, 704/200, 704/245, 704/231, 704/256
• US Class Current: 704/270.1
• CPC Class Codes: G10L 15/065 Adaptation
• CPC Class Codes: G10L 15/142 Hidden Markov Models [HMMs]
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 15/34 Adaptation of a single recogniser for parallel processing, e.g. by use of multiple processors or cloud computing
• CPC Class Codes: G10L 2015/088 Word spotting
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: H04M 1/271 controlled by voice recognition
• CPC Class Codes: H04M 2201/40 using speech recognition speech recognition per se G10L15/00
• CPC Class Codes: H04M 3/42 Systems providing special services or facilities to subscribers specially adapted for wireless communication networks H04W4/00
• CPC Class Codes: H04M 3/42204 Arrangements at the exchange for service or number selection by voice at the terminal H04M1/27
• CPC Class Codes: H04M 3/44 Additional connecting arrangements for providing access to frequently-wanted subscribers, e.g. abbreviated dialling at the subscriber's set H04M1/27; automatic redialling H04M3/424

Abstract

Methods and apparatus for generating and using both speaker dependent and speaker independent garbage models in speaker dependent speech recognition applications are described. The present invention recognizes that in some speech recognition systems, e.g., systems where multiple speech recognition operations are performed on the same signal, it may be desirable to recognize and treat words or phrases in one part of the speech recognition system as garbage or out of vocabulary utterances with the understanding that the very same words or phrases will be recognized and treated as in-vocabulary by another portion of the system. In accordance with the present invention, in systems where both speaker independent and speaker dependent speech recognition operations are performed independently, e.g., in parallel, one or more speaker independent models of words or phrases which are to be recognized by the speaker independent speech recognizer are included as garbage (OOV) models in the speaker dependent speech recognizer. This reduces the risk of obtaining conflicting speech recognition results from the speaker independent and speaker dependent speech recognition circuits. The present invention also provides for the generation of speaker dependent garbage models from the very same data used to generate speaker dependent speech recognition models, e.g., word models. The technique involves processing the data included in the speaker dependent speech recognition models to create one or more speaker dependent garbage models.

Claims

1. A speech recognition method, comprising the steps of:

performing a speaker independent speech recognition operation on a signal using a first set of speaker independent speech recognition models corresponding to a first set of words;

performing a speaker dependent speech recognition operation on said signal using a set of speaker dependent speech recognition models corresponding to a second set of words and using, as an out of vocabulary word model, at least one speaker independent speech recognition model corresponding to a word included in the first set of words.

2. The method of claim 1, wherein the speaker independent and speaker dependent speech recognition operations are performed in parallel.

3. The method of claim 2, further comprising the step of:

generating, using Hidden Markov modeling techniques the first set of speaker independent speech recognition models and the speaker independent speech recognition model used as an out of vocabulary word model.

4. The method of claim 3, wherein the speaker independent speech recognition model used in the speaker dependent speech recognition operation has a structure corresponding to the structure of the speaker dependent speech recognition models.

5. The method of claim 3, wherein the first set of speaker independent speech recognition models includes information about more signal characteristics than the speaker independent speech recognition model used in the speaker dependent speech recognition operation.

6. The method of claim 1, wherein the speaker independent and speaker dependent speech recognition operations are performed simultaneously.

7. A speech recognition method, comprising the steps of:

performing a speaker independent speech recognition operation on a signal to identify words or phrases included therein using a first set of speaker independent speech recognition models corresponding to a first set of words or phrases;

performing a speaker dependent speech recognition operation on the signal to identify words or phrases included therein using as garbage models a second set of speaker independent speech recognition models corresponding to the words or phrases included in the first set of words or phrases and using a set of speaker dependent speech recognition models to identify words or phrases included in the signal corresponding to a second set of words or phrases, the speaker dependent speech recognition operation including the step of rejecting words or phases identified using the second set of speaker independent speech recognition models as being out of vocabulary words or phrases.

8. The method of claim 7, wherein the speaker independent and speaker dependent speech recognition operations are performed in parallel.

9. The method of claim 7, further comprising the step of:

generating the first and second sets of speaker independent speech recognition models, the generating step being performed prior to performing the speaker independent and speaker dependent speech recognition operations and involving the steps of;

i. using statistical modeling techniques to generate the first set of speaker independent speech recognition models to have a first resolution and structure;

ii. using statistical modeling techniques to generate the second set of speaker independent speech recognition models to have a second resolution and structure, the second resolution being lower than the first resolution.

10. The method of claim 9, wherein the first and second sets of speaker independent speech recognition models are generated using Hidden Markov Model training techniques.

11. The method of claim 9, wherein the second set of generated speaker independent speech recognition models have the same structure as the speaker dependent speech recognition models used in the speaker dependent speech recognition operation.

12. A speech recognition system, comprising:

speaker independent means for identifying a first set of words in a signal by performing a speaker independent speech recognition operation on the signal;

speaker dependent means for identifying a second set of words in the signal by performing a speaker dependent speech recognition operation on the signal, the speaker dependent means including speaker independent means for identifying and rejecting a word included in the first set of words.

13. The speech recognition system of claim 12,

wherein the speaker independent means includes a first set of speaker independent speech recognition models;

and

wherein the speaker dependent means includes a first set of speaker dependent speech recognition models and at least one speaker independent speech recognition model used for identifying and rejecting the word included in the first set of words.

14. The speech recognition system of claim 13,

wherein the first set of words includes explicit telephone network control commands;

wherein the second set of words includes names; and

wherein the word included in the first set of words which is identified and rejected by the speaker dependent means is a word included in one of the explicit telephone network control commands.

15. The speech recognition system of claim 13,

wherein the speaker independent means and speaker dependent means are arranged in parallel.

16. The speech recognition system of claim 15, further comprising an information database coupled to the speaker independent means and the speaker dependent means for storing the first set of speaker independent speech recognition models, the set of speaker dependent speech recognition model used as an out of vocabulary word model.