Speech recognition method and system in which said method is implemented

• US 6,138,094 A
• Filed: 01/27/1998
• Issued: 10/24/2000
• Est. Priority Date: 02/03/1997
• Status: Expired due to Fees

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Miet, Gilles, Guilhaumon, Benoit
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): Azad, Abul K.
• Application Number:
  US09/014,172
• Time in Patent Office: 1,001 Days
• Field of Search: 704/231, 704/233, 704/238, 704/241, 704/251, 704/253, 704/254, 704/255, 704/226
• US Class Current: 704/233
• CPC Class Codes: G10L 15/10 using distance or distortion measures between unknown speech and reference templates
• 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

Abstract

In a speech recognition method and system a spoken word to be recognized is broken down into input vectors (K2a), ambient noise is evaluated (K1), a recognized word is chosen from a dictionary having associated reference vectors which are separated from the input vectors by a shortest distance (K3), and the recognized word is validated by a comparison of this distance with a threshold value which is derived as a function of the result of the evaluation of ambient noise. The ambient noise evaluation may be carried out at an instant when the speaker is silent, i.e. either before or after the speaker speaks the word to be recognized.

Claims

1. A method of recognizing spoken words, utilizing a dictionary containing words and associated reference vectors, the method comprising:

breaking down a spoken word to be recognized into input vectors,

evaluating ambient noise measured when the speaker is not speaking,

choosing a recognized word from the dictionary having associated reference vectors which are separated from the input vectors by a smallest distance, and

validating the recognized word by a comparison of said smallest distance with a threshold value,

wherein the threshold value is derived as a function of the result of the evaluation of ambient noise.

2. The method as claimed in claim 1, wherein the evaluation of ambient noise is carried out before the speaker speaks the word to be recognized.

3. The method as claimed in claim 1, wherein the evaluation of ambient noise is carried out after the speaker has spoken the word to be recognized.

4. The method as claimed in claim 1, wherein an invitation-to-speak signal perceptible to the speaker is issued and the evaluation of ambient noise is carried out before this signal is issued.

5. A method as claimed in claim 1, wherein the threshold increases substantially linearly with an increase in the ambient noise level measured in dB.

6. A speech recognition device comprising a user interface including a microphone for capturing words spoken by a user to be recognized, and a microprocessor assembly programmed for carrying out the following acts utilizing a stored dictionary containing words and associated reference vectors:

breaking down a spoken word to be recognized into input vectors,

evaluating ambient noise measured when the user is not speaking,

choosing a recognized word from the dictionary having associated reference vectors which are separated from the input vectors by a smallest distance, and

validating the recognized word by a comparison of said smallest distance with a threshold value,

wherein the threshold value is derived as a function of the result of the evaluation of ambient noise.

7. The device as claimed in claim 6, wherein the microprocessor assembly is programmed such that the evaluation of ambient noise is carried out once speech recognition is activated by the user via the user interface.

8. A device as claimed in claim 6, wherein the microprocessor assembly is programmed such that an invitation-to-speak signal perceptible to the user is issued via the user interface and the evaluation of ambient noise is carried out before this signal is issued.

9. A device as claimed in claim 6, wherein said device is a cellular telephone station.