Voice Recognition Method and System Adapted to the Characteristics of Non-Native Speakers

• US 20070294082A1
• Filed: 07/22/2004
• Published: 12/20/2007
• Est. Priority Date: 07/22/2004
• Status: Abandoned Application

• Current Assignee: Orange S.A.
• Original Assignee: Orange S.A.
• Inventors: Jouvet, Denis, Bartkova, Katarina
• Application Number:
  US11/658,010
• Publication Number:
  US 20070294082A1
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G10L 15/005 Language recognition
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems

Abstract

The invention relates to a voice signal recognition method comprising a step of producing an iterative learning procedure of acoustic models representing a standard set of models of voice units pronounced in a given target language and a step of using the acoustic models to recognize the voice signal by comparing said signal with the acoustic models previously obtained. The method consists in further producing an additional set of voice units in the target language adapted to the characteristics of a foreign language during the production of the acoustic models.

Claims

1. A method of recognizing a voice signal, comprising a step of generation by an iterative learning procedure of acoustic models representing a standard set of models of vocal units uttered in a given target language and a step of using the acoustic models to recognize the voice signal by comparison of that signal with the acoustic models obtained beforehand, characterized in that during the generation of the acoustic models, there is further generated an additional set of models of vocal units in the target language adapted to the characteristics of a foreign language.

2. A method according to claim 1, characterized in that the additional set of models of vocal units of the target language adapted to the characteristics of a foreign language is estimated from pronunciations of words or phrases in a foreign language.

3. A method according to claim 2, characterized in that a voice signal uttered in a target language and comprising phonemes pronounced in accordance with characteristics of a foreign language is recognized by comparing each utterance with the vocal unit models of the additional set (BR2) and with the vocal unit models of the standard set (BR1).

4. A method according to claim 2, characterized in that a voice signal uttered in a target language and comprising phonemes pronounced in accordance with characteristics of a foreign language is recognized by comparing the signal to be recognized with a combination of vocal unit models of the standard set and vocal unit models of the additional set.

5. A method according to claim 3, characterized in that the acoustic models further comprise a set of models of vocal units uttered in a foreign language.

6. A method according to claim 5, characterized in that a voice signal uttered in a target language and modified in accordance with characteristics of a foreign language is recognized by comparing it with a combination of models of vocal units further comprising a set of models of vocal units uttered in a foreign language.

7. A method according to claim 1, characterized in that a voice signal is compared either to models of vocal units or to combinations of models of vocal units belonging to the standard set or to another set of models of vocal units in a foreign language.

9. A method according to claim 4, characterized in that the acoustic models further comprise a set of models of vocal units uttered in a foreign language.

10. A method according to claim 9, characterized in that a voice signal uttered in a target language and modified in accordance with characteristics of a foreign language is recognized by comparing it with a combination of models of vocal units further comprising a set of models of vocal units uttered in a foreign language.

8. A voice recognition system, comprising means for analyzing voice signals (SV_MA) by using an iterative learning procedure to generate acoustic models representing a standard set of models of vocal units (UV_MA) uttered in a given target language, and means for comparing a voice signal to be recognized with the acoustic models of vocal units obtained beforehand, characterized in that the acoustic models further comprise an additional set of models of vocal units in the target language adapted to the characteristics of a foreign language.