Method and system for speech recognition

• US 20030050779A1
• Filed: 08/31/2001
• Published: 03/13/2003
• Est. Priority Date: 08/31/2001
• Status: Active Grant

• Current Assignee: Nokia Technologies Oy (Nokia Corporation)
• Original Assignee: Nokia Corporation
• Inventors: Jensen, Kare Jean, Pedersen, Morten With, Riis, Soren
• Granted Patent: US 7,043,431 B2
• Time in Patent Office: Days
• US Class Current: 704/236
• CPC Class Codes: G10L 13/08 Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination
• CPC Class Codes: G10L 15/144 Training of HMMs
• CPC Class Codes: G10L 25/30 using neural networks

Abstract

There is provided a novel approach for generating multilingual text-to-phoneme mappings for use in multilingual speech recognition systems. The multilingual mappings are based on the weighted outputs from a neural network text-to-phoneme model, trained on data mixed from several languages. The multilingual mappings used together with a branched grammar decoding scheme is able to capture both inter- and intra-language pronunciation variations which is ideal for multilingual speaker independent speech recognition systems. A significant improvement in overall system performance is obtained for a multilingual speaker independent name dialing task when applying multilingual instead of language dependent text-to-phoneme mapping.

Claims

1. Method of speech recognition in order to identify a speech command as a match to a written text command, and comprising steps of:

providing a text input from a text database;

receiving an acoustic input;

generating sequences of multilingual phoneme symbols based on said text input by means of a multilingual text-to phoneme module;

generating pronunciations in response to said sequences of multilingual phoneme symbols; and

comparing said pronunciations with the acoustic input in order to find a match.

2. Method according to claim 1 wherein the text input is processed letter by letter, and wherein a neural network provides an estimate of the posterior probabilities of the different phonemes for each letter.

3. Method according to claim 1 comprising deriving said text input from a database containing user entered text strings.

4. System for speech recognition and comprising:

a text database for providing a text input;

transducer means for receiving an acoustic input;

a multilingual text-to phoneme module for outputting sequences of multilingual phoneme symbols based on said text input;

pronunciation lexicon module receiving said sequences of multilingual phoneme symbols from said multilingual text-to phoneme module, and for generating pronunciations in response thereto; and

a multilingual recognizer based on multilingual acoustic phoneme models for comparing said pronunciations generated by the pronunciation lexicon module with the acoustic input in order to find a match.

5. System according to claim 4, wherein the multilingual text-to phoneme module processes said text input letter by letter, and comprises a neural network for giving an estimate of the posterior probabilities of the different phonemes for each letter.

6. System according to claim 5 wherein the neural network is a standard fully connected feed-forward multi-layer perceptron neural network.

7. System according to claim 4 wherein the text input is derived from a database containing user entered text strings.

8. System according to claim 7 wherein the database containing user entered text strings is an electronic phonebook including phone numbers and associated name labels.

9. Communication terminal having for speech recognition unit comprising:

a text database for providing a text input;

transducer means for receiving an acoustic input;

a multilingual text-to phoneme module for outputting sequences of multilingual phoneme symbols based on said text input;

pronunciation lexicon module receiving said sequences of multilingual phoneme symbols from said multilingual text-to phoneme module, and for generating pronunciations in response thereto; and

a multilingual recognizer based on multilingual acoustic phoneme models for comparing said pronunciations generated by the pronunciation lexicon module with the acoustic input in order to find a match.

10. Communication terminal according to claim 9, wherein the multilingual text-to phoneme module processes said text input letter by letter, and comprises a neural network for giving an estimate of the posterior probabilities of the different phonemes for each letter.

11. Communication terminal according to claim 10 wherein the neural network is a standard fully connected feed-forward multi-layer perceptron neural network.

12. Communication terminal according to claim 9 wherein the text input is derived from a database containing user entered text strings.

13. Communication terminal according to claim 12 wherein the database containing user entered text strings is an electronic phonebook including phone numbers and associated name labels.