METHOD AND SYSTEM FOR CREATING OR UPDATING ENTRIES IN A SPEECH RECOGNITION LEXICON

• US 20100057461A1
• Filed: 02/04/2008
• Published: 03/04/2010
• Est. Priority Date: 02/06/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications Austria Gmbh (Microsoft Corporation)
• Inventors: Neubacher, Andreas, Grobauer, Gerhard
• Granted Patent: US 8,447,606 B2
• Time in Patent Office: Days
• US Class Current: 704/243
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 2015/0638 Interactive procedures

Abstract

In a method and a system (20) for creating or updating entries in a speech recognition (SR) lexicon (7) of a speech recognition system, said entries mapping speech recognition (SR) phoneme sequences to words, said method comprising entering a respective word, and in the case that the word is a new word to be added to the SR lexicon, also entering at least one associated SR phoneme sequence through input means (26), it is provided that the SR phoneme sequence associated with the respective word is converted into speech by phoneme to speech conversion means (4.4), and the speech is played back by playback means (28), to control the match of the phoneme sequence and the word.

Claims

1. A method for providing entries in a speech recognition (SR) lexicon (7) of a speech recognition system (1), said entries mapping speech recognition (SR) phoneme sequences to words, said method comprising receiving a respective word, and if the respective word is a new word to be added to the speech recognition (SR) lexicon, also receiving at least one associated SR phoneme sequence through input means (26), and converting the associated SR phoneme sequence into speech by phoneme to speech conversion means (4.4), and playing back the speech to enable control of a match of the SR phoneme sequence and the respective word.

2. The method according to claim 1, characterized in that the associated SR phoneme sequence as received by the SR lexicon (7) is automatically generated by phonetic transcription.

3. The method according to claim 2, characterized in that the respective word to be added is converted into the associated SR phoneme sequence by phonetic transcription.

4. The method according to claim 2, characterized in that a spoken word like the respective word to be added is converted into the associated SR phoneme sequence by phonetic transcription.

5. The method according to claim 1, characterized in that the associated SR phoneme sequence is manually entered into the lexicon.

6. The method according to claim 2, characterized in that the conversion as applied to the associated SR phoneme sequence to obtain the speech is at least substantially an inverse of the conversion as realized by the phonetic transcription.

7. The method according to claim 1, characterized in that when there is a mismatch between the speech as played back and the respective word, a modified SR phoneme sequence is entered, or a modified spoken word is received and converted into a modified SR phoneme sequence by phonetic transcription, said modified SR phoneme sequence being received thereafter into the SR lexicon (7).

8. The method according to claim 1, characterized in that the entered SR phoneme sequence is enhanced with prosody when being converted into speech.

9. A system (20) for providing entries in a speech recognition (SR) lexicon (7) of a speech recognition system, said entries mapping speech recognition (SR) phoneme sequences to words, said system comprising word and SR phoneme sequence input means (21;

25, 4.3) which are associated with said SR lexicon (7), characterized in that phoneme to speech conversion and playback means (27;

4.4, 28) are associated with said SR lexicon (7) for selectively converting phoneme sequences stored in the SR lexicon into speech, and for playing back the speech.

10. The system according to claim 9, characterized in that the phoneme to speech conversion means (4.4) for converting SR phoneme sequences into speech are at least substantially an inverse of the phonetic transcription means (4.1).

11. The system according to claim 9, characterized in that the word and SR phoneme sequence input means (25, 4.3) comprise means (25) for manually entering words and/or texts which are spoken like said words, and means (4.3) for converting the words or spoken-like texts into SR phoneme sequences.

12. The system according to claim 9, characterized by input means (21) for manually entering phoneme sequences into the SR lexicon (7).

13. The system according to claim 9, characterized in that the phoneme to speech conversion means (4.4) comprise phoneme enhancement means (44) for enhancement of generated SR phoneme sequences with prosody.

14. A speech recognition system (1) comprising a system (20) according to anyone of claims 9 to 13.