Verbal, fully automatic dictionary updates by end-users of speech synthesis and recognition systems

• US 6,078,885 A
• Filed: 05/08/1998
• Issued: 06/20/2000
• Est. Priority Date: 05/08/1998
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Beutnagel, Mark C.
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US09/075,162
• Time in Patent Office: 774 Days
• Field of Search: 704/258, 704/260, 704/209, 704/231
• US Class Current: 704/258
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 2015/025 Phonemes, fenemes or fenones being the recognition units
• CPC Class Codes: G10L 2015/0633 using lexical or orthographic knowledge sources

Abstract

A method and system that allows users, or maintainers, of a speech-based application to revise the phonetic transcription of words in a phonetic dictionary, or to add transcriptions for words not yet present in the dictionary. The application is assumed to communicate with the user or maintainer audibly by means of speech recognition and/or speech synthesis systems, both of which rely on a dictionary of phonetic transcriptions to accurately recognize speech and pronunciation of a given word. The method automatically determines the phonetic transcription based on the word'"'"'s spelling and the recorded preferred pronunciation, and updates the dictionary accordingly. Moreover, both speech synthesis and recognition performance are improved through use of the updated dictionary.

Claims

1. A method of allowing an end-user of a text-to-speech system or a speech recognition system to verbally update a phonetic dictionary, comprising the steps of:

recording a verbal pronunciation of at least one word, as spoken by the user;

generating a phonetic transcription of the at least one word based on the verbal pronunciation;

augmenting the phonetic transcription with syllable stress markers based on the verbal pronunciation; and

entering the phonetic transcription into the dictionary.

2. The method of claim 1, wherein the verbal pronunciation is digitally recorded.

3. The method of claim 1, further comprising the steps of:

receiving an orthography of the at least one word;

generating candidate pronunciations;

generating a recognition grammar for the at least one word from the candidate pronunciations; and

comparing the verbal pronunciation against the recognition grammar to generate the phonetic transcription.

4. The method of claim 1, further comprising the step of:

permitting the individual to associate a part of speech tag with the phonetic transcription.

5. The method of claim 1, wherein the step of augmenting the transcription with syllable stress markers is based on acoustical features of a phoneme of the verbal pronunciation of the at least one word.

6. The method of claim 5, wherein the acoustical features are selected from the group consisting of the identity of the phoneme, the duration of the phoneme, the energy of the phoneme, the normalized energy of the phoneme, the normalized duration of the phoneme, the fundamental frequency of the phoneme, and the normalized fundamental frequency of the phoneme.

7. The method of claim 1, further comprising the steps of:

using the phonetic transcription to speak the at least one word back to the individual for validation; and

receiving an acceptance or a rejection of the phonetic transcription from the individual.

8. An article of manufacture for allowing an end-user of a text-to-speech system or a speech recognition system to verbally update a phonetic dictionary, comprising:

a computer readable medium having computer readable program code stored therein, the computer readable code for causing a computer system to receive a speech signal corresponding to at least one word spoken by the end-user, convert the speech signal into a phonetic transcription of the at least one word augment the phonetic transcription with syllable stress markers based on the speech signal, and enter the phonetic transcription into the dictionary.

9. A method for a text-to-speech (TTS) system to update individual entries in a phonetic dictionary, comprising the steps of:

receiving an indication from an end-user that the TTS system has mispronounced at least one word;

after receiving said indication, recording a verbal pronunciation of the at least one word as spoken by the end-user;

determining a phonetic transcription that corresponds to the at least one word as spoken by the end-user; and

storing the phonetic transcription in the dictionary.

10. The method of claim 9, wherein the at least one word as spoken by the end-user is digitally recorded.

11. The method of claim 9, wherein the steps of determining a phonetic transcription includes the steps of:

receiving an orthography of the at least one word;

generating candidate pronunciations;

generating a recognition grammar for the at least one word based on the candidate pronunciations; and

selecting a member of the recognition grammar whose phonemes match sounds in the recording of the at least one word.

12. The method of claim 11, wherein the step of determining a phonetic transcription includes the step of augmenting the transcription with syllable stress markers.

13. The method of claim 9, further comprising the steps of:

using the phonetic transcription to speak the at least one word back to the end-user for validation; and

receiving an acceptance or a rejection of the phonetic transcription from the end-user.

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

using the phonetic transcription as a default transcription, when the at least one word is encountered by the TTS system in text.

15. A method for a speech recognition system to update individual entries in a phonetic dictionary, comprising the steps of:

receiving an indication from an end-user that a phonetic transcription of at least one word in the dictionary should be updated;

after receiving said indication, recording a verbal pronunciation of the at least one word as spoken by the end-user;

determining a phonetic transcription that corresponds to the at least one word as spoken by the end-user; and

storing the phonetic transcription in the dictionary.

16. The method of claim 15, wherein the at least one word as spoken by the end-user is digitally recorded.

17. The method of claim 15, wherein the steps of determining a phonetic transcription includes the steps of:

generating candidate pronunciations;

generating a recognition grammar for the at least one word based on the candidate pronunciations; and

selecting a member of the recognition grammar whose phonemes match sounds in the recording of the at least one word.

18. The method of claim 15, wherein the step of determining a phonetic transcription includes the step of augmenting the transcription with syllable stress markers.

19. The method of claim 15, further comprising the steps of:

using the phonetic transcription to speak the at least one word back to the end-user for validation; and

receiving an acceptance or a rejection of the phonetic transcription from the end-user.

20. The method of claim 15, further comprising the step of:

using the phonetic transcription as a default transcription, when the at least one word is encountered by the speech recognition system in speech.

21. A system for allowing an end-user of a text-to-speech system or a speech recognition system to verbally update a phonetic dictionary, comprising:

a memory device storing the phonetic dictionary;

a processor in communication with said memory device, the processor configured to receive a recorded verbal pronunciation of at least one word, as spoken by the end-user, generate a phonetic transcription of the at least one word based on the verbal pronunciation, augment the phonetic transcription with syllable stress markers based on the verbal pronunciation, and enter the phonetic transcription into the dictionary.

22. The system of claim 21, wherein the processor is further configured to receive an orthography of the at least one word whose pronunciation is to be recorded.

23. The system of claim 21, wherein the processor is further configured to generate a recognition grammar for the at least one word and compare the recorded verbal pronunciation against the recognition grammar to generate the phonetic transcription.

24. The system of claim 21, wherein the processor is further configured to speak the at least one word back to the individual using the phonetic transcription, and receive an acceptance or a rejection of the phonetic transcription from the individual.