System and method for preparing a pronunciation dictionary for a text-to-speech voice

• US 7,630,898 B1
• Filed: 09/27/2005
• Issued: 12/08/2009
• Est. Priority Date: 09/27/2005
• Status: Active Grant

• Current Assignee: Cerence Inc., Cerence Operating Company (Cerence Inc.)
• Original Assignee: AT&T Intellectual Property II LP (AT&T, Inc.)
• Inventors: Davis, Steven Lawrence, Fetters, Shane, Schulz, David Eugene, Gustafson, Beverly, Loney, Louise
• Primary Examiner(s): Hudspeth, David R
• Assistant Examiner(s): ALBERTALLI, BRIAN LOUIS
• Application Number:
  US11/235,817
• Time in Patent Office: 1,533 Days
• Field of Search: None
• US Class Current: 704/266
• CPC Class Codes: G06F 40/242 Dictionaries
• 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 15/187 Phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams

Abstract

Disclosed are various elements of a toolkit used for generating a TTS voice for use in a spoken dialog system. The embodiments in each case may be in the form of the system, a computer-readable medium or a method for generating the TTS voice. One embodiment of the invention relates to a method of generating a database for a TTS voice. The method comprises matching every spoken word associated with a TTS voice database with a smallest set of possible pronunciations for each word. The smallest set is generated by automatically determining a dialect and linguistic context using linguistic rules, empirically determining idiosyncratic speaker characteristics and determining a subject domain. The method further comprises dynamically generating a pronunciation dictionary on a word-by-word basis using the smallest set.

Claims

1. A computer-implemented method of generating a database for a text-to-speech (TTS) voice, the method comprising:

matching via a processor every spoken word associated with a TTS voice database with a smallest set of possible pronunciations for each word, the smallest set being generated by;

automatically via the processor determining a dialect and linguistic context using linguistic rules;

empirically determining idiosyncratic speaker characteristics; and

determining a subject domain; and

dynamically generating a pronunciation dictionary on a word-by-word basis using the smallest set.

2. The computer-implemented method of claim 1, wherein the dynamically generated pronunciation dictionary accounts for reading errors and idiosyncrasies of speakers.

3. The computer-implemented method of claim 1, further comprising:

forcing an automatic speech recognition (ASR) module to choose from a subset of one or more variants of a word when more than one pronunciation variants exists for a given word.

4. The computer-implemented method of claim 1, further comprising:

automatically generating phonetic variant pronunciations for the pronunciation dictionary for any given word.

5. The computer-implemented method of claim 4, wherein generating phonetic variant pronunciations is based on a surrounding linguistic context for each word.

6. The computer-implemented method of claim 1, wherein the linguistic contexts are associated with foreign languages.

7. The computer-implemented method of claim 1, further comprising:

tracking whether each lexical pronunciation is either machine generated or human entered.

8. The computer-implemented method of claim 7, further comprising:

flagging machine generated lexical pronunciations for human inspection.

9. The computer-implemented method of claim 1, further comprising adding default pronunciations to the pronunciation dictionary based on TTS letter-to-sound rules.

15. The computing device of claim 9, wherein the linguistic contexts are associated with foreign languages.

10. A computing device for generating a database for a text-to-speech (TTS) voice, the computing device comprising:

a module configured to control the processor to match every spoken word associated with a TTS voice database with a smallest set of possible pronunciations for each word, the smallest set being generated by;

automatically via the processor determining a dialect and linguistic context using linguistic rules;

empirically determining idiosyncratic speaker characteristics; and

determining a subject domain; and

a module configured to control the processor to dynamically generate a pronunciation dictionary on a word-by-word basis using the smallest set.

11. The computing device of claim 10, wherein the dynamically generated pronunciation dictionary accounts for reading errors and idiosyncrasies of speakers.

12. The computing device of claim 10, further comprising:

a module configured to force an automatic speech recognition (ASR) module to choose from a subset of one or more variants of a word when more than one pronunciation variants exist for a given word.

13. The computing device of claim 10, further comprising:

a module configured to automatically generate phonetic variant pronunciations for the pronunciation dictionary for any given word.

14. The computing device of claim 13, wherein generating phonetic variant pronunciations is based on a surrounding linguistic context for each word.

16. The computing device of claim 10, further comprising:

a module configured to track whether each lexical pronunciation is either machine generated or human entered; and

a module configured to flag machine generated lexical pronunciations for human inspection.

17. The computing device of claim 10, further comprising a module configured to add default pronunciations to the pronunciation dictionary based on TTS letter-to-sound rules.

18. A tangible computer-readable storage medium storing instructions for controlling a computing device for generating a database for a text-to-speech (TTS) voice, the instructions comprising:

matching via a processor every spoken word associated with a TTS voice database with a smallest set of possible pronunciations for each word, the smallest set being generated by;

automatically determining a dialect and linguistic context using linguistic rules;

empirically determining idiosyncratic speaker characteristics; and

determining a subject domain; and

dynamically generating a pronunciation dictionary on a word-by-word basis using the smallest set.

19. The computer-readable medium of claim 18, wherein the dynamically generated pronunciation dictionary accounts for reading errors and idiosyncrasies of speakers.

20. The computer-readable medium of claim 19, the instructions further comprising:

forcing an automatic speech recognition (ASR) module to choose from a subset of one or more variants of a word when more than one pronunciation variants exists for a given word.