System and method of developing a TTS voice

• US 7,693,716 B1
• Filed: 09/27/2005
• Issued: 04/06/2010
• Est. Priority Date: 09/27/2005
• Status: Active Grant

• Current Assignee: Cerence Operating Company (Cerence Inc.)
• Original Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Inventors: Davis, Steven Lawrence, Schulz, David Eugene, Loney, Louise, Fetters, Shane, Gustafson, Beverly
• Primary Examiner(s): Vo, Huyen X.
• Application Number:
  US11/235,954
• Time in Patent Office: 1,652 Days
• Field of Search: 704/268, 704/231, 704/243, 704/260, 704/258, 704/266, 704/261, 704/267, 704/270, 704/272
• US Class Current: 704/260
• CPC Class Codes: G10L 13/04 Details of speech synthesis systems, e.g. synthesiser structure or memory management

Abstract

Disclosed herein are various aspects 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. An embodiment of the invention relates to a method of tracking progress in developing a text-to-speech (TTS) voice. The method comprises insuring that a corpus of recorded speech contains reading errors and matches an associated written text, creating a tuple for each utterance in the corpus and tracking progress for each utterance utilizing the tuple. Various parameters may be tracked using the tuple but the tuple provides a means for enabling multiple workers to efficiently process a database of utterance in preparation of a TTS voice.

Claims

1. A method of tracking progress in developing a text-to-speech (TTS) voice, the method comprising:

ensuring that a corpus of recorded speech contains reading errors and matches an associated written text, wherein each of the reading errors comprises at least a half phenome incorrectly matched to a word of the written text;

creating, via a processor, a tuple of files for each utterance in the corpus; and

tracking, via the processor, work done for each utterance utilizing the tuple.

2. The method of claim 1, wherein each tuple comprises ASR-generated phonemes, pronunciation lists, confidence scores and a progress matrix.

3. The method of claim 2, wherein the progress matrix stores and tracks work performed on the tuple.

4. The method of claim 3, wherein the progress matrix further stores information about which person has performed work on the tuple.

5. The method of claim 4, wherein when different people perform work on each tuple, work-tracking information is stored in the progress matrix such that several people may simultaneously work on the corpus.

6. A tangible computer-readable storage medium storing instructions for controlling a computing device to track progress in developing a text-to-speech (TTS) voice, the instructions comprising:

ensuring that a corpus of recorded speech contains reading errors and matches an associated written text, wherein each of the reading errors comprises at least a half phenome incorrectly matched to a word of the written text;

creating a tuple of files for each utterance in the corpus; and

tracking work done for each utterance utilizing the tuple.

7. The tangible computer-readable storage medium of claim 6, wherein each tuple comprises ASR-generated phonemes, pronunciation lists, confidence scores and a progress matrix.

8. The tangible computer-readable storage medium of claim 7, wherein the progress matrix stores and tracks work performed on the tuple.

9. The tangible computer-readable storage medium of claim 8, wherein the progress matrix further stores information about which person has performed work on the tuple.

10. The tangible computer-readable storage medium of claim 9, wherein when different people perform work on each tuple, work-tracking information is stored in the progress matrix such that several people may simultaneously work on the corpus.

11. A computing device that tracks progress in developing a text-to-speech (TTS) voice, the computing device comprising:

a processor;

a module configured to control the processor to ensure that a corpus of recorded speech contains reading errors and matches an associated written text, wherein each of the reading errors comprises at least a half phenome incorrectly matched to a word of the written text;

a module configured to control the processor to create a tuple of files for each utterance in the corpus;

a module configured to control the processor to track work done for each utterance utilizing the tuple.

12. The computing device of claim 11, wherein each tuple comprises ASR-generated phonemes, pronunciation lists, confidence scores and a progress matrix.

13. The computing device of claim 12, wherein the progress matrix stores and tracks work performed on the tuple.

14. The computing device of claim 13, wherein the progress matrix further stores information about which person has performed work on the tuple.

15. The computing device of claim 14, wherein when different people perform work on each tuple, work-tracking information is stored in the progress matrix such that several people may simultaneously work on the corpus.