Method for speech recognition

• US 8,527,271 B2
• Filed: 06/18/2008
• Issued: 09/03/2013
• Est. Priority Date: 07/18/2007
• Status: Active Grant

• Current Assignee: Cerence Operating Company (Cerence Inc.)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Wandinger, Michael, Perez, Jesus Fernando Guitarte, Littel, Bernhard
• Primary Examiner(s): Saint Cyr, Leonard
• Application Number:
  US12/452,707
• Publication Number:
  US 20100286984A1
• Time in Patent Office: 1,903 Days
• Field of Search: 704/246, 704/247, 704/251, 704/252
• US Class Current: 704/251
• CPC Class Codes: G01C 21/3608 using speech input, e.g. using speech recognition
• CPC Class Codes: G10L 15/04 Segmentation; Word boundary detection
• CPC Class Codes: G10L 15/142 Hidden Markov Models [HMMs]
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command

Abstract

A method for the voice recognition of a spoken expression to be recognized, comprising a plurality of expression parts that are to be recognized. Partial voice recognition takes place on a first selected expression part, and depending on a selection of hits for the first expression part detected by the partial voice recognition, voice recognition on the first and further expression parts is executed.

Claims

1. A computer-implemented method for speech recognition of a spoken expression, which comprises a first expression part and remaining expression parts to be recognized, comprising performing by a processor operations of:

conducting a partial speech recognition on the first expression part using a Hidden-Markov-Model, wherein the remaining expression parts are covered by fill data, the partial speech recognition using a first vocabulary having a list of search terms for the first expression part;

selecting N best fitting hits for the first expression part from the list of search terms;

creating a second vocabulary, different than the first vocabulary, the second vocabulary comprising a list of search term combinations depending on the selected N best fitting hits for the first expression part, the second vocabulary having a list of search term combinations that covers the remaining expression parts previously covered by the fill data;

conducting a second speech recognition on the remaining expression parts using the second vocabulary; and

providing a result, including a speech-recognized result of the first expression part and a speech-recognized result of the remaining expression parts.

2. The method according to claim 1, wherein the first expression part is selected from the spoken expression based on a temporal energy contour of the spoken expression.

3. The method according to claim 1, wherein the first expression part is selected from the spoken expression based on a speech activity recognition in the expression.

4. The method according to claim 1, wherein the method recognizes the spoken expression using a hierarchically built database in which a generic city name term has further street address terms assigned thereto, and the further street address terms have even further house number subtopics are assigned thereto.

5. The method according to claim 1, wherein after conducting the partial speech recognition, the first vocabulary is deleted from a memory and the second vocabulary is loaded into the memory.

6. The method according to claim 1, wherein the first expression part is located at the beginning of the spoken expression.

7. The method according to claim 1 wherein at least parts of said second vocabulary are precompiled.

8. The method according to claim 1 wherein a partial vocabulary is precompiled for the second vocabulary, and reference vectors of a Hidden-Markov-Model are calculated and stored for the partial vocabulary.

9. The method according to claim 1, wherein the list of search terms of the first vocabulary comprises a list of city names, and the list of search term combinations of the second vocabulary comprises a list of city name and street address combinations.

10. A speech recognition device to recognize a spoken expression, which comprises a first expression part and remaining expression parts to be recognized, the speech recognition device comprising:

a database including a memory and providing at least one of search terms and a phonemic transcription of the search terms; and

a control unit connected to the database and configured to;

conduct a partial speech recognition on the first expression part using a Hidden-Markov-Model, wherein the remaining expression parts are covered by fill data, the partial speech recognition using a first vocabulary having a list of search terms for the first expression part;

select N best fitting hits for the first expression part from the list of search terms;

create a second vocabulary, different than the first vocabulary, the second vocabulary comprising a list of search term combinations depending on the selected N best fitting hits for the first expression part, the second vocabulary having a list of search term combinations that covers the remaining expression parts previously covered by the fill data;

conduct a second speech recognition on the remaining expression parts using the second vocabulary; and

provide a result, including a speech-recognized result of the first expression part and a speech-recognized result of the remaining expression parts.

11. The speech recognition device according to claim 10, wherein the speech recognition device is embedded in a motor vehicle.

12. The speech recognition device according to claim 11, wherein said speech recognition device is part of a vehicle navigation system.

13. The speech recognition device according to claim 11, wherein said speech recognition device is formed as a part of a mobile phone and/or of an audio player.

14. The speech recognition device according to claim 10, wherein the list of search terms of the first vocabulary comprises a list of city names, and the list of search term combinations of the second vocabulary comprises a list of city name and street address combinations.

15. A computer readable storage medium storing a program, which when executed by a computer performs a method for speech recognition of a spoken expression, which comprises a first expression part and remaining expression parts to be recognized, the method comprising:

conducting a partial speech recognition on the first expression part using a Hidden-Markov-Model wherein the remaining expression parts are covered by fill data, the partial speech recognition using a first vocabulary having a list of search terms for the first expression part;

selecting N best fitting hits for the first expression part from the list of search terms;

creating a second vocabulary, different than the first vocabulary, the second vocabulary comprising a list of search term combinations depending on the selected N best fitting hits for the first expression part, the second vocabulary having a list of search term combinations that covers the remaining expression parts previously covered by the fill data;

conducting a second speech recognition on the remaining expression parts using the second vocabulary; and

providing a result, including a speech-recognized result of the first expression part and a speech-recognized result of the remaining expression parts.

16. The computer readable storage medium according to claim 15, wherein the list of search terms of the first vocabulary comprises a list of city names, and the list of search term combinations of the second vocabulary comprises a list of city name and street address combinations.