Processing received data in a distributed speech recognition process

• US 6,732,072 B1
• Filed: 07/25/2001
• Issued: 05/04/2004
• Est. Priority Date: 11/13/1998
• Status: Expired due to Term

• Current Assignee: Google Technology Holdings LLC (Alphabet Inc.)
• Original Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Inventors: Evans, Alun Christopher, Pearce, David John Benjamin, Gibbs, Jon Alastair
• Primary Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US09/830,305
• Time in Patent Office: 1,014 Days
• Field of Search: 704/200, 704/251, 704/270, 704/231, 375/220, 375/240.27
• US Class Current: 704/231
• CPC Class Codes: H04L 1/0083 Formatting with frames or packets; Protocol or part of protocol for error control

Abstract

A method of processing received data in a distributed speech recognition process. The method comprises the steps of comparing a received form of the common header information from each of a plurality of header frames; classifying a predetermined number of the same received form as a reference; and processing one or more received data frames whose received common header information is different to the reference by using the reference form instead of the received different form. Also described is an apparatus for processing received data in a distributed speech recognition process.

Claims

1. A method of processing received data in a distributed speech recognition process, the distributed speech recognition process being one in which data is arranged in a frame structure comprising a plurality of data frames each preceded by a respective header frame comprising header information including common header information, said data is received at a second location having been transmitted from a first location, and said data is processed using respective header frame information;

the method comprising the steps of;

comparing a received form of said common header information from each of a plurality of header frames;

classifying, when a predetermined number of said header frames are found to have the same received form of said common header information, the same received form as a reference; and

processing one or more received data frames corresponding to respective header frames whose received form of the common header information is different to said reference by using the reference form instead of the received different form.

2. A method according to claim 1, wherein said header information further comprises frame-specific header information, and wherein the frame-specific header information of the one or more header frames whose received form of the common header information is different to said reference is excluded from use.

3. A method according to claim 1, wherein said header information further comprises frame-specific header information, and wherein the frame-specific header information of the one or more header frames whose received form of the common header information is different to said reference is used in a different way compared to the frame-specific header information of those header frames whose received form of the common header information is the same as said reference.

4. A method according to claim 3, wherein the frame-specific information used in a different way is an end-of-message indicator of a first end of message recognition method, and responsive thereto is triggered a second end-of-message recognition method.

5. A method according to any preceding claim 4, wherein said header information is forward error correction encoded using a block-based coding scheme.

6. A method according to any preceding claim 5, wherein data frames received before said reference is determined are buffered prior to processing after said reference is determined.

7. An apparatus for processing received data in a distributed speech recognition process, the distributed speech recognition process being one in which data is arranged in a frame structure comprising a plurality of data frames each preceded by a respective header frame comprising header information including common header information, said data is received at a second location having been transmitted from a first location, and said data is processed using respective header frame information;

the apparatus comprising;

means for comparing a received form of said common header information from each of a plurality of header frames;

means for classifying, when a predetermined number of said header frames are found to have the same received form of said common header information, the same received form as a reference; and

means for processing one or more received data frames corresponding to respective header frames whose received form of the common header information is different to said reference by using the reference form instead of the received different form.

8. An apparatus according to claim 7, wherein said header information further comprises frame-specific header information, and wherein the frame-specific header information of the one or more header frames whose received form of the common header information is different to said reference is excluded from use.

9. An apparatus according to claim 7, wherein said header information further comprises frame-specific header information, and wherein the frame-specific header information of the one or more header frames whose received form of the common header information is different to said reference is used in a different way compared to the frame-specific header information of those header frames whose received form of the common header information is the same as said reference.

10. An apparatus according to claim 9, wherein the frame-specific information used in a different way is an end-of-message indicator of a first end of message recognition method, and responsive thereto is triggered a second end-of-message recognition method.

11. An apparatus according to any of claims 7-10, wherein said header information is forward error correction encoded using a block-based coding scheme.

12. An apparatus according to any of claims 7-11, wherein data frames received before said reference is determined are buffered prior to processing after said reference is determined.