Speech recognition system, speech recognition method and storage medium

• US 8,010,359 B2
• Filed: 06/24/2005
• Issued: 08/30/2011
• Est. Priority Date: 03/16/2005
• Status: Expired due to Fees

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Matsuo, Naoshi
• Primary Examiner(s): Saint Cyr; Leonard
• Application Number:
  US11/165,120
• Publication Number:
  US 20060212291A1
• Time in Patent Office: 2,258 Days
• Field of Search: None
• US Class Current: 704/251
• CPC Class Codes: G10L 2015/228 of application context
• CPC Class Codes: G10L 21/028 using properties of sound source

Abstract

Provided are a speech recognition system, a method and a storage medium capable of, even in a case where plural speakers input superimposed speeches, recognizing a speech of an individual each speaker and making a single application program sharable among the speakers in execution. In a speech recognition system receiving speeches of plural speakers to execute a predetermined application program, the received speeches are separated according to the respective speakers if necessary, the received speeches of individual speakers are speech-recognized, results of speech recognition are matched with data items necessary for executing the application program, one of results of recognition of plural speeches which are found as a result of the matching to be overlapping is selected, and the results of recognition of plural speeches which are found as a result of the matching not to be overlapping are linked to the selected result of speech recognition.

Claims

1. A speech recognition system comprising:

an input part for receiving speeches from each of plural speakers;

a speech recognition part for speech-recognizing a speech received from each of the plural speakers;

a matching part for matching the results of speech recognition with data items necessary for executing an application program;

a selecting part for selecting one of the results of recognition of plural speeches which are found as a result of the matching to be overlapping in a data item necessary for executing the application program as a result of the matching; and

a linkage part for linking the selected result of speech recognition the results of recognition of the plural speeches which are found as results of the matching not to be overlapping in data items necessary to execute the application program based on the linked results of speech recognition, wherein

a priority level showing a precedence of speech of one speaker over speech of another speaker in selection of a result of speech recognition for each speaker is stored,

the selecting part preferentially selects a result of speech recognition of a speech uttered by a speaker with a highest priority level, the priority level being stored in advance of the speech recognition,

based on a result of speech recognition of one speaker, it is determined as to whether or not an empty data item exists among data items necessary for execution of an application program,

when it is determined that there is an empty data item, it is determined as to whether or not the result of speech recognition of the one speaker can be linked to a result of speech recognition on another speaker, and

linking the result of speech recognition of the one speaker to the result of speech recognition on another speaker when it is determined to be possible.

2. The speech recognition system of claim 1, wherein the speech recognition part calculates an evaluation value representing a degree of coincidence with a speech pattern stored in advance and outputs a character sequence having a largest calculated evaluation value as a result of recognition, and

the selecting part selects a result of speech recognition having the largest evaluation value among overlapping results of speech recognition of plural speeches.

3. The speech recognition system of claim 2, wherein the selecting part preferentially selects a result of speech recognition of a speech uttered later.

4. The speech recognition system of claim 1, wherein the selecting part preferentially selects a result of speech recognition of a speech uttered later.

5. The speech recognition system of claim 1, comprising speech separation part for separating received speeches according to the respective speakers.

6. A speech recognition system comprising a processor capable of performing:

receiving speeches from each of plural speakers;

speech-recognizing the received speeches;

matching results of speech recognition with data items necessary for executing an application program;

selecting one of results of recognition of plural speeches which are found as a result of the matching to be overlapping in a data item necessary for execution of the application program;

linking the selected result of speech recognition the results of recognition of plural speeches which are found as results of the matching not to be overlapping in data items necessary to execute the application program based on the linked results of speech recognition;

storing a priority level indicating a precedence of speech of one speaker over speech of another speaker in selection of a result of speech recognition for each speaker; and

preferentially selecting a result of speech recognition of a speech uttered by a speaker with a higher priority level, the priority level being stored before the speech recognition;

based on a result of speech recognition of one speaker, determining as to whether or not an empty data item exists among data items necessary for execution of an application program,

when it is determined that there is an empty data item, determining as to whether or not the result of speech recognition of the one speaker can be linked to a result of speech recognition on another speaker; and

linking the result of speech recognition of the one speaker to the result of speech recognition on another speaker when it is determined to be possible.

7. The speech recognition system of claim 6, comprising a processor further capable of performing:

calculating an evaluation value representing a degree of coincidence with patterns stored in advance;

outputting a character sequence having a largest calculated evaluation value, and

selecting a result of speech recognition having the largest evaluation value among overlapping results of recognition of plural speeches.

8. The speech recognition system of claim 7, comprising a processor further capable of performing:

preferentially selecting a result of recognition of a speech uttered later.

9. The speech recognition system of claim 6, comprising a processor further capable of performing:

preferentially selecting a result of recognition of a speech uttered later.

10. The speech recognition system of claim 6, comprising a processor further capable of performing:

separating received speeches according to the respective speakers.

11. A speech recognition method for causing a computer to function as a speech recognition system, the speech recognition method performed by the computer comprising steps of:

receiving speeches from each of plural speakers;

speech-recognizing the received speeches from each of the plural speakers;

matching results of speech recognition with data items necessary for executing an application program;

selecting one of results of recognition of plural speeches which are found as a result of the matching to be overlapping in a data item necessary for execution of the application program;

linking the selected result of speech recognition the results of recognition of plural speeches which are found as results of the matching not to be overlapping in data items necessary to execute the application program based on the linked results of speech recognition;

storing a priority level indicating a precedence of speech of one speaker over speech of another speaker in selection of a result of speech recognition for each speaker;

preferentially selecting a result of speech recognition of a speech uttered by a speaker with a higher priority level, the priority level being stored in advance of the speech recognition;

based on a result of speech recognition of one speaker, determining as to whether or not an empty data item exists among data items necessary for execution of an application program;

when it is determined that there is an empty data item, determining as to whether or not the result of speech recognition of the one speaker can be linked to a result of speech recognition on another speaker; and

linking the result of speech recognition of the one speaker to the result of speech recognition on another speaker when it is determined to be possible.

12. The speech recognition method of claim 11, the application program further comprising:

in a case where results of recognition of plural speeches overlapping in data items necessary for executing the application program are to be selected,

calculating an evaluation value representing a degree of coincidence with a speech pattern stored in advance;

outputting a character sequence having a largest calculated evaluation value, and

selecting a result of speech recognition having the largest evaluation value among overlapping results of recognition of plural speeches.

13. The speech recognition method of claim 11, the application program further comprising:

separating received speeches according to the respective speakers.

14. A non-transitory computer-readable storage medium for a given application program causing a computer to function as a given speech recognition system, the application program causing the computer-to execute:

receiving speeches of plural speakers;

executing the speech recognition program based on results of recognition of the received speeches;

speech-recognizing the received speeches of individual speakers;

matching results of recognition of speeches with data items necessary to execute the application program;

selecting one of results of recognition of plural speeches which are found as a result of the matching to be overlapping in a data item necessary to execute the application program; and

linking the selected result of speech recognition the results of recognition of plural speeches which are found as results of the matching not to be overlapping in data items necessary to execute the application program based on the linked results of recognition of the received speech, wherein

a priority level showing a precedence of speech of one speaker over speech of another speaker in selection of a result of speech recognition for each speaker is stored,

the selecting part preferentially selects a result of speech recognition of a speech uttered by a speaker with a highest priority level, the priority level stored in advance of the speech recognition,

based on a result of speech recognition of one speaker, determining as to whether or not an empty data item exists among data items necessary for execution of an application program,

when it is determined that there is an empty data item, determining as to whether or not the result of speech recognition of the one speaker can be linked to a result of speech recognition on another speaker, and

linking the result of speech recognition of the one speaker to the result of speech recognition on another speaker when it is determined to be possible.

15. The non-transitory computer-readable storage medium of claim 14, storing the computer application program comprising:

calculating an evaluation value representing a degree of coincidence with a speech pattern stored in advance;

outputting a character sequence having a largest calculated evaluation value; and

selecting a result of speech recognition having the largest evaluation value among overlapping results of recognition of plural speeches.

16. The non-transitory computer-readable storage medium of claim 14, the application program further comprising:

separating received speeches according to the respective speakers.

17. The non-transitory computer-readable storage medium of claim 15, the application program further comprising:

separating received speeches according to the respective speakers.