Server-client type speech recognition apparatus and method

• US 20040243414A1
• Filed: 07/13/2004
• Published: 12/02/2004
• Est. Priority Date: 06/20/2001
• Status: Active Grant

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Yamada, Eiko, Hagane, Hiroshi, Yoshida, Kazunaga
• Granted Patent: US 7,478,046 B2
• Time in Patent Office: Days
• US Class Current: 704/270.1
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems

Abstract

To provide a speech recognition apparatus which enables the reduction of transmission time and of costs. A terminal-side apparatus (100) includes a speech detection portion (101) for detecting a speech interval of inputted data, a waveform compression portion (102) for compressing waveform data at the detected speech interval, and a waveform transmission portion (103) for producing the compressed waveform data. A server-side apparatus (200) includes a waveform reception portion (201) for receiving the waveform data transmitted from the terminal-side apparatus, a waveform decompression portion (202) for decompressing the received waveform data, an analyzing portion (203) for analyzing the decompressed waveform data, and a recognizing portion (204) for performing recognition processing to produce a recognition result.

Claims

1. A speech recognition apparatus comprising a terminal-side apparatus (100) and a server-side apparatus (200), wherein said terminal-side apparatus (100) comprises:

a speech detection portion (101) for detecting a speech interval of received speech data to produce waveform data of the detected speech interval;

a compression portion (102) for compressing the waveform data at the detected speech interval to produce compressed waveform data; and

a waveform transmission portion (103) for transmitting the compressed waveform data to said server-side apparatus, and

wherein said server-side apparatus (200) comprises;

a waveform reception portion (201) for receiving the compressed waveform data transmitted from said terminal-side apparatus to produce received waveform data;

a waveform decompression portion (202) for decompressing the received waveform data to produce decompressed waveform data; and

recognizing means (203, 204, 205) for performing recognition processing by using the decompressed waveform data to produce a recognition result.

2. A speech recognition apparatus comprising a terminal-side apparatus (100A) and a server-side apparatus (200A), wherein said terminal-side apparatus (100A) comprises:

a waveform and signal reception portion (104) for receiving waveform data of a received speech to produce received waveform data and for receiving a waveform data re-transmission request signal transmitted from said server-side apparatus to produce a received waveform date re-transmission request signal;

a speech section portion (101) for detecting a speech interval of the received waveform data to produce waveform data at the detected speech interval;

a waveform compression portion (102) for compressing the waveform data at the detected speech interval to produce compressed waveform data;

a waveform storing portion (105) for temporarily storing the compressed waveform data as the stored waveform data to simultaneously produce the stored waveform data and for producing the stored waveform data in response to the received waveform data re-transmission request signal; and

a waveform transmission portion (103) for transmitting the stored waveform data to said server-side apparatus, and

wherein said server-side apparatus (200A) comprises;

a waveform reception portion (201A) for receiving the compressed waveform data transmitted from said terminal-side apparatus to produce received waveform data and for producing the waveform data re-transmission request signal when the reception of the compressed waveform data fails;

a waveform decompression portion (202) for decompressing the received waveform data to produce decompressed waveform data;

recognizing means (203, 204, 205) for performing recognition processing by using the decompressed waveform data to produce a recognition result; and

a waveform data re-transmission request signal transmission portion (206) for transmitting, to said server-side apparatus 200, the waveform data re-transmission request signal received from said waveform reception portion.

3. A speech recognition apparatus comprising a terminal-side apparatus (100B, 100C) and a server-side apparatus (200B, 200C), wherein said terminal-side apparatus (100B, 100C) comprises:

a waveform and signal reception portion (104) for receiving waveform data of a received speech to produce received waveform data and for receiving a waveform data re-transmission request signal transmitted from said server-side apparatus to produce received waveform data re-transmission request signal;

a speech detection portion (101A) for detecting a speech interval of the received waveform data to produce waveform data at the detected speech interval and for producing a start-point cancel signal when and the speech is detected and thereafter the detection is canceled;

a waveform compression portion (102, 102A) for compressing the waveform data at the detected speech interval to produce compressed waveform data;

a waveform storing portion (105) for temporarily storing the compressed waveform data as the stored waveform data to simultaneously produce the stored waveform data and for producing the stored waveform data in response to the received waveform data re-transmission request signal;

a waveform transmission portion (103) for transmitting the stored waveform data to said server-side apparatus;

a start-point cancel signal transmission portion (106) for transmitting the start-point cancel signal outputted from said speech detection portion to said server-side apparatus, and

wherein said server-side apparatus (200B, 200C) comprises;

a waveform and signal reception portion (201B) for receiving compressed waveform data and the start-point cancel signal from said terminal-side apparatus to produce received waveform data and received start-point cancel signal and for producing a waveform data re-transmission request signal when the reception of the compressed waveform data fails;

a waveform decompression portion (202, 202A) for decompressing the received waveform data to produce decompressed waveform data;

recognizing means (203, 204A, 204B, 205) for performing recognition processing by using the decompressed waveform data to produce a recognition result and for stopping the recognition processing in response to the received start-point cancel signal; and

a waveform data re-transmission request signal transmission portion (206) for transmitting, to said server-side apparatus, the waveform data re-transmission request signal from said waveform and signal reception portion.

4. A speech recognition apparatus as claimed in claim 3, wherein said terminal-side apparatus (100C) further comprises:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis sound to produce formed synthesis sound information and for producing the synthesis sound,

wherein said waveform compression portion (102A) sends the formed synthesis sound information with it contained in a part of the compressed waveform data,

wherein said server-side apparatus (200C) further comprises;

a synthesis sound information obtaining portion (207) for obtaining the synthesis sound information from the waveform data decompressed by said waveform decompression portion to produce obtained synthesis sound information, and

wherein said recognizing means (204B) associates the obtained synthesis sound information with the recognition result upon ending the recognition to produce associated recognition result or the recognition result and the synthesis sound information.

5. A speech recognition apparatus comprising a terminal-side apparatus (100D) and a server-side apparatus (200D), wherein said terminal-side apparatus (100D) comprises:

a waveform, signal and compressing method reception portion (104A) for receiving at least inputted waveform data, a waveform data re-transmission request signal transmitted from said server-side apparatus, and compressing method information available to said server-side apparatus transmitted from said server-side apparatus to produce received waveform data, a received waveform data re-transmission request signal, and received compressing method information;

a speech detection portion (101A) for detecting a speech interval of the received waveform data to produce waveform data at the detected speech interval;

a compressing method selection portion (110) for selecting an optimum compressing method from the received compressing method information to produce a selected compressing method;

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index;

a waveform compression portion (102B) for compressing the waveform data at the detected speech interval to produce compressed waveform data with the formed compressing method index contained in a part of the compressed waveform data;

a waveform storing portion (105) for temporarily storing the compressed waveform data as the stored waveform data to produce simultaneously the stored waveform data and for producing the stored waveform data in response to the received waveform data re-transmission request signal;

a waveform transmission portion (103) for transmitting the stored waveform data to said server-side apparatus; and

a compressing method request signal transmission portion (112) for transmitting a compressing method request signal to said server-side apparatus,

wherein said server-side apparatus (200D) comprises;

a waveform and signal reception portion (201C) for receiving compressed waveform data and the compressing method request signal transmitted from said terminal-side apparatus to produce received waveform data and a received compressing method request signal and for producing a waveform data re-transmission request signal when the reception of the compressed waveform data fails;

a waveform decompression portion (202B) for decompressing the received waveform data to produce decompressed waveform data;

recognizing means (203A, 204C, 205A) for performing recognition processing by using the decompressed waveform data to produce a recognition result;

a waveform data re-transmission request signal transmission portion (206) for transmitting, to said server-side apparatus, the waveform data re-transmission request signal outputted from said waveform and signal reception portion;

a compressing method storing portion (212) for storing compressing method information available to said server-side apparatus;

a compressing method obtaining portion (211) for obtaining, in response to the received compressing method request signal, the compressing method information stored in said compressing method storing portion to transmit the compressing method information to said terminal-side apparatus;

a compressing method index obtaining portion (208) for obtaining an index of the compressing method from the decompressed waveform data to produce an obtained compressing method index;

a recognition engine selection portion (210) for selecting a recognition engine from the obtained compressing method index to produce a selected engine; and

a recognition engine setting portion (210) for setting the selected engine to said recognizing means from stored engines.

6. A speech recognition apparatus as claimed in claim 5, wherein said terminal-side apparatus (100D) further comprises:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis sound to produce formed synthesis sound information on the synthesis sound and for producing the synthesis sound,

wherein said waveform compression portion (102B) sends, to said waveform storing portion, the formed synthesis sound information with it contained in a part of the compressed waveform data,

wherein said server-side apparatus (200D) further comprises;

a synthesis sound information obtaining portion (207) for obtaining the synthesis sound information from the decompressed waveform data to produce obtained synthesis sound information, and

wherein said recognizing means (204C) associates the synthesis sound with the recognition result from the obtained synthesis sound information upon ending the recognition to produce an associated recognition result or the recognition result and the synthesis sound information.

7. A speech recognition apparatus comprising a terminal-side apparatus (100E) and a server-side apparatus (200E), wherein said terminal-side apparatus (100E) comprises:

a waveform, signal and compressing method reception portion (104A) for receiving at least inputted waveform data, a waveform data re-transmission request signal transmitted from said server-side apparatus, and compressing method information available to said server-side apparatus transmitted from said server-side apparatus to produce received waveform data, a received waveform data re-transmission request signal, and received compressing method information;

a speech detection portion (101A) for detecting a speech interval of the received waveform data to produce waveform data at the detected speech interval and for producing a start-point cancel signal when the speech is detected and thereafter the detection is canceled;

a compressing method selection portion (110) for selecting an optimum compressing method from the received compressing method information to produce a selected compressing method;

a compassing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index;

a waveform compression portion (102B) for compressing the waveform data at the detected speech interval to produce compressed waveform data with the formed compressing method index contained in a part of the compressed waveform data;

a waveform storing portion (105) for temporarily storing the compressed waveform data as the stored waveform data to produce simultaneously the stored waveform data and for producing the stored waveform data in response to the received waveform data re-transmission request signal;

a waveform transmission portion (103) for transmitting the stored waveform data to said server-side apparatus;

a start-point cancel signal transmission portion (106) for transmitting, to said server-side apparatus, the start-point cancel signal outputted from said speech detection portion; and

a compressing method request signal transmission portion (112) for transmitting a compressing method request signal to said server-side apparatus,

wherein said server-side apparatus (200E) comprises;

a waveform, signal, and task information reception portion (201D) for receiving compressed waveform data, the start-point cancel signal, the compressing method request signal from said terminal-side apparatus, and task information transmitted from a contents side to produce received waveform data, a received start-point cancel signal, a received compressing method request signal, and a received task information, and for producing a waveform data re-transmission request signal when the reception of the compressed waveform data fails;

a waveform decompression portion (202B) for decompressing the received waveform data to produce decompressed waveform data;

recognizing means (203A, 204C, 205A) for performing recognition processing by using the decompressed waveform data to produce a recognition result and for stopping the recognition processing in response to the received start-point cancel signal;

a waveform data re-transmission request signal transmission portion (206) for transmitting, to said server-side apparatus, the waveform data re-transmission request signal outputted from said waveform and signal reception portion;

a task information storing portion (213) for storing the received task information to produce stored task information;

a compressing method and task information corresponding table storing portion (212A) for storing the task information and one or more compressing methods available to the use of a task;

a compressing method obtaining portion (211A) for obtaining, in response to the received compressing method request signal, available compressing method information from the stored task information and the corresponding table between the task information and the compressing method transmitted from said compressing method and task information corresponding table storing portion to transmitting the compressing method information to said terminal-side apparatus;

a compressing method index obtaining portion (208) for obtaining an index of the compressing method from the decompressed waveform data to produce an obtained compressing method index;

a recognition engine selection portion (209) for selecting a recognition engine from the obtained compressing method index to produce a selected engine; and

a recognition engine setting portion (210) for setting the selected engine to said recognizing means from stored engines.

8. A speech recognition apparatus as claimed in claim 7, wherein said terminal-side apparatus (100E) further comprises:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis sound to produce formed synthesis sound information and for producing the synthesis sound,

wherein said waveform compression portion (102B) sends the formed synthesis sound information with it contained in a part of the compressed waveform data,

wherein said server-side apparatus (200E) further comprises;

a synthesis sound information obtaining portion (207) for obtaining the synthesis sound information from the decompressed waveform data to produce obtained synthesis sound information, and

wherein said recognizing means (204C) associates the synthesis sound with the recognition result from the obtained synthesis sound information upon ending the recognition to produce associated recognition result or the recognition result and the synthesis sound information.

9. A speech recognition apparatus comprising a terminal-side apparatus (100F) and a server-side apparatus (200F), wherein said terminal-side apparatus (100F) comprises:

a waveform, signal, compressing method, and task information reception portion (104B) for receiving inputted waveform data, task information transmitted from the contents side, a waveform data re-transmission request signal transmitted from said server-side apparatus, and compressing method information available to said server-side apparatus transmitted from said server-side apparatus to produce received waveform data, received task information, received waveform data re-transmission request signal, and received compressing method information;

a task information storing portion (113) for storing the received task information to produce stored task information;

a compressing method and task information corresponding table storing portion (111A) for storing a corresponding table between the task information and at least one ore more compressing methods available to the use a task;

a compressing method selection portion (110A) for selecting, in response to the received compressing method information, an optimum compressing method based on the stored task information and the corresponding table between the task information and the compressing method transmitted from said compressing method and task information corresponding table storing portion to produce a selected compressing method;

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index;

a speech detection portion (101A) for detecting a speech interval of the received waveform data to produce waveform data at the detected speech interval;

a waveform compressing portion (102B) for compressing the waveform data at the detected speech interval to produce compressed waveform data with the formed compressing method index contained in a part of the compressed waveform data;

a waveform storing portion (105) for temporarily storing t the compressed waveform data as the stored waveform data to produce the stored waveform data and for producing the stored waveform data in response to the received waveform data re-transmission request signal;

a waveform transmission portion (103) for transmitting the stored waveform data to said server-side apparatus; and

a compressing method request signal transmission portion (112) for transmitting a compressing method request signal to said server-side apparatus,

wherein said server-side apparatus (200F) comprises;

a waveform and signal reception portion (201C) for receiving compressed waveform data transmitted from said terminal-side apparatus and the compressing method request signal to produce received waveform data and a received compressing method request signal and for producing a waveform data re-transmission request signal when the reception of the compressed waveform data fails;

a waveform decompression portion (202B) for decompressing the received waveform data to produce decompressed waveform data;

recognizing means (203A, 204C, 205A) for performing recognition processing by using the decompressed waveform data to produce a recognition result;

a waveform data re-transmission request signal transmission portion (206) for transmitting, to said server-side apparatus, the waveform data re-transmission request signal outputted from said waveform and signal reception portion;

a compressing method storing portion (212) for storing information on the compressing methods available to said server-side apparatus;

a compressing method obtaining portion (211) for obtaining, in response to the received compressing method request signal, the compressing method information stored in said compressing method storing portion to transmit the compressing method information to said terminal-side apparatus;

a compressing method index obtaining portion (208) for obtaining an index of the compressing method from the decompressed waveform data to produce obtained compressing method index;

a recognition engine selection portion (210) for selecting a recognition engine from the obtained compressing method index to produce a selected engine; and

a recognition engine setting portion (210) for setting the selected engine to said recognizing means from stored engines.

10. A speech recognition apparatus as claimed in claim 9, wherein said terminal-side apparatus (100F) further comprises:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis sound to produce formed synthesis sound information and for producing the synthesis sound,

wherein said waveform compressing portion (102B) sends the formed synthesis sound information to said waveform storing portion with it contained in a part of the compressed waveform data,

wherein said server-side apparatus (200F) further comprises;

a synthesis sound information obtaining portion (207) for obtaining the synthesis sound information from the decompressed waveform data to produce obtained synthesis sound information, and

wherein said recognizing means (204C) associates the synthesis sound with the recognition result from the obtained synthesis sound information upon ending the recognition to produce an associated recognition result or the recognition result and the synthesis sound information.

11. A terminal (100) connected to a server apparatus (200) which receives and decompresses compressed waveform data transmitted therefrom, performs recognition processing by using the decompressed waveform data, and produces a recognition result, said terminal and said server apparatus (200) constituting a server-client speech recognition apparatus, said terminal comprising:

a speed detection portion (101) for detecting a speech interval of inputted speech data to produce waveform data of the detected speech interval;

a waveform compression portion (102) for compressing the waveform data at the detected speech interval to produce compressed waveform data; and

a waveform transmission portion (103) for transmitting the compressed waveform data to said server apparatus.

12. A terminal (100A, 100B, 100C, 100D, 100F) connected to a server apparatus (200A) which receives and decompresses compressed waveform data transmitted therefrom, performs recognition processing by using the decompressed waveform data, and produce a recognition result, said terminal (100A, 100B, 100C, 100D, 100F) and said server apparatus constituting a server-client speech recognition apparatus, said terminal comprising:

a waveform and signal reception portion (104, 104A, 104B) for receiving waveform data of an inputted speech and a waveform data re-transmission request signal transmitted from said server apparatus to produce received waveform data and a received waveform data re-transmission request signal;

a speech detection portion (101, 101A) for detecting a speech interval of the received waveform data to produce waveform data at the detected speech interval;

a waveform compressing portion (102, 102A, 102B) for compressing the waveform data at the detected speech interval to produce compressed waveform data;

a waveform storing portion (105) for temporarily storing the compressed waveform data to produce stored waveform data and for producing the stored waveform data in response to the received waveform data re-transmission request signal; and

a waveform transmission portion (103) for transmitting the stored waveform data to the server apparatus.

13. A terminal as claimed in claim 12, further comprising:

a start-point cancel signal transmission portion (106) for receiving a signal at the timing of a start point transmitted from said speech detection portion when said speech detection portion (101A) detects the speech and thereafter cancels the detection to transmit a start-point cancel signal to said server apparatus.

14. A terminal as claimed in claim 12, further comprising:

a compressing method selection portion (110) for selecting an optimum compressing method from the compressing method information to produce compressed compressing method when said waveform and signal reception portion (104A) receives the compressing method information available to said server-side apparatus transmitted from said serve-side apparatus; and

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index,

wherein said waveform compression portion (102B) contains the formed compressing method index in a part of the compressed waveform data.

15. A terminal as claimed in claim 13, further comprising:

a compressing method selection portion (110) for selecting an optimum compressing method from the compressing method information to produce a compressed compressing method when said waveform and signal reception portion (104A) receives the compressing method information available to said server-side apparatus transmitted from said server-side apparatus; and

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index,

wherein said waveform compression portion (102B) contains the formed compressing method index in a part of the compressed waveform data.

16. A terminal as claimed in claim 12, further comprising:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis information to produce formed synthesis sound information and for producing the synthesis sound,

wherein said waveform compressing portion (102A, 102B) contains the formed synthesis sound information in a part of the compressed waveform data.

17. A terminal as claimed in claim 13, further comprising:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis sound to produce formed synthesis sound information and for producing the synthesis sound,

wherein said waveform compression portion (102A, 102B) contains the formed synthesis sound information in a part of the compressed waveform data.

18. A terminal as claimed in claim 14, further comprising:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis sound to produce formed synthesis sound information and for producing the synthesis sound,

wherein said waveform compression portion (102B) contains the formed synthesis sound information in a part of the compressed waveform data.

19. A terminal as claimed in claim 15, further comprising:

a speech synthesizing portion (107) for synthesizing synthesis sound to produce synthesized synthesis sound; and

a synthesis sound information forming portion (108) for forming information on the synthesized synthesis sound to produce formed synthesis sound information and for producing the synthesis sound,

wherein said waveform compression portion (102B) contains the formed synthesis sound information in a part of the compressed waveform data.

20. A terminal as claimed in claim 12, wherein said waveform and signal reception portion (104B) for receiving the inputted waveform data, task information transmitted from a contents side, a waveform data re-transmission request signal transmitted from said server apparatus, and compressing method information available to said server apparatus transmitted from said server apparatus,

wherein said terminal further comprises:

a task information storing portion (113) for storing the received task information to produce stored task information;

a compressing method and task information corresponding table storing portion (111A) for storing a corresponding table between the task information and one or more compressing methods available to the use of a task;

a compressing method selection portion (110A) for selecting an optimum compressing method based on the stored task information, the corresponding table between the task information and the compressing methods transmitted from said compressing method and task corresponding table storing portion, and the received compressing method information to produce a selected compressing method when said waveform and signal reception portion receives the compressing method information available to said server apparatus; and

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index, and

wherein said waveform compression portion (102B) contains the formed compressing method index in a part of the compressed waveform data.

21. A terminal as claimed in claim 13, wherein said waveform and signal reception portion (104B) receives the inputted waveform data, task information transmitted from a contents side, a waveform data re-transmission request signal transmitted from said server apparatus, and compressing method information available to said server apparatus transmitted from said server apparatus,

wherein said terminal further comprises:

a task information storing portion (113) for storing the received task information to produce stored task information;

a compressing method and task information corresponding table storing portion (111A) for storing a corresponding table between the task information and one or more compressing methods available to the use of a task;

a compressing method selection portion (110A) for selecting an optimum compressing method based on the stored task information, the corresponding table between the task information and the compressing method transmitted from said compressing method and task corresponding table storing portion, and the received compressing method information to produce a selected compressing method when said waveform and signal reception portion receives the compressing method information available to said server apparatus; and

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index, and

wherein said waveform compression portion (102B) contains the formed compressing method index in a part of the compressed waveform data.

22. A terminal as claimed in claim 16, wherein said waveform and signal reception portion (104B) receives the inputted waveform data, task information transmitted from a contents side, a waveform data re-transmission request signal transmitted from said server apparatus, and compressing method information available to said server apparatus transmitted from said server apparatus,

wherein said terminal further comprises:

a task information storing portion (113) for storing the received task information to produce stored task information;

a compressing method and task information corresponding table storing portion (111A) for storing a corresponding table between the task information and one or more compressing methods available to the use of a task;

a compressing method selection portion (110A) for selecting an optimum compressing method based on the stored task information, the corresponding table between the task information and the compressing method transmitted from said compressing method and task corresponding table storing portion, and the received compressing method information to produce a selected compressing method when said waveform and signal reception portion receives the compressing method information available to said server apparatus; and

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce formed compressing method index, and

wherein said waveform compression portion (102B) contains the formed compressing method index in a part of the compressed waveform data.

23. A terminal as claimed in claim 17, wherein said waveform and signal reception portion (104B) receives the inputted waveform data, task information transmitted from a contents side, a waveform data re-transmission request signal transmitted from said server apparatus, and compressing method information available to said server apparatus transmitted from said server apparatus,

wherein said terminal further comprises:

a task information storing portion (113) for storing the received task information to produce stored task information;

a compressing method and task information corresponding table storing portion (111A) for storing a corresponding table between the task information and one or more compressing methods available to the use of a task;

a compressing method selection portion (110A) for selecting an optimum compressing method based on the stored task information, the corresponding table between the task information and the compressing method transmitted from said compressing method and task corresponding table storing portion, and the received compressing method information to produce a selected compressing method when said waveform and signal reception portion receives the compressing method information available to said server apparatus; and

a compressing method index forming portion (109) for forming an index of the selected compressing method to produce a formed compressing method index, and

wherein said waveform compression portion (102B) contains the formed compressing method index in a part of the compressed waveform data.

24. A server apparatus (200, 200A, 200B, 200C, 200D, 200E) connected to a terminal (100, 100A, 100B, 100C, 100D, 100E) which detects a speech interval of inputted data, compresses waveform data at the detected speech interval, and transmits the compressed waveform data, said server apparatus and said terminal constituting a server-client speech recognition apparatus, said server apparatus comprising:

a reception portion (201, 201A, 201B, 201C, 201D) for receiving the waveform data transmitted from said terminal to produce the received waveform data;

a waveform decompression portion (202, 202A, 202B) for decompressing the received waveform data to produce decompressed waveform data; and

recognizing means (203, 201A, 204, 204A, 204B, 204C, 205, 205A) for performing recognition processing by using the decompressed waveform data to produce a recognition result.

25. A server apparatus as claimed in claim 24, further comprising:

a waveform data re-transmission request signal transmission portion (206) for transmitting a waveform data re-transmission request signal to said terminal when the reception of the compressed waveform data fails in said reception portion (201A, 201B, 201C, 201D).

26. A server apparatus as claimed in claim 24, wherein when said reception portion (201B, 201C, 201D) receives a start-point cancel signal transmitted when said terminal detects the speech and thereafter the detection is canceled, said recognizing means (204A, 204B, 204C) stops the recognition processing based on the notification from said reception portion.

27. A server apparatus as claimed in claim 24, wherein said reception portion (210C) receives a compressing method request signal transmitted from said terminal to produce a received compressing method request signal, and wherein said server apparatus further comprises:

a compressing method storing portion (212) for storing information on the compressing method available to the server side;

a compressing method obtaining portion (211) for obtaining, in response to the received compressing method request signal, the compressing method information stored in said compressing method storing portion to transmit the compressing method information to said terminal;

a compressing method index obtaining portion (208) for obtaining an index of the compressing method from the decompressed data to produce an obtained compressing method index;

a recognition engine selection portion (209) for selecting a recognition engine from the obtained compressing method index to produce a selected recognition engine; and

a recognition engine setting portion (210) for setting the selected engine from stored engines.

28. A server apparatus as claimed in claim 25, wherein said reception portion (210C) receives a compressing method request signal transmitted from said terminal to produce a received compressing method request signal, and

wherein said server apparatus further comprises:

a compressing method storing portion (212) for storing information on the compressing method available to the server side;

a compressing method obtaining portion (211) for obtaining, in response to the received compressing method request signal, the compressing method information stored in said compressing method storing portion to transmit the compressing method information to said terminal;

a compressing method index obtaining portion (208) for obtaining an index of the compressing method from the decompressed data to produce an obtained compressing method index;

a recognition engine selection portion (209) for selecting a recognition engine from the obtained compressing method index to produce a selected recognition engine; and

a recognition engine setting portion (210) for setting the selected engine from stored engines.

29. A server apparatus as claimed in claim 26, wherein said reception portion (210C) receives a compressing method request signal transmitted from said terminal to produce a received compressing method request signal, and

wherein said server apparatus comprises:

a compressing method storing portion (212) for storing information on the compressing method available to the server side;

a compressing method obtaining portion (211) for obtaining, in response to the received compressing method request signal, the compressing method information stored in said compressing method storing portion to transmit the compressing method information to said terminal;

a compressing method index portion (208) for obtaining an index of the compressing method from the decompressed data to produce an obtained compressing method index;

a recognition engine selection portion (209) for selecting a recognition engine from the obtained compressing method index to produce a selected recognition engine; and

a recognition engine setting portion (210) for setting the selected engine from stored engines.

30. A server apparatus as claimed in claim 24, further comprising:

a synthesis sound information obtaining portion (207) for obtaining synthesis sound information from the waveform data decompressed by said waveform decompression portion (202A) to produce obtained synthesis sound information,

wherein said recognizing block (204B) associates the synthesis sound with the recognition result from the obtained synthesis sound information upon ending the recognition to produce an associated recognition result or the recognition result and the synthesis sound information.

31. A server apparatus as claimed in claim 25, further comprising:

a synthesis sound information obtaining portion (207) for obtaining synthesis sound information from the waveform data decompressed by said decompression portion (202A) to produce obtained synthesis sound information, and

wherein said recognizing portion (204B) associates the synthesis sound with the recognition result from the obtained synthesis sound information upon ending the recognition to produce an associated recognition result or the recognition result and the synthesis sound information.

32. A server apparatus as claimed in claim 26, further comprising:

a synthesis sound information obtaining portion (207) for obtaining synthesis sound information from the waveform data decompressed by said waveform decompression portion (202A) to produce obtained synthesis sound information,

wherein said recognizing portion (204B) associates the synthesis sound with the recognition result from the obtained synthesis sound information upon ending the recognition to produce an associated recognition result or the recognition result and the synthesis sound information.

33. A server apparatus as claimed in claim 27, further comprising:

a synthesis sound information obtaining portion (207) for obtaining synthesis sound information from the waveform data decompressed by said waveform decompression portion (202B) to produce obtained synthesis sound information,

wherein said recognizing portion (204C) associates the synthesis sound with the recognition result from the obtained synthesis sound information upon ending the recognition to produce an associated recognition result or the recognition result and the synthesis sound information.

34. A server apparatus as claimed in claim 24, wherein said reception portion (201D) receives waveform data transmitted from said terminal, a start-point cancel signal, a compressing method request signal, and task information transmitted from a contents side, and

wherein said server apparatus further comprises:

a task information storing portion (213) for storing the task information received by said reception portion to produce stored task information;

a compressing method and task information corresponding table storing portion (212A) for storing the task information and one or more compressing methods available to the use of the task; and

a compressing method obtaining portion (211A) for obtaining available compressing method information from the stored task information and the corresponding table between the task information and the compressing method transmitted from said compressing method and task information corresponding table storing portion to transmit it to said terminal when said reception portion receives the compressing method request signal.

35. A server apparatus as claimed in claim 25, wherein said reception portion (201D) receives the waveform data transmitted from said terminal, a start-point cancel signal, a compressing method request signal, and task information transmitted from a contents side, and

wherein said server apparatus further comprises:

a task information storing portion (213) for storing the task information received by said reception portion to produce stored task information;

a compressing method and task information corresponding table storing portion (212A) for storing the task information and one or more compressing methods available to the use of the task; and

a compressing method obtaining portion (211A) for obtaining available compressing method information from the stored task information and the corresponding table between the task information and the compressing method transmitted from said compressing method and task information corresponding table storing portion to transmit it to said terminal when said reception portion receives the compressing method request signal.

36. A server apparatus as claimed in claim 26, wherein said reception portion (201D) receives the waveform data transmitted from said terminal, a start-point cancel signal, a compressing method request signal, and task information transmitted from a contents side, and

wherein said server apparatus further comprises:

a task information storing portion (213) for storing the task information received by said reception portion to produce stored task information;

a compressing method and task information corresponding table storing portion (212A) for storing the task information and one or more compressing methods available to the use of the task; and

a compressing method obtaining portion (211A) for obtaining available compressing method information from the stored task information and the corresponding table between the task information and the compressing method transmitted from said compressing method and task information corresponding table storing portion to transmit it to said terminal when said reception portion receives the compressing method request signal.

37. A speech recognition method of a server-client system comprising a server apparatus (200) and a terminal (100, 100B), said speech recognition method comprising:

in said terminal (100, 100B),

a step (101) of detecting a speech interval of inputted data;

a step (102) of compressing waveform data of the detected speech interval; and

a step (103) of transmitting the compressed waveform data to said server apparatus, and

in said server apparatus (200, 200B),

a step (201) of receiving the waveform data outputted from said terminal;

a step (202) of decompressing the received waveform data; and

a step (203, 204, 204A, 205) of performing recognition processing by using the decompressed waveform data to produce a recognition result.

39. A speech recognition method as claimed in claim 37, further comprising:

in said terminal (100B),

a step (106) of transmitting a start-point cancel signal to said server apparatus when the speech is detected and thereafter the detection is canceled; and

in said server apparatus (200B),

a step (201B, 204A) of stopping the recognition processing when the start-point cancel signal from said terminal is received.

38. A speech recognition method of a server-client system comprising a server apparatus (200A, 200B, 200C, 200D, 200E, 200F) and a terminal (100A, 100B, 100C, 100D, 100E, 100F), said speech recognition method comprising:

in said terminal (100A, 100B, 100C, 100D, 100F),

a step (104, 104A, 104B) of receiving waveform data of an inputted speech;

a step (101, 101A) of detecting a speech interval of the received waveform data;

a step (102,102B) of compressing the waveform data of the detected speech interval;

a step (103) of temporarily storing the compressed waveform data into a waveform storing portion (105) to transmit the compressed waveform data to said server apparatus;

a step (104, 104A, 104B, 103) of transmitting, to said server apparatus, the waveform data stored in said waveform storing portion (105) on reception of a waveform data re-transmission request signal transmitted from said serer apparatus, and

in said server apparatus (200A, 200B, 200C, 200D, 200E),

a step (201A, 201B, 201C, 201D) of receiving the waveform data outputted from said terminal;

a step (202, 202B) of decompressing the received waveform data;

a step (203, 203A, 204, 204A, 204C, 205, 205A) of performing recognition processing by using the decompressed waveform data to produce a recognition result; and

a step (206) of transmitting the waveform data re-transmission request signal to said terminal when the reception of the compressed waveform data transmitted from said terminal fails.

40. A speech recognition method as claimed in claim 38, further comprising:

in said terminal (100B, 100D, 100C, 100E, 100F),

a step (106) of transmitting a start-point cancel signal to said server apparatus when the speech is detected and thereafter the detection is canceled; and

in said server apparatus (200B, 200D, 200C, 200E, 200F),

a step (201B, 201C, 201D, 204A, 204C) of stopping the recognition processing when the start-point cancel signal from said terminal is received.

41. A speech recognition method as claimed in claim 38, further comprising:

a step (104A) of receiving compressing method information available to the server side which is transmitted from said server apparatus;

a step (110) of selecting an optimum compressing method from the received compressing method information;

a step (109) of forming an index of the selected compressing method;

a step (102B, 105, 103) of compressing the waveform data at the speech interval to transmit it to said server apparatus with the formed compressing method index contained in a part of the compressed waveform data; and

in said server apparatus (200D),

a step (211) of, when the compressing method request signal transmitted from said terminal is received, obtaining the compressing method information stored in a compressing method storing portion (212) for storing the compressing method information available to the server side to transmit the compressing method information to said terminal;

a step (208) of obtaining an index of the compressing method from the decompressed data;

a step (209) of selecting a recognition engine from the obtained compressing method index; and

a step (210) of setting the selected engine from stored engines.

42. A speech recognition method as claimed in claim 40, further comprising:

in said terminal (100D, 100F),

a step (104A) of receiving compressing method information available to the server side transmitted from said server apparatus;

a step (110) of selecting an optimum compressing method from the receiving compressing method information;

a step (109) of forming an index of the selected compressing method; and

a step (102B, 105, 103) of compressing the waveform data of the speech interval to transmit it to said server apparatus with the formed compressing method index contained in a part of the compressed waveform data; and

in said server apparatus (200D, 200F),

a step (211) of, when a compressing method request signal transmitted from said terminal is received, obtaining the compressing method information stored in a compressing method storing portion (212) for storing the compressing method information available to the server side to transmit the compressing method information to said terminal;

a step (208) of obtaining an index of the compressing method from the decompressed data;

a step (209) of selecting an recognition engine from the obtained compressing method index; and

a step (210) of setting the selected engine from stored engines.

43. A speech recognition method as claimed in claim 38, further comprising:

in said terminal (100D, 100E, 100F),

a step (107) of synthesizing a synthesis sound;

a step (108) of forming information on the synthesized synthesis sound to produce the synthesis sound;

a step (102B, 105, 103) of compressing the waveform data at the detected speech interval to transmit it to said server apparatus with the formed synthesis sound information contained in a part of the waveform data; and

in said server apparatus (200D, 200E, 200F),

a step (207) of obtaining the synthesis sound information from the decompressed data,

wherein upon ending the recognition, the synthesis sound is associated with the recognition result from the obtained synthesis sound information to produce an associated recognition result or the recognition result and the synthesis sound information (204C).

44. A speech recognition method as claimed in claim 40, further comprising:

in said terminal (100D, 100E, 100F),

a step (107) of synthesizing a synthesis sound;

a step (108) of forming information on the synthesized synthesis sound to produce the synthesis sound;

a step (102B, 105, 103) of compressing the waveform data at the detected speech interval to transmit it to said server apparatus with the formed synthesis sound information contained in a part of the waveform data; and

in said server apparatus (200D, 200E, 200F),

a step (207) of obtaining the synthesis sound information from the decompressed data,

wherein upon ending the recognition, the synthesis sound is associated with the recognition result from the obtained synthesis sound information to produce an associated recognition result or the recognition result and the synthesis sound information (204C).

45. A speech recognition method as claimed in claim 42, further comprising:

in said terminal (100D, 100F),

a step (107) of synthesizing a synthesis sound;

a step (108) of forming information on the synthesized synthesis sound to produce the synthesis sound;

a step (102B, 105, 103) of compressing the waveform data at the detected speech interval to transmit it to said server apparatus with the formed synthesis sound information contained in a part of the waveform data; and

in said server apparatus (200D, 200F),

a step (207) of obtaining the synthesis sound information from the decompressed data,

wherein upon ending the recognition, the synthesis sound is associated with the recognition result from the obtained synthesis sound information to produce an associated recognition result or the recognition result and the synthesis sound information (204C).

46. A speech recognition method as claimed in claim 38, further comprising:

in said server apparatus (200E),

a step (201D) of receiving the task information transmitted from a contents side to store it in a task information storing portion (213); and

a step (201D, 211A) of, when a compressing method request signal is received, obtaining available compressing method information from a corresponding table (212A) between the task information and compressing methods to transmit it to said terminal.

47. A speech recognition method as claimed in claim 40, further comprising:

in said server apparatus (200E),

a step (201D) of receiving task information transmitted from a contents side to store it in a task information storing portion (213); and

a step (201D, 211A) of, when a compressing method request signal is received, obtaining available compressing method information from a corresponding table (212A) between the task information and compressing methods to transmit it to said terminal.

48. A speech recognition method as claimed in claim 43, further comprising:

in said server apparatus (200E),

a step (201D) of receiving task information transmitted from a contents side to store it in a task information storing portion (213); and

a step (201D, 211A) of, when a compressing method request signal is received, obtaining available compressing method information from a corresponding table (212A) between the task information and compressing methods to transmits it to said terminal.

49. A speech recognition method as claimed in claim 44, further comprising:

in said server apparatus (200E),

a step (201D) of receiving task information transmitted from a contents side to store it in a task information storing portion (213); and

a step (201D, 211A) of, when a compressing method request signal is received, obtaining available compressing method information from a corresponding table (212A) between the task information and compressing methods to transmit it to said terminal.

50. A speech recognition method as claimed in claim 40, further comprising:

in said terminal (100F),

a step (104B) of receiving task information transmitted from a contents side and compressing method information available to the server side transmitted from said server apparatus; and

a step (110A) of, when the compressing method information available to the server side is received, selecting an optimum compressing method based on the task information, a corresponding table between the task information and compressing methods transmitted from a compressing method and task corresponding table storing portion (111A), and the compressing method information available to said server apparatus.

51. A speech recognition method as claimed in claim 40, further comprising:

in said terminal (100F),

a step (104B) of receiving task information transmitted from a contents side and compressing method information available to the server side transmitted from said server apparatus; and

a step (110A) of, when the compressing method information available to the server side is received, selecting an optimum compressing method based on the task information, a corresponding table between the task information and compressing methods transmitted from a compressing method and task corresponding table storing portion (111A), and the compressing method information available to said server apparatus.

52. A speech recognition method as claimed in claim 43, further comprising:

in said terminal (100F),

a step (104B) of receiving task information transmitted from a contents side and compressing method information available to the server side transmitted from said server apparatus; and

a step (110A) of, when the compressing method information available to the server side is received, selecting an optimum compressing method based on the task information, a corresponding table between the task information and compressing methods transmitted from a compressing method and task corresponding table storing portion (111A), and the compressing method information available to said server apparatus.

53. A speech recognition method as claimed in claim 44, further comprising:

in said terminal (100F),

a step (104B) of receiving task information transmitted from a contents side and the compressing method information available to the server side transmitted from said server apparatus; and

a step (110A) of, when the compressing method information available to the server side is received, selecting an optimum compressing method based on the task information, a corresponding table between the task information and compressing methods transmitted from a compressing method and task corresponding table storing portion (111A), and the compressing method information available to said server apparatus.

54. A speech recognition method as claimed in claim 45, further comprising:

in said terminal (100F),

a step (104B) of receiving task information transmitted from a contents side and compressing method information available to the server side transmitted from said server apparatus; and

a step (110A) of, when the compressing method information available to the server side, selecting an optimum compressing method based on the task information, a corresponding table between the task information and compressing methods transmitted from a compressing method and task corresponding table storing portion (111A), and the compressing method information available to said server apparatus.