SOUND SIGNAL RECOGNITION SYSTEM AND SOUND SIGNAL RECOGNITON METHOD, AND DIALOG CONTROL SYSTEM AND DIALOG CONTROL METHOD USING SOUNG SIGNAL RECOGNITION SYSTEM

• US 20030101053A1
• Filed: 02/26/2002
• Published: 05/29/2003
• Est. Priority Date: 11/28/2001
• Status: Active Grant

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Washio, Nobuyuki
• Granted Patent: US 7,177,806 B2
• Time in Patent Office: Days
• US Class Current: 704/226
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 25/78 Detection of presence or absence of voice signals switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10

Abstract

A sound signal including either one or both of a voice signal section and a DTMF signal section is inputted to a sound signal matching/recognizing part via a sound signal input part. The sound signal is divided into a sound signal section by a sound signal analyzing part. A matching part conducts a matching process of a sound signal with reference to both a DTMF signal model and a voice signal model. A recognizing part is provided with a language model including a word dictionary and grammar information, and recognizes a sound signal by using the language model based on the matching result of the matching part.

Claims

1. A sound signal recognition system, comprising:

a sound signal input part for receiving a sound signal including either one selected from a voice signal section and a DTMF signal section or both sections;

a matching part including a voice signal model and a DTMF signal model, for conducting a matching process of the sound signal inputted from the sound signal input part by using both the voice signal model and the DTMF signal model for reference; and

a sound signal recognizing part including a language model, for recognizing the sound signal by using the matching result of the matching part and the language model,

wherein a sound signal recognition process is conducted with respect to the sound signal including either one selected from the voice signal section and the DTMF signal section or both sections.

2. A sound signal recognition system according to claim 1, wherein the sound signal recognizing part selects a better result by comparing the matching result using the voice signal model with the matching result using the DTMF signal model in the matching part for each segment of a sound signal section serving as a recognition unit, the sound signal recognition system further comprising an integrating part for connecting sound signal recognition results selected by the sound signal recognizing part and integrating them as a total sound signal recognition result with respect to all the sections of the input sound signal.

3. A sound signal recognition system according to claim 2, wherein the language model is capable of including a DTMF signal as sound signal recognition vocabulary.

4. A sound signal recognition system according to claim 2, further comprising a guidance part for providing a user who performs sound signal input via the sound signal input part with guidance on whether a specific vocabulary is to be input as sound signal input by a voice or sound signal input by a DTMF signal.

5. A dialog control system including a sound signal recognition system of claim 2, which controls a dialog flow with a user, based on a sound signal recognition result according to the sound signal recognition system.

6. A sound signal recognition system according to claim 1, wherein the language model is capable of including a DTMF signal as sound signal recognition vocabulary.

7. A sound signal recognition system according to claim 6, further comprising a guidance part for providing a user who performs sound signal input via the sound signal input part with guidance on whether a specific vocabulary is to be input as sound signal input by a voice or sound signal input by a DTMF signal.

8. A dialog control system including a sound signal recognition system of claim 6, which controls a dialog flow with a user, based on a sound signal recognition result according to the sound signal recognition system.

9. A sound signal recognition system according to claim 1, further comprising a guidance part for providing a user who performs sound signal input via the sound signal input part with guidance on whether a specific vocabulary is to be input as sound signal input by a voice or sound signal input by a DTMF signal.

10. A sound signal recognition system according to claim 9, wherein upon detecting that a misidentification rate of a sound signal inputted by a voice for a specific vocabulary is high under predetermined conditions, the integrating part notifies the guidance part of instruction information for outputting guidance for asking the user to conduct re-input of the sound signal by a DTMF signal for the specific vocabulary.

11. A dialog control system including a sound signal recognition system of claim 10, which controls a dialog flow with a user, based on a sound signal recognition result according to the sound signal recognition system.

12. A sound signal recognition system according to claim 9, wherein when the integrating part estimates and holds a misidentification rate in the matching result for the sound signal by a voice and a misidentification rate in the matching result for the sound signal by a DTMF signal, and either one of the misidentification rates becomes higher than a predetermined value, the integrating part notifies the guidance part of instruction information for displaying guidance to the user to conduct input by the other sound signal.

13. A dialog control system including a sound signal recognition system of claim 12, which controls a dialog flow with a user, based on a sound signal recognition result according to the sound signal recognition system.

14. A sound signal recognition system according to claim 9, wherein the guidance part has a function of notifying a user of correspondence between a DTMF signal and a vocabulary in advance.

15. A dialog control system including a sound signal recognition system of claim 14, which controls a dialog flow with a user, based on a sound signal recognition result according to the sound signal recognition system.

16. A dialog control system including a sound signal recognition system of claims 9, which controls a dialog flow with a user, based on a sound signal recognition result according to the sound signal recognition system.

17. A dialog control system including a sound signal recognition system of claim 1, which controls a dialog flow with a user, based on a sound signal recognition result according to the sound signal recognition system.

18. A sound signal recognition method, comprising:

inputting a sound signal including either one selected from a voice signal section and a DTMF signal section or both sections;

matching the input sound signal by using both a voice signal model and a DTMF signal model;

recognizing the sound signal by using the matching result and a language model; and

conducting a sound signal recognition process with respect to the sound signal including either one selected from the voice signal section and the DTMF signal section or both sections.

19. A dialog control method including the sound signal recognition method of claim 18, which controls a dialog flow with a user, based on a sound signal recognition result using the sound signal recognition method.

20. A sound signal recognition program for executing a sound signal recognition process with respect to an input sound signal including either one selected from a voice signal section and a DTMF signal section or both sections, the program comprising:

a sound signal input processing operation of inputting a sound signal including either one selected from a voice signal section and a DTMF signal section or both sections;

a matching processing operation of conducting a matching process of the sound signal inputted in the sound signal input processing operation by using both a voice signal model and a DTMF signal model; and

a sound signal recognition processing operation of performing recognition of the sound signal by using a language model based on a matching result in the matching processing operation, the language model including a word dictionary and grammar.