Method, apparatus, and program of dialog presentation steps for agents

Patent metadata

Field	Value
Publication number	US-10872609-B2
Application number	US-201716301611-A
Country	US
Kind code	B2
Filing date	May 19, 2017
Priority date	May 20, 2016
Publication date	Dec 22, 2020
Grant date	Dec 22, 2020

Abstract

A dialog method carried out by a dialog system includes an agent that performs a dialog with a user. The dialog method carried out by the dialog system includes a speech receiving step in which the dialog system receives input of a user speech which is a speech of the user, a first presentation step in which when the dialog system cannot obtain any recognition result of a desired level corresponding to the user speech, the dialog system presents a speech which does not include any content words as a first agent speech which is a speech of the agent uttered immediately after the user speech and a second speech step in which the dialog system presents a speech generated or selected not based on the user speech as a second agent speech which is a speech of an agent uttered after uttering the first agent speech.

First claim

What is claimed is:

1. A dialog method carried out by a dialog system comprising an agent that performs a dialog with a user, the dialog method comprising:
a speech receiving step in which the dialog system receives input of a user speech which is a speech of the user;
a first presentation step in which where the dialog system cannot obtain any recognition result of a desired level corresponding to the user speech, the dialog system presents a speech which does not include any content words as a first agent speech which is a speech of the agent uttered immediately after the user speech; and
a second presentation step in which the dialog system presents a speech generated or selected not based on the user speech as a second agent speech, which is a speech of an agent which is different from the agent uttered after uttering the first agent speech, wherein the second agent speech is a speech associated with at least part of at least one of a speech uttered by the user before the user speech to which the dialog system cannot obtain any recognition result of the desired level corresponding and a speech uttered by the agent before the user speech to which the dialog system cannot obtain any recognition result of the desired level corresponding.

2. The dialog method according to claim 1 , wherein the second agent speech is a speech generated or selected based on at least part of at least one of a speech uttered by the user before the user speech and a speech uttered by the agent before the user speech.

3. The dialog method according to claim 1 , wherein the second agent speech is a speech based on a topic of a dialog between the user and the agent before the user speech.

4. The dialog method according to claim 1 , wherein the second agent speech is a speech with a topic different from a topic of a dialog between the user and the agent before the user speech.

5. The dialog method according to claim 2 , wherein the second agent speech is a speech based on a topic of a dialog between the user and the agent before the user speech.

6. The dialog method according to any one of claims 1 to 5 further comprising a recognition result decision step in which the dialog system decides that a recognition result of a desired level corresponding to the user speech is not obtained when an index indicating connectivity of topics between a text of a voice recognition result corresponding to the user speech and a text of a speech uttered by the dialog system before the user speech is less than a predetermined threshold or/and when a degree of deviation of the text of the voice recognition result corresponding to the user speech from an estimated response to the text of the speech uttered by the dialog system before the user speech exceeds a predetermined threshold.

7. A computer-readable non-transitory recording medium that records a program for causing a computer to execute each step of the dialog method according to any one of claims 1 to 5 .

8. A dialog apparatus that obtains a speech uttered by an agent that performs a dialog with a user, the dialog apparatus comprising:
a recognition part that recognizes a user speech which is a speech of the user and obtains a recognition result;
a recognition result decision part that decides whether or not the recognition result of the user speech is a recognition result of a desired level; and
a speech determination part that obtains, where the recognition result of the desired level corresponding to the user speech is not obtained, a speech which does not include any content words as a first agent speech which is a speech of the agent uttered immediately after the user speech, and generates or selects a speech not based on the user speech as a second agent speech, which is a speech of an agent which is different from the agent uttered after uttering the first agent speech, wherein the speech determination part generates or selects a speech associated with at least part of at least one of a speech uttered by the user before the user speech to which the dialog apparatus cannot obtain any recognition result of the desired level corresponding and a speech uttered by the agent before the user speech to which the dialog apparatus cannot obtain any recognition result of the desired level corresponding, as the second agent speech.

9. The dialog apparatus according to claim 8 , wherein the speech determination part generates or selects a speech with a topic different from a topic of a dialog between the user and the agent before the user speech as the second agent speech.

10. A computer-readable non-transitory recording medium that records a program for causing a computer to function as the dialog apparatus according to any one of claims 8 and 9 .

Assignees

• Nippon Telegraph & Telephone
• Univ Osaka

Inventors

• Sugiyama Hiroaki
• Meguro Toyomi
• Yamato Junji
• Yoshikawa Yuichiro
• Ishiguro Hiroshi

Classifications

• G10L15/32

  Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems · CPC title

• G10L13/00

  Speech synthesis; Text to speech systems · CPC title

• G10L2015/225

  Feedback of the input speech · CPC title

• G10L15/1815

  Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning · CPC title

• G10L15/22 Primary

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title