Speech dialogue device and speech dialogue method

1. A speech dialogue device comprising:

a speech recognizer to recognize uttered speech;

a response-mode selector to select a mode of a response corresponding to a keyword included in a recognition result of the speech recognizer using a response-mode correspondence table defining a correspondence relationship between the keyword for instructing start of a speech dialogue and the mode of the response;

a dialogue controller to start the speech dialogue when the keyword is included in the recognition result of the speech recognizer, determine a response in accordance with a subsequent recognition result from the speech recognizer, and control a mode of the response in such a manner as to match the mode selected by the response-mode selector; and

a speech output controller to generate speech data on a basis of the response and the mode that are controlled by the dialogue controller and output the speech data to a speaker,

wherein the speech recognizer includes:

a local recognizer to recognize the uttered speech using a local speech recognition dictionary in the speech dialogue device; and

a server recognizer to use and cause an external speech recognition server to recognize the uttered speech and obtain a recognition result,

the response-mode selector selects the local recognizer or the server recognizer that corresponds to the keyword using the response-mode correspondence table defining a correspondence relationship between the keyword and the local recognizer or the server recognizer, and

the dialogue controller switches to the local recognizer or the server recognizer selected by the response-mode selector and determines the response in accordance with a recognition result from the local recognizer or the server recognizer after the switching.

2. The speech dialogue device according to claim 1 , further comprising:

an individual identification unit to identify a user who has operated a button for instructing start of a speech dialogue,

wherein the response-mode selector selects a mode of a response corresponding to the user identified by the individual identification unit using a user response-mode correspondence table defining a correspondence relationship between the user and the mode of the response, and

the dialogue controller starts the speech dialogue when the button is operated, determines a response in accordance with a subsequent recognition result from the speech recognizer and controls a mode of the response in such a manner as to match the mode selected by the response-mode selector.

3. The speech dialogue device according to claim 1 ,

wherein the response-mode selector selects, as the mode of the response, speed, gender, age, volume, or a musical interval of speech of the response.

4. The speech dialogue device according to claim 1 ,

wherein the response-mode selector selects, as the mode of the response, a language of the response or a dialect in each language.

5. The speech dialogue device according to claim 1 ,

wherein the response-mode selector selects, as the mode of the response, an amount of information in the response corresponding to a level of proficiency of a user in the speech dialogue.

6. The speech dialogue device according to claim 1 , further comprising:

a display output controller to generate display data on a basis of the response controlled by the dialogue controller and output the display data to a display,

wherein the response-mode selector selects, as the mode of the response, either one of or both of a speech response from the speaker and a display response on the display.

7. The speech dialogue device according to claim 2 ,

wherein the individual identification unit identifies a user who has uttered the keyword for instructing the start of the speech dialogue, and

the response-mode selector registers the user identified by the individual identification unit and the mode of the response corresponding to the keyword uttered by the user in association with each other in the user response-mode correspondence table.

8. A speech dialogue method in a speech dialogue device, comprising:

recognizing, by a speech recognizer, uttered speech;

selecting, by a response-mode selector, a mode of a response corresponding to a keyword included in a recognition result of the speech recognizer using a response-mode correspondence table defining a correspondence relationship between the keyword for instructing start of a speech dialogue and the mode of the response;

starting, by a dialogue controller, the speech dialogue when the keyword is included in the recognition result of the speech recognizer,

determining, by the dialogue controller, a response in accordance with a subsequent recognition result from the speech recognizer, and

controlling, by the dialogue controller, a mode of the response in such a manner as to match the mode selected by the response-mode selector; and

generating, by a speech output controller, speech data on a basis of the response and the mode that are controlled by the dialogue controller and outputting the speech data to a speaker,

wherein the speech dialogue method further includes:

recognizing, by the speech recognizer, the uttered speech using a local speech recognition dictionary in the speech dialogue device; and

using and causing, by the speech recognizer, an external speech recognition server to recognize the uttered speech and obtaining a recognition result,

selecting, by the response-mode selector, a local recognizer or a server recognizer that corresponds to the keyword using the response-mode correspondence table defining a correspondence relationship between the keyword and the local recognizer or the server recognizer, and

switching, by the dialogue controller, to the local recognizer or the server recognizer selected by the response-mode selector and

determining, by the dialogue controller, the response in accordance with a recognition result from the local recognizer or the server recognizer after the switching step.