JP2003271194A - Voice interaction device and controlling method thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide more familiar voice interaction by changing the speaking tone and/or voice quality of combined voice of a responding side with respect to the speaking tone and/or voice quality of user's voice. <P>SOLUTION: A recognition model 103 includes a recognition model corresponding to a plurality of types of speaking tones. A voice recognizing part 104 refers to the recognition model 103 to recognize the speaking tone and contents of input voice from a voice inputting part 101. A speaking tone determining part 108 determines the speaking tone of combined voice on the basis of the speaking tone recognized by the voice recognizing part 104. A response sentence preparing part 105 prepares a response sentence to the recognized contents, and a response sentence analyzing part 106 converts the response sentence into reading. A waveform generating part 109 generates a voice waveform on the basis of the speaking tone determined by the speaking tone determining part 108 and the reading of the response sentence obtained by the response sentence analyzing part 106, and a voice outputting part 111 outputs the voice waveform by voice. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice interactive apparatus which recognizes a user's voice and responds with a synthesized voice according to the recognition result, and a control method thereof.

[0002]

2. Description of the Related Art In a conventional voice dialogue system, a mainstream is one that converts a user's voice into character information as a recognition result and outputs a response sentence corresponding to it as a synthetic voice.
It was always possible to obtain only a voice response with a constant tone. In recent years, advances in speech recognition and speech synthesis technologies have made it possible to realize voice dialogue systems in various situations, and a more familiar voice dialogue system that returns expressive responses is expected.

For example, information such as emotion, sex, and age is recognized from the voice of the user, the system side creates a response sentence in a style suitable for the opponent based on the recognition result, and uses the face image of the CG. Suggestions have been made such as responding expressively.

Further, it is said that the accent type of the sentence uttered by the user, the pitch and loudness of the voice and the speaking speed are detected, and the accent type of the synthetic voice, the pitch and loudness of the voice and the speaking speed are adapted to the utterance of the user. Proposals have also been made.

[0005]

[Problems to be Solved by the Invention] In the method of recognizing emotions, sex, age, etc. included in a user's voice and enhancing the expressiveness of a response by the facial expression of a facial image, a telephone or the like having no image display portion is used. Cannot be used with the voice dialogue system. Further, the means for changing the style of the response contents cannot be used in a situation where it is not preferable to change the original contents such as reading an e-mail or reading a literary work.

[0006] Further, in the method of adapting the accent type of the synthesized voice, the pitch and loudness of the voice, and the speaking speed to the accent type of the user, the pitch and loudness of the voice, and the speaking speed, the expression of emotion cannot be performed, and the familiarity becomes familiar. There is a limit in terms of ease.

The present invention has been made in view of the above problems, and it is possible to change the tone and / or the voice quality of the synthesized voice of the response side in accordance with the tone and / or the voice quality of the voice of the user, The purpose is to realize more familiar synthetic speech dialogue.

[0008]

A speech dialogue apparatus according to the present invention for achieving the above object has the following configuration.
That is, a voice dialog device that recognizes an input voice and outputs a response sentence according to a recognition result as a synthesized voice, and a recognition unit that recognizes a tone and content of the input voice, and a recognition unit that recognizes the input voice. The system includes a determining unit that determines the tone of the synthesized voice based on the tone, and a generating unit that generates the synthesized voice of the response sentence with respect to the content of the input voice with the tone determined by the determining unit. Further, a voice interaction device according to another configuration of the present invention for achieving the above object has the following configuration. That is, a voice dialog device for recognizing an input voice and outputting a response sentence according to a recognition result as a synthetic voice, the recognition means recognizing the voice quality and content of the input voice, and the voice quality recognized by the recognition means. Determining means for determining the voice quality of the synthetic voice based on the above, and generating means for generating the synthetic voice of the response sentence with respect to the content of the input voice with the voice quality determined by the determining means.

Further, a control method of a voice interactive apparatus according to the present invention for achieving the above object, recognizes an input voice,
A method of controlling a voice interaction device for outputting a response sentence according to a recognition result as a synthetic voice, comprising: a recognition step of recognizing a tone and content of the input voice; and a tone recognized in the recognition step. The method further comprises a determining step of determining a speech tone of the synthetic voice, and a generating step of generating a synthetic speech of a response sentence with respect to the content of the input voice with the tone determined in the determining step.
Furthermore, a control method for a voice dialogue apparatus according to another configuration of the present invention that achieves the above object is a control method for a voice dialogue apparatus which recognizes an input voice and outputs a response sentence according to the recognition result as a synthetic voice. A recognition step of recognizing the voice quality and content of the input voice; a determination step of determining the voice quality of the synthesized voice based on the voice quality recognized in the recognition step; and a voice quality determined in the determination step, And a generation step of generating a synthetic voice of a response sentence corresponding to the content of the voice.

[0010]

DETAILED DESCRIPTION OF THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

In the first to third embodiments described below, a user recognizes a plurality of types of voices (a plurality of types of voices such as a whispering voice, a bright voice, or a sad voice) in speech recognition by using a recognition model. The speech tone is recognized together with the content of the voice uttered by. Then, in the voice synthesis process,
A voice waveform dictionary for a plurality of types of voices is used to synthesize voices with voices according to the user's voice. Note that the synthesized voice of the response sentence may be adapted to the voice quality of the user in addition to being adapted to the tone of the user (fourth to sixth embodiments).

<First Embodiment> FIG. 1 is a block diagram showing the arrangement of a voice interactive apparatus according to the first embodiment. Figure 1
In the figure, 101 is a voice input unit for inputting the voice of the user. Reference numeral 102 denotes a database, which is used to generate a response sentence corresponding to the recognition result. A recognition model 103 is used to recognize the user's utterance content and its tone. A voice recognition unit 104 recognizes the user's utterance content and tone using the recognition model 103.

Reference numeral 105 denotes a response sentence creating unit, which creates a response sentence from the utterance contents recognized by the voice recognizing unit 104. 10
Reference numeral 6 is a response sentence analysis unit, which analyzes the response sentence created by the response sentence creation unit 105 and gives a reading. A language analysis dictionary 107 is used by the response sentence analysis unit 106 to analyze the response sentence.

Reference numeral 108 is a speech tone determining unit which determines the speech tone used in the speech synthesis to be the closest to the speech tone recognized by the voice recognition unit 104 from the user's voice. 10
Reference numeral 9 denotes a waveform generation unit, which generates a voice waveform with the reading of the response sentence obtained by the response sentence analysis unit 106 and the speech tone determined by the speech tone determination unit 108. Reference numeral 110 is a voice waveform dictionary, which stores voice units of a plurality of types of speech tones. A voice output unit 111 outputs the voice of the response sentence according to the voice waveform generated by the waveform generation unit 109.

The above-described configuration of each unit is realized in the general-purpose personal computer by the CPU executing the program stored in the memory and described in the following flowchart. Also, the database 102, the recognition model 103, the language analysis dictionary 107, and the voice waveform dictionary 110.
Are stored in an external storage device accessible by the personal computer.

FIG. 7 is a flow chart for explaining the procedure of the voice dialogue processing according to the first embodiment.

First, in step S701, the voice input unit 10
1 inputs the user's voice. And step S
At 702, the voice recognition unit 104 recognizes the user's utterance content and its tone. The recognition model 103 for recognition is a plurality of types of Hidden Markov Models (HMM) created by using voice data uttered by a plurality of humans in a plurality of types of tones. Matching with the HMM of each tone is performed, and the tone and utterance content of the best matched model are used as the recognition result.

In step S703, the tone determination unit 1
08 determines the tone of voice synthesis. Here, of the tones prepared in the voice waveform dictionary 110, the tone closest to the tone of the recognition result of the user's voice is selected.

In step S704, the response sentence creating section 105
Creates a response sentence from the utterance content of the user. Response statement, for "Hello", "Hello", questions such as "thank you" there are also those that are well-correspondence on the conversation, such as "You're welcome" to it, "the weather today?" Of the user On the other hand, “September 25, the weather in Yokohama was fine, cloudy in the afternoon, and the probability of precipitation was 0 in the morning.
%, 10% in the afternoon, minimum temperature is 15 degrees, maximum temperature is 28%
There are various things such as those that are created by referring to external data such as "degree." Database 102 of FIG.
Is the source of information needed to create these responses.

In step S705, the response sentence analysis unit 10
6 analyzes the language of the response sentence. For example, in Japanese speech synthesis, a phonetic notation for a kanji / kana mixed sentence is created. In English speech synthesis, phonetic transcription is created from spelling.

In step S706, the waveform generator 109
Is extracted from the speech waveform dictionary corresponding to the speech determined by the speech determination unit 108, and phonemes corresponding to the reading and phonetic symbols are extracted, and prosody is generated and poses are arranged in accordance with the speech. To create. In step S707, the audio output unit 111 outputs audio using the audio waveform generated by the waveform generation unit 109.

As described above, according to the first embodiment, the voice of the tone closest to the tone of the input voice is synthesized according to the tone of the input voice, so that it is more natural with the machine. Voice conversation is realized.

<Second Embodiment> In the first embodiment, the tone of the synthetic voice is determined so that the tone is closest to the tone recognized by the voice recognition unit 104. In the second embodiment, a table that associates the tone of the input voice with the tone of the synthetic voice is prepared, and the tone of the synthetic voice is determined based on the table.

FIG. 2 is a block diagram showing the structure of the voice interactive apparatus according to the second embodiment. The same components as those in the first embodiment are designated by the same reference numerals. Reference numeral 208 denotes a speech tone determination unit, which searches the speech tone correspondence table 211 for the tone of the input voice recognized by the voice recognition unit 104 and determines the tone of the response sentence. In the speech tone correspondence table 211, the correspondence between the speech tone of the user and the speech tone of the response voice (synthesized voice) is registered in advance.

The operation of the voice interactive apparatus according to the second embodiment having the above-mentioned configuration is the same as that of the first embodiment, but in step S703, the speech tone determination unit 208 causes the speech recognition unit 104 to recognize the input voice. The tone associated with the tone is acquired by referring to the tone correspondence table 211,
Determine the tone of the response voice. FIG. 9A shows an example of the tone correspondence table 211. When the user has a normal voice, the system also responds with a normal voice. When the user whispers, the system side also responds with a whisper. When the user has a sad voice, the system side responds with a soft voice, and when the user has an angry voice, the system side responds with a tense voice. This response is in tune with the user, and is to encourage further utterance in the dialogue strategy without making the user feel negative.

In FIGS. 9A and 9B, the user means (input tone) of the input voice, and the system means (tone of) the response voice. Further, in the second embodiment, the correspondence of the tone is set to be synchronized with the user as shown in FIG. 9A. However, as shown in FIG. It is also possible to use a correspondence in which the emotion of the user side is guided in the opposite direction, such as the system side responding with a cheerful voice and the angry voice responding with a whisper.

<Third Embodiment> In the second embodiment,
Although it has been described that the speech tone of the synthetic voice is determined with reference to the speech tone correspondence table 211, a plurality of types of tone tone correspondence tables are prepared in the third embodiment, and the user can select a desired table. To do so.

FIG. 3 is a block diagram showing the structure of the voice interactive apparatus according to the third embodiment. First embodiment or second
The same components as those in the embodiment are designated by the same reference numerals. Reference numeral 311 is a speech tone correspondence selection unit that selects a desired table from a plurality of types of speech tone correspondence tables stored in the speech tone correspondence table 312. Step S80
3, the speech tone determination unit 308 has substantially the same function as the speech length determination unit 208 of the second embodiment, but refers to the speech tone correspondence table selected by the speech tone correspondence selection unit 311 to synthesize the synthesized speech. Determine the tone.

For example, in the speech tone correspondence table, FIG.
Tables such as those shown in FIGS. 9A and 9B are registered, and the speech tone correspondence selection unit 311 selects a desired table according to a user's instruction. The speech tone determination unit 308 refers to the table selected by the speech tone correspondence selection unit 311 to determine the speech tone of the synthetic voice. In addition, FIG.
Although two types of tables are shown in (b), the number of tables is three.
It doesn't matter if there are more than one.

When selecting a table, an interface for displaying a table as shown in FIGS. 9 (a) and 9 (b) on the display device provided in the voice dialog device and allowing the user to select a desired table is provided. To be provided.

As described above, according to the third embodiment, when the response sentence corresponding to the content of the input voice is output, it is possible to perform voice synthesis in a desired tone according to the tone of the input voice. Become.

In the third embodiment, the table is selected, but the tone correspondence table can be arbitrarily created (the tone correspondence table can be edited).
You may do it. For example, in FIG. 9A, an arbitrary combination of the tone of the user and the tone of the system may be set. In this case, one type of synthesized voice may be used for two types of input voice tones (for example, “normal voice” and “whispering voice” on the user side may be “normal”). It may be possible to assign a "voice").

<Fourth Embodiment> In the first to third embodiments, the tone of the synthesized voice is determined according to the tone of the input voice. In the fourth embodiment, the voice quality of the synthesized voice is determined according to the voice quality of the input voice.

FIG. 4 is a block diagram showing the structure of the voice dialogue system according to the fourth embodiment. In FIG. 4, the same components as those of the voice dialogue devices of the first to third embodiments are designated by the same reference numerals.

The recognition model 403 is registered with a model for recognizing the utterance content of the user and a plurality of types of voice qualities. The voice recognition unit 404 refers to the recognition model 403 to recognize the utterance content and voice quality of the user. Voice quality determination unit 408
Determines the voice quality closest to the voice quality recognized from the user's voice as the voice quality used in the synthesized voice of the response sentence to the input voice. The waveform generation unit 409 reads the response sentence obtained by the response sentence analysis unit 106, and the voice quality determination unit 408.
On the basis of the voice quality determined in step 1, a voice waveform is generated by referring to the voice waveform dictionary 410. The voice waveform dictionary 410 stores voice units having a plurality of voice qualities.

FIG. 8 is a flow chart for explaining the voice dialogue processing according to the fourth embodiment.

First, in step S801, the voice input unit 10
1 inputs the user's voice. In step S802, the voice recognition unit 404 refers to the recognition model 403 to recognize the utterance content of the input voice and its voice quality.

The recognition model 403 for recognizing utterance contents and voice quality is a plurality of types of Hidden Markov Models (HMM) created by using human uttered voices grouped by age and sex. The input voice is matched with the HMM of each voice quality, and the voice quality and utterance content of the best matched recognition model are used as the recognition result.

In step S803, the voice quality determination unit 408
Determines the voice quality for voice synthesis based on the voice quality recognized in step S802. In the fourth embodiment, of the voice qualities registered in the voice waveform dictionary 410, the voice recognition unit 40
The one that is closest to the voice quality of the user voice recognized in 4 is selected.

In step S804, the response sentence creating unit 10
Reference numeral 5 creates a response sentence from the user's uttered content recognized by the voice recognition unit 404. Response statement, for "Hello", "Hello", questions such as "thank you" there are also those that are well-correspondence on the conversation, such as "You're welcome" to it, "the weather today?" Of the user On the other hand, such as "September 25, the weather in Yokohama is sunny, cloudy in the afternoon, the probability of precipitation is 0% in the morning, 10% in the afternoon, the minimum temperature is 15 degrees, and the maximum temperature is 28 degrees". There are various things to create by referring to the data of. Database 102 shows the sources of information needed to create these responses.

In step S805, the response sentence analysis unit 106
The linguistic analysis of the response sentence is performed by. In Japanese, create a phonetic notation for mixed kanji and kana sentences. In English, create phonetic transcription from spelling.

In step S806, the waveform generator 409
However, the speech waveform corresponding to the voice quality of the response voice is extracted from the voice waveform dictionary corresponding to the voice quality, and the phoneme corresponding to the voice quality is generated, the prosody is generated and the poses are arranged to create the voice waveform. In step S807, the audio output unit 111 causes the waveform generation unit 4
The voice is output based on the voice waveform generated in 09.

As described above, according to the fourth embodiment, it is possible to utter a response sentence with a voice quality according to the input voice.

<Fifth Embodiment> In the fourth embodiment, the voice quality of the input voice is recognized, and a synthetic voice having a voice quality closest to the recognized voice quality is generated. In the fifth embodiment, the voice quality of the synthetic speech is determined using a voice quality matching table similar to the tone quality matching table described in the second embodiment.

FIG. 5 is a block diagram showing the structure of the voice dialogue system according to the fifth embodiment. The same components as those in the first to fourth embodiments are designated by the same reference numerals. A voice quality determination unit 508 searches the voice quality correspondence table 511 with the voice quality of the input voice recognized by the voice recognition unit 404 and determines the voice quality of the response sentence. In the voice quality correspondence table 511, the correspondence between the voice quality of the user and the voice quality of the response voice (synthesized voice) is registered in advance.

The operation of the voice interactive apparatus according to the fifth embodiment having the above configuration is the same as that of the fourth embodiment, but in step S803, it is associated with the tone of the input voice recognized by the voice recognition unit 104. The obtained voice quality is acquired by referring to the voice quality correspondence table 511, and is determined as the voice quality of the response voice. For example, FIG. 10A is an example of this voice quality correspondence table 511. When the user is an adult male voice, the system responds with an adult female voice, and when the user is an adult female voice, the system side responds with an adult male voice.

In FIGS. 10A and 10B, the user means (input voice quality) of the input voice, and the system means (voice quality) of the response voice.

<Sixth Embodiment> In the fifth embodiment,
Although it has been described that the voice quality of synthesized speech is determined with reference to the voice quality association table 511, in the sixth embodiment, a plurality of types of voice quality association tables are prepared so that the user can select a desired table. .

FIG. 6 is a block diagram showing the arrangement of a voice dialogue system according to the sixth embodiment. The same components as those in the first to fifth embodiments are designated by the same reference numerals. Reference numeral 611 denotes a voice quality correspondence selection unit, which selects a desired table from a plurality of types of voice quality correspondence tables stored in the voice quality correspondence table 612.

In step S803, the voice quality determining unit 6
08 refers to the voice quality correspondence table selected by the voice quality correspondence selection unit 611 to determine the voice quality of the synthesized voice.

For example, the voice quality correspondence table 612 has registered therein the tables as shown in FIGS. 10A and 10B, and the voice quality correspondence selection unit 311 selects a desired table according to a user's instruction. To do. Voice quality determination unit 608
Determines the voice quality of the synthesized voice by referring to the table selected by the voice quality correspondence selection unit 611. In addition,
Although two types of tables are shown in FIG. 10, the number of tables is three.
It doesn't matter if there are more than one.

When selecting a table, an interface for displaying a table as shown in FIGS. 10 (a) and 10 (b) on the display device provided in the voice dialog device and allowing the user to select a desired table is provided. To be provided.

As described above, according to the sixth embodiment, when outputting a response sentence corresponding to the contents of the input voice, it is possible to perform voice synthesis with a desired voice quality according to the voice quality of the input voice.

In the sixth embodiment, the table is selected, but the voice quality correspondence table can be arbitrarily created (the voice quality correspondence table can be edited).
You may do it. For example, in FIG. 10A, an arbitrary combination of the voice quality on the user side and the voice quality on the system side may be set. In this case, the voice quality of the synthesized voice may be set to one for the two voice qualities of the input voice (for example, “adult male voice” and “adult female voice” on the user side may be “children”). "Male voice" may be assigned).

In each of the first to sixth embodiments, the hidden Markov model is used as the recognition model for speech recognition.
Another model such as a neural network may be used.

In the first to third embodiments, whispering voices and emotional voices such as emotions are used as speech tones, but more prosody parts such as DJ tone, narrator tone and reading tone are adopted. Characters that are characteristic of may be used as the type of speech tone.

In the first to sixth embodiments, only the model of the best matching tone and sound quality is adopted as the recognition result. However, when there are a plurality of candidates, the tone of the higher-ranked one is adopted. Alternatively, a combination of the sound quality and the sound quality may be used as the recognition result, and the tone of the response may be determined based on the combination. For example, as a result of recognition of a user's voice, when both a sad voice and an angry voice are in the upper rank, it is possible to associate the user with a whisper.

In the above embodiment, the combination of age and gender is adopted as the voice quality, but the pitch of the base voice of bass, tenor alto, soprano, etc. may be adopted as the type of voice quality. FIG. 10B shows an example of the correspondence.

Further, in the above-described embodiment, the description has been divided into the ones corresponding to the tone (first to third embodiments) and the ones corresponding to the voice quality (fourth to sixth embodiments). And that the voice quality may correspond to the input voice, and in that case any of the first to third embodiments may be combined with any of the fourth to sixth embodiments. Is clear from.

As described above, according to the above-described embodiments, the voice quality and tone of the responding voice are changed in accordance with the voice quality of the user and the tone of the user at that time. For this reason,
There is an effect that it is possible to generate a synthetic voice that does not give a feeling of repulsion and gives a more familiar impression, without using a facial image showing a facial expression and without changing the utterance content.

The object of the present invention is to supply a storage medium having a program code of software for realizing the functions of the above-described embodiments to a system or apparatus, and to supply a computer (or CPU) of the system or apparatus.
It is needless to say that it can be achieved by reading and executing the program code stored in the storage medium.

In this case, the program code itself read from the storage medium realizes the function of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

As a storage medium for supplying the program code, for example, a floppy disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD
-R, magnetic tape, non-volatile memory card, ROM, etc. can be used.

Moreover, not only the functions of the above-described embodiments are realized by executing the program code read by the computer, but also the OS (operating system) running on the computer based on the instructions of the program code. It is needless to say that this also includes a case where the above) performs a part or all of the actual processing and the processing realizes the functions of the above-described embodiments.

Further, after the program code read from the storage medium is written in the memory provided in the function expansion board inserted into the computer or the function expansion unit connected to the computer, based on the instruction of the program code, It goes without saying that a case where the CPU or the like included in the function expansion board or the function expansion unit performs some or all of the actual processing and the processing realizes the functions of the above-described embodiments is also included.

[0066]

As described above, according to the present invention,
By changing the tone and / or voice quality of the synthesized voice of the responding side with respect to the tone and / or voice quality of the user's voice, it becomes possible to provide a more familiar voice dialogue.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a voice interaction device according to a first embodiment.

FIG. 2 is a block diagram showing a configuration of a voice interaction device according to a second embodiment.

FIG. 3 is a block diagram showing a configuration of a voice dialogue device according to a third embodiment.

FIG. 4 is a block diagram showing a configuration of a voice dialogue device according to a fourth embodiment.

FIG. 5 is a block diagram showing a configuration of a voice dialogue device according to a fifth embodiment.

FIG. 6 is a block diagram showing a configuration of a voice dialogue device according to a sixth embodiment.

FIG. 7 is a flowchart illustrating a voice interaction process according to the first embodiment.

FIG. 8 is a flowchart illustrating a voice interaction process according to a fourth embodiment.

FIG. 9 is a diagram showing an example of a correspondence table of user and system tone.

FIG. 10 is a diagram showing an example of a correspondence table of voice qualities of a user and a system.

─────────────────────────────────────────────────── ─── Continued Front Page (51) Int.Cl. ⁷ Identification Code FI Theme Coat (Reference) G10L 15/14 G10L 3/00 R

Claims (18)

[Claims] 1. A voice interactive device for recognizing an input voice and outputting a response sentence according to a recognition result as a synthetic voice, comprising: a recognizing means for recognizing a tone and content of the input voice, and the recognizing means. The system includes: a determining unit that determines the tone of the synthesized voice based on the recognized tone; and a generating unit that generates the synthesized voice of the response sentence with respect to the content of the input voice with the tone determined by the determining unit. A voice dialogue device characterized by the above. 2. A voice waveform dictionary corresponding to a plurality of types of tones is further provided, and the generating unit uses the voice waveform dictionary corresponding to the tone determined by the determining unit to synthesize a previously-synthesized voice. The voice interaction device according to claim 1, wherein the voice interaction device is generated. 3. The voice interactive apparatus according to claim 1, wherein the deciding means decides the tone of the synthetic voice to a tone similar to the tone recognized by the recognizing means. 4. A speech tone table for registering a speech tone of a synthetic speech in association with a speech tone of an input speech, wherein the deciding means is for a synthetic speech corresponding to the speech tone recognized by the recognizing means. The voice interaction apparatus according to claim 1, wherein the voice tone is acquired from the voice tone table. 5. The method according to claim 4, further comprising a selection unit that has a plurality of speech tone tables and that selects a desired table used by the determination unit from the plurality of types of speech tone tables. Spoken dialogue device. 6. The voice interactive apparatus according to claim 4, further comprising an editing unit that edits the speech tone table. 7. The recognizing means recognizes an input voice using a recognition model corresponding to a plurality of types of tones, and determines the tone corresponding to the best matching recognition model as the tone of the input voice. The voice interaction device according to claim 1, wherein 8. A voice interactive device for recognizing an input voice and outputting a response sentence according to a recognition result as a synthetic voice, the recognition means recognizing a voice quality and content of the input voice, and the recognition means. Determining means for determining the voice quality of the synthesized voice based on the determined voice quality, and generating means for generating the synthesized voice of the response sentence with respect to the content of the input voice with the voice quality determined by the determining means. Voice interaction device. 9. A voice waveform dictionary corresponding to a plurality of types of voice qualities is further provided, and the generating means generates a synthesized voice of previous and previous equal parts using the voice waveform dictionary corresponding to the voice quality determined by the determining means. 9. The voice interaction device according to claim 8, wherein: 10. The deciding means determines the voice quality of the synthesized voice,
9. The voice interaction device according to claim 8, wherein the voice quality is determined to be similar to the voice quality recognized by the recognition means. 11. A voice quality table for registering a voice quality of a synthesized voice in association with a voice quality of an input voice, wherein the deciding means has a voice quality for synthetic voice corresponding to the voice quality recognized by the recognizing means. The voice interaction device according to claim 8, wherein the voice interaction device is obtained from a table. 12. The voice dialog according to claim 11, further comprising a selection unit that has a plurality of types of voice quality tables and that selects a desired table to be used by the determination unit from the plurality of types of voice quality tables. apparatus. 13. The voice interactive apparatus according to claim 11, further comprising an editing unit that edits the voice quality table. 14. The recognizing means recognizes an input voice by using a recognition model corresponding to a plurality of types of voice qualities, and determines a voice quality corresponding to the best matching recognition model as a voice quality of the input voice. The voice interaction device according to claim 8. 15. A control method of a voice interaction device for recognizing an input voice and outputting a response sentence according to a recognition result as a synthetic voice, comprising: a recognition step of recognizing a tone and content of the input voice, A determining step of determining the tone of the synthetic voice based on the tone recognized in the recognizing step, and a generating step of generating a synthetic voice of a response sentence with respect to the content of the input voice with the tone determined in the determining step. A method for controlling a voice interaction device, comprising: 16. A method of controlling a voice interaction device for recognizing an input voice and outputting a response sentence according to a recognition result as a synthetic voice, comprising: a recognition step of recognizing a voice quality and content of the input voice; A determining step of determining a voice quality of the synthetic voice based on the voice quality recognized in the step; and a generating step of generating a synthetic voice of a response sentence with respect to the content of the input voice with the voice quality determined in the determining step. And a method for controlling a voice interaction device. 17. A control program for causing a computer to execute the control method according to claim 15 or 16. 18. A storage medium storing a control program for causing a computer to execute the control method according to claim 15 or 16.