JP4586566B2 - Spoken dialogue system - Google Patents

Description

  The present invention relates to a voice dialogue system that performs a voice dialogue with a user, and more particularly, to one that includes a dialogue with a communication partner connected via wireless communication.

2. Description of the Related Art Conventionally, a vehicular communication proxy device that performs a call during driving on behalf of an occupant is known (for example, Patent Document 1). The vehicular communication proxy device has a voice interaction function, and an anthropomorphic agent makes a call with a call partner and transmits information acquired from the call partner to the occupant.
JP 2003-32329 A

  In the above prior art, when the agent talks with the other party, the voice of the agent is not output (cannot be heard) to the passenger compartment so as not to misunderstand that the passenger is talking to himself / herself. However, the occupant may be inconvenient for the occupant because the agent may not be able to hear the content spoken to the other party.

  Accordingly, an object of the present invention is to provide a voice dialogue system that is more convenient for the user.

In order to solve the above problems, according to one aspect of the present invention,
In a voice dialogue system that performs voice dialogue with the user,
The first voice used in the dialogue between the communication partner connected via the wireless communication means and the voice dialogue system and the second voice used in the dialogue between the passenger in the vehicle cabin and the voice dialogue system And voice output means for outputting to the vehicle interior,
A voice dialogue system is provided, wherein the timbres of the first and second voices output by the voice output means are different from each other.

  According to this aspect, the occupant can listen to what the voice dialogue system speaks to the communication partner, and whether the voice dialogue system is talking to himself or to the communication partner. It can be easily identified by the difference in timbre, and there is no misunderstanding.

  Moreover, it is preferable that the said audio | voice output means outputs the 3rd audio | voice which the said communicating party emits in a vehicle interior. As a result, the occupant can listen to the voice of the other party, and can understand the contents of the dialogue between the voice conversation system and the other party.

  Moreover, it is preferable that the timbres of the first to third voices output by the voice output means are different from each other. As a result, the voice uttered by the other party can be easily identified by the difference in timbre. In particular, if the other party is also the same voice dialogue system as this system, it is possible that both systems are the same voice, but the occupant can easily identify which voice dialogue system is the voice by the difference in the timbre become.

  In addition, it is preferable to include selection means for selecting a sound to be output from the first to third sounds in response to an input from a passenger instructing whether or not sound output is necessary. As a result, the occupant can arbitrarily select the voice to be output, and can select whether or not to listen to the conversation.

  And determining means for determining whether or not the conversation with the call partner is successful. When the determination means determines that the conversation is not established, the sound output means outputs at least one of the first and third sounds. It is preferable to output. As a result, even when the voice dialogue system cannot communicate, the content of the dialogue can be grasped by listening to the passenger. At this time, even if it is selected not to listen to the conversation, the occupant can grasp the contents of the conversation.

  In addition, it is preferable to include a voice control means for controlling the magnitude of each voice output by the voice output means. As a result, even if the sound volume changes due to different timbres, the sound volume can be adjusted.

  Moreover, it is preferable to provide a voice control means for controlling the height of each voice output by the voice output means. As a result, even if the sound level changes due to different timbres, the height can be adjusted.

  According to the present invention, the convenience for the user can be further increased.

  The best mode for carrying out the present invention will be described below.

[1. Overview of this embodiment]
FIG. 1 is a flowchart showing a schematic operation of the voice interaction system of the present embodiment. The voice interaction system of the present embodiment makes a call on behalf of the occupant based on the request of the occupant such as the driver, acquires information requested by the occupant during the conversation with the other party, and transmits the acquired information to the occupant. It is something to convey. A detailed description of each step in FIG. 1 will be described later.

[2. Details of this embodiment]
FIG. 2 is a diagram showing an overall schematic configuration surrounding the voice interaction system. As shown in FIG. 2, the vehicle 100 includes a control device 10, an input device 20, an output device 30, a storage device 40, and a communication device 50 as a voice interaction system. In addition, it is shown that the facility 200 located at a place away from the vehicle 100 can be accessed via the communication device 50.

  The control device 10 includes a speech recognition unit 11, a speech synthesis unit 12, a dialogue control unit 13, and a speech control unit 14.

  The voice recognition means 11 recognizes the voice spoken by the user who has a voice conversation with the voice dialogue system and converts it into a character string. The input speech is converted into a character string by performing feature extraction, phoneme analysis, word analysis, and syntax analysis on the speech input from the speech input means 22.

  The speech synthesizer 12 generates speech from a character string. The voice synthesizing unit 12 generates a voice output from the voice output unit 31 and a voice transmitted via the mobile phone 51 from the character string by performing character string analysis, prosody control, and voice waveform generation.

  The dialogue control means 13 controls the flow of dialogue by voice between the voice dialogue system and its user. The dialogue control means 13 recognizes the character string converted from the input voice by the voice recognition means 11, and performs control according to the content of the recognized character string. Further, when there is a voice to be output, the dialogue control means 13 generates a character string that is a source of the voice generated by the voice synthesis means 12.

  Such voice interaction technologies include, for example, an automatic response system (IVR: Interactive Voice Response) adopted at customer service by telephone, voice recognition used for route search in car navigation systems, and voice when equipped with communication functions. E-mail reading by synthesis is well known.

  The voice control unit 14 controls the voice waveform (sound wave) generated by the voice synthesis unit 12. By controlling the speech waveform, it is possible to change the “sound volume (sound pressure, sound pressure level)”, “sound pitch”, and “timbre”. “Sound volume” is determined by the amplitude of the speech waveform, “Pitch” is determined by the frequency of the speech waveform, and “Tone” is determined by the shape of the speech waveform. When comparing two sound waveforms (that is, having the same amplitude and frequency) with respect to the volume and pitch of the sound, if the shapes of the sound waveforms are different, the timbres of the sounds are different from each other.

  The input device 20 includes a switch (SW) input unit 21 and a voice input unit 22. Examples of the SW input means 21 include a push SW, a lever SW, and a touch panel display. It is an interface for accepting an operation input by a passenger to the control device 10. The voice input unit 22 collects the voice in the passenger compartment and supplies the collected voice to the voice recognition unit 11. The voice input unit 22 supplies the voice of the communication partner collected through the communication device 50 to the voice recognition unit 11.

  The output device 30 includes audio output means 31 and image output means 32. The sound output means 31 is a so-called speaker that outputs actual sound into the vehicle interior based on the sound waveform described above. The image output means 32 is a display device that provides visual information to the occupant. More specifically, the image output means 32 is a TFT-LCD (thin film transistor liquid crystal display), a self-luminous EL (Electro Luminescence) panel, a VFD. (Fluorescent display tube) and head-up display. The image output means 32 displays, for example, a navigation system screen (a menu screen or a route guidance screen), an anthropomorphic agent image, and a character string recognized and converted in a voice dialogue exchange.

  The storage device 40 stores a conversation scenario 41, facility information 42, selected facility information 43, and map information 44. The conversation scenario 41 is word and syntax dictionary data and sentence examples, and is referred to when controlling the above-described voice dialogue. The facility information 42 is information such as position information and fees related to facilities such as restaurants and parks. The selected facility information 43 is information relating to the facility selected by the passenger, which will be described in detail later. The map information 44 is the position of the vehicle 100 from the GPS, map data, building data, road data, and the like, and is referred to when the navigation system performs a route search or the like.

  The communication device 50 is a device that enables wireless communication with the outside of the vehicle, and more specifically, is a mobile phone 51. The control device 10 makes a call to the facility 200 including the telephone 60 outside the vehicle via the mobile phone 51, and allows the user to interact with the other party.

[3. Operation of this embodiment]
An operation example of the present embodiment will be described with reference to the drawings. FIG. 1 is a flowchart showing a schematic operation of the voice interaction system of the present embodiment. Here, it is assumed that the operation of the spoken dialogue system is expressed by an anthropomorphic agent. Agents interact with the occupants and provide optimal recommendations based on the occupant's preferences, requirements, and learning results (eg, if a nearby restaurant is likely to be preferred by the occupant) I will do it. There are various types of agents such as humans, animals, robots, cartoon characters, and the like, which can be selected according to user preferences. The agent may move on the display or may be like holography.

  Hereinafter, in a situation where the occupant is looking for a set meal while driving, the voice dialogue system (agent) himself acquires information about the set meal and provides the information to the occupant. An example will be described.

  In FIG. 1, first, the occupant selects a call partner (step 100). The other party here is a canteen. The selection of the communication partner will be described in detail with reference to FIG. FIG. 3 is a flowchart for selecting a communication partner. The occupant causes the navigation system to display the facility list on the image output means 32, and selects a canteen from the facility list (step 110). The occupant presses the information acquisition button to select information to be acquired from the menu, for example, “business hours” or “waiting time” (step 120). Here, a plurality of set restaurants may be selected with priorities. The order in which the voice conversation system makes a call (the order in which information is to be acquired) changes according to the priority. However, the order of making a call may be automatically set according to the facility information 42 (distance from the current value, setting of a restaurant fee, etc.) stored in the storage device 40. Then, when the occupant presses the facility selection completion button, selection of the other party is completed (step 130).

  Returning to FIG. 1, the agent calls the selected set restaurant (step 200) and starts a dialogue when the other party answers the call (step 300). FIG. 4 is a flowchart of the voice dialogue between the agent and the other party. The agent recognizes what the other party has spoken and identifies the other party who answered the call. When the agent cannot be identified, the agent identifies the other party by asking the other party, “What if you want to be a restaurant?”

  Next, the agent says, “This is a driver agent system. There are things I want you to tell me on behalf of the driver. Please tell me about“ business hours ”first,” and introduce yourself and ask about the information you want to get ( Step 310). The other party answers “Business hours are □□”. Then, the agent receives the content of the response voice of the other party to the question, and recognizes the voice (step 320). The agent who has performed the speech recognition confirms with the other party that "Are you sure you want the business hours?" (Step 330). When confirming that it is OK, a question about the next information to be acquired is made (step 350; No).

  The agent asked the next question, “How much is the“ waiting time ”? (Step 310). The called party answers “waiting time is △△”. As described above, the agent who recognizes the answer by voice confirms with the other party that "waiting time is xx?" (Steps 320 and 330). Here, when the other party indicates that the confirmation content is wrong or when the other party cannot recognize the voice, the agent tells the other party that he / she wants to wait for a while (step 330; No). ).

  The agent confirms whether or not to listen to the dialogue content because the information cannot be obtained from the occupant, saying “Cannot recognize waiting time. Do you want to listen to the dialogue content?” (Step 340). When the occupant answers “listen”, the agent who has recognized the voice enables the occupant to hear the content of the conversation through the speaker in the vehicle interior. When the agent asks again, “How long is the waiting time?”, The other party answers “The waiting time is △△”. This time, even the passenger in the passenger compartment can hear the voice of the other party, so that the passenger can know the waiting time by directly listening to the voice of the other party. At that time, of course, the agent is also listening to the conversation of the other party, and recognizing the voice for “waiting time ΔΔ”. If the agent is still unable to recognize the voice, apologize to the occupant, “After all, I could not recognize it. Further, if other information that the occupant wants to hear comes out, the occupant may directly talk to the other party and hear the voice that the other party responds. If all the required information is acquired, the agent informs the other party that the call is to be terminated, saying "Thank you for the question. Thank you."

  The agent then informs the passenger of the obtained information with a display on the display, saying, “Business hours are □□, wait time is △△. Do you want to check the next candidate?” Is confirmed (step 400 in FIG. 1). If the occupant replied, “No more confirmation”, the voice is recognized and the information acquisition is completed.

  In this way, the voice that flows from the speaker in the passenger compartment includes the voice that the agent talks to the other party and the voice that talks to the occupant, the voice that the other party talks to the agent and the voice that talks to the occupant, and the conversation is complicated. It is out. Therefore, it may be difficult for passengers, particularly drivers who are driving, to understand who is talking to whom. This is especially true if the language spoken by the agent or the other party is a question format. Therefore, in the voice dialogue system of the present invention, the voices used in the dialogue with the call partner output from the speaker in the vehicle interior, the voices used in the dialogue with the passenger in the vehicle cabin, and the voice emitted by the call partner are different from each other. I try to make it.

  FIG. 5 is a flowchart showing timbre control of dialogue voice. The voice synthesizer 12 generates a voice waveform from the character string of the sentence spoken by the agent (step 510). Based on the generated voice waveform, the voice of the agent is output to the occupant through the speaker in the vehicle compartment and to the call partner via the mobile phone 51 (step 520). The voice control unit 14 adjusts the amplitude, frequency, and shape of the voice waveform to change the timbre of the voice that is spoken to the occupant and the voice that is spoken to the other party (step 530).

  When the voice of the other party is input via the mobile phone 51 (step 540), the other party's voice is output via the speaker in the passenger compartment (step 550), and the voice recognition means 11 Is recognized (step 560). The voice control unit 14 adjusts the amplitude, frequency, and shape of the voice waveform of the voice of the other party of the call so that the timbre differs from the timbre of the voice of the agent performed in step 530 (step 570).

  In this way, the voice dialogue system makes it possible for the user (occupant) to change the timbre of the voice used in the dialogue with the other party that is output from the speaker in the passenger compartment and the voice used in the dialogue with the passenger in the passenger compartment. You can easily identify whether you are talking to a communication partner or talking to a communication partner.

  The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the above-described embodiments, and various modifications and substitutions can be made to the above-described embodiments without departing from the scope of the present invention. Can be added.

  For example, although the case where a call is made from the voice interactive system side has been described above, the case where the call is received is also possible to carry out a voice conversation in the same manner, and the timbre of the voice at that time may be made different from each other.

  If the agent does not make a call when making a call, the call may be made again after a certain period of time. In this case, the agent himself / herself may automatically make a call again after a certain time, or the passenger may be inquired about whether or not to call again. By inquiring the occupant, the occupant can select whether to acquire information even after waiting or to shift to acquiring information of the next facility without waiting.

  Moreover, you may provide the means to estimate the language of the other party. The conversation scenario 41 in the storage device 40 may be stored for each language, and the language spoken by the agent may be switched according to the language of the other party. As a result, if it is recognized that the conversation cannot be performed due to language mismatch, information can be acquired by switching to a language that can be understood by the other party. The language in this case is Japanese or English spoken by humans. However, when the communication partner such as the above-mentioned IVR is a machine, the communication protocol may be switched according to the partner.

It is a flowchart which shows schematic operation | movement of the speech dialogue system of this embodiment. It is the figure which showed the whole schematic structure surrounding a voice dialogue system. It is a flowchart about selection of a communicating party. It is a flowchart about the voice dialogue between an agent and a call partner. It is a flowchart which shows the timbre control of a dialog voice.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Speech recognition means 12 Speech synthesis means 13 Dialogue control means 14 Voice control means 100 Vehicle 200 Facility

Claims (4)

In a voice dialogue system that performs voice dialogue with the user,
The first voice used in the dialogue between the communication partner connected via the wireless communication means and the voice dialogue system and the second voice used in the dialogue between the passenger in the vehicle cabin and the voice dialogue system And a voice output means for outputting the third voice emitted by the communication partner to the passenger compartment,
Speech dialogue system, characterized in that different from each other the sound of the first to third sound output by the sound output unit. The voice dialogue system according to claim 1 , further comprising voice control means for controlling the magnitude of each voice output by the voice output means. The voice dialogue system according to claim 1 , further comprising voice control means for controlling the height of each voice output by the voice output means. In a voice dialogue system that performs voice dialogue with the user,
The first voice used in the dialogue between the communication partner connected via the wireless communication means and the voice dialogue system and the second voice used in the dialogue between the passenger in the vehicle cabin and the voice dialogue system And voice output means for outputting the third voice uttered by the communication partner to the vehicle interior ;
Determining means for determining success or failure of the dialogue with the other party ;
Making the timbres of the first and second voices output by the voice output means different from each other ;
The voice dialogue system is characterized in that the voice output means outputs at least one of the first and third voices when the judgment means judges that the dialogue is not established .

Images