JP4765394B2 - Spoken dialogue device - Google Patents

Description

  The present invention relates to a voice interactive apparatus that exchanges information with a user by voice.

Conventionally, a voice input unit that inputs voice, a voice recognition unit that recognizes input voice, a control unit that changes at least the content displayed on the display unit based on the recognized voice, and the vehicle A speed detecting unit for detecting the speed of the voice, a gain changing unit for changing the gain of the voice input from the voice input unit based on the vehicle speed, and an utterance when the gain of the input voice is changed by the gain changing unit. There is known a voice interaction device provided with a notification means for performing guidance to increase the volume of the voice (for example, see Patent Document 1).
JP 2003-344088 A

  By the way, in this type of voice interactive apparatus, when the user's utterance volume is insufficient, the voice level (or S / N ratio) necessary to realize the desired recognition accuracy as in the above-described conventional technology. ), A voice message such as “Please speak again with a high (or low) voice level” may be output to request the speaker to speak again.

  However, this method of requesting recurrent speech is not effective for a user who feels resistance to speaking in a loud voice, and induces the speech volume to a desired level in order to eliminate the shortage of the user's speech volume. There is a problem that it is difficult.

  Therefore, an object of the present invention is to provide a voice interactive apparatus capable of effectively guiding the user's utterance volume to a desired level in consideration of the user's psychology.

In order to solve the above-described problem, according to one aspect of the present invention, a voice interaction apparatus that performs voice exchange of information with a user, and the voice recognition process when the user's utterance volume has reached an appropriate level In a voice interactive apparatus in which
Voice input means for inputting the utterance from the user;
Utterance volume calculation means for calculating a user's utterance volume based on utterance data input to the voice input means;
Music output means for outputting music;
Output control means for controlling the music output means,
The output control means starts the music output by the music output means or maintains the music output state, and if the user's utterance volume is lower than a predetermined volume according to the calculation result of the user's utterance volume, the music The music output by the music output means in such a manner that the volume of the music output by the output means is increased or the volume of the music output by the music output means is reduced when the user's utterance volume is higher than a predetermined volume. There is provided a voice interactive device characterized by adjusting the volume of the voice.

Further, in the above aspect, further comprising response sound output means for outputting a response sound for voice conversation with the user,
  The output control means controls the music output means and the response sound output means,
  The output control means further starts the output of music by the music output means or maintains the music output state, and further according to the calculation result of the user's utterance volume, the volume of the response sound output by the response sound output means It is good also as adjusting.

In each aspect described above, when speech volume Yu over THE is smaller than the predetermined volume, good as possible to increase the volume of the response Kotaeon you output by response Kotaeon output means. If speech volume is greater than the predetermined volume users, good as possible to reduce the volume of the response Kotaeon you output by response Kotaeon output means. The volume of the music or response sound by the output control means may be adjusted during the user's speech. When used in a passenger compartment, the music output means uses, as the music, music recorded on a music CD or DVD set by the user in the in-vehicle audio apparatus, or predetermined music prepared in advance. May be output via a speaker.

  According to the present invention, it is possible to obtain a voice interactive apparatus capable of effectively guiding a user's utterance volume to a desired level.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  FIG. 1 is a system configuration diagram showing an embodiment of a voice dialogue system in which a voice dialogue apparatus according to the present invention is incorporated. The voice interaction device 10 includes a dialogue control ECU 20, an in-vehicle microphone (microphone) 30 that picks up sound (voice) in the vehicle interior, an amplifier 42, and a speaker 40. The amplifier 42 and the speaker 40 may be the same as those used in the in-vehicle audio system.

  The in-vehicle audio system is connected to the dialogue control ECU 20 via an appropriate bus such as a CAN (controller area network). As will be described in detail later, the dialogue control ECU 20 controls the volume of music output from the in-vehicle audio system via the speaker 40 as necessary.

  The dialog control ECU 20 includes a CPU, a memory, and an A / D (analog-to-digital) converter connected via a bus as its basic configuration. The memory stores programs and data for realizing the functions of the voice interaction apparatus 10 described below.

  The analog sound input to the in-vehicle microphone 30 is subjected to predetermined processing such as amplification processing and noise removal by a microphone amplifier, converted into a digital audio signal by an A / D converter, and sent to the dialogue control ECU 20. The dialogue control ECU 20 extracts a feature amount from the audio signal, and then obtains a recognition result by a matching process using a given acoustic / language model. The dialogue control ECU 20 amplifies a response sound (system sound) corresponding to the recognition result to a predetermined level by the amplifier 42 as necessary, and then outputs the response sound to the vehicle interior via the speaker 40.

  Note that the present invention is not particularly limited by the speech recognition method, and can be applied to speech recognition processing using any software (speech recognition engine) with any hardware configuration.

  FIG. 2 is a flowchart showing main processing realized by the voice interaction system of this embodiment.

  In step 100, the dialogue control ECU 20 detects a stage before the start of the voice dialogue with the user, and outputs a predetermined system voice (guidance voice). This pre-dialogue stage may be when the talk switch is turned on, for example, in a system in which a voice dialog is started when the user turns on the talk switch, and the user needs to make some inquiry from the system side. It may be when. As an example of the latter, the system voice “Where is Ikisaki?” Is output via the speaker 40 when the navigation system asks the user where to go.

  When the pre-dialogue start stage is detected in step 100, the dialog control ECU 20 outputs system music that is not system voice through the speaker 40 at a predetermined volume L0 as step 110. As the system music, an appropriate material is selected as background music (BGM) and prepared in advance in the memory of the dialogue control ECU 20. Thereby, the user can recognize from the auditory that the voice interaction device 10 has entered the state of waiting for the user to input the speech, and can speak with a relaxed feeling because the system music flows behind. (As a result, it is possible to expect an utterance volume that is easy to hear and easy to recognize.) The system music may be downloadable by the user or may be selectable by the user from a plurality of types.

  Regarding this step 110, if the user is already listening to music by the in-vehicle audio system when the pre-dialogue stage is detected, instead of playing the music CD or DVD by the in-vehicle audio system, The system music may be output (that is, switching to system music is performed), or the reproduction of music CD, DVD, or the like by the in-vehicle audio system is continued without outputting the system music. Good. In the latter case, the volume of music output from the in-vehicle audio system may be maintained as it is, or may be reduced or increased to a level equivalent to the initial output volume L0 of system music. Hereinafter, music reproduced by the in-vehicle audio system is referred to as “audio music” in order to distinguish it from the system music.

  When the system music or audio music is being output (in the state of waiting for the user's utterance input), for example, when the user utters by answering "Toyota City Station" (step 120), the utterance volume is increased. It is calculated and it is determined whether or not the speech volume is at an appropriate level (step 130). At this time, the reproduction signal of the system music or the audio music is subtracted from the analog sound (or the digital audio signal by the A / D converter) input to the in-vehicle microphone 30 (that is, the reproduction of the system music or the audio music). By canceling the signal by superimposing its opposite phase), the system music or audio music component contained in the voice input to the in-vehicle microphone 30 is removed. This may be realized by using a technique similar to the barge-in (interrupt utterance) function for detecting the utterance from the user who is outputting the system voice.

  Next, an utterance period is detected from the audio signal from which the components of the system music or audio music are removed in this way. The utterance section is a section that includes the voice portion of the speaker, that is, the speech to be recognized. In the above example, the utterance section is an utterance section related to the utterance “Toyoyoshi”. The detection of the utterance period may be realized by any method, and for example, a method as disclosed in Japanese Patent Application Laid-Open No. 2004-271607 may be used. In this case, the utterance period is detected based on the audio signal from which noise is removed or reduced. This is because the noise of the voice signal is removed or reduced by the filter, so that the amplitude of the voiceless part (speechless part) becomes extremely small, and only the amplitude of the voiced part (voiced part) remains. Is emphasized compared to the non-speech part.

  In step 130, it may be determined whether or not the amount of sound S [dB] in the utterance section has reached a predetermined volume. The voice amount S in the utterance section may be calculated as an average value of the sound volume over all the sections in the utterance section. The calculated value of the volume of speech within the utterance interval may be calculated using appropriate parameters representing the volume and intensity of the sound, such as power and sound pressure. Instead of the average value, the integrated value (integrated value), maximum Values, minimum values, etc. may be used.

  Alternatively, in this step 130, it may be determined whether or not the ratio of the amount of sound S [dB] in the utterance section and the volume N [dB] of the noise, that is, the S / N ratio satisfies a predetermined criterion. . The volume N of the noise may be calculated as an average value of the volume outside the utterance section (that is, the silent part). The average value N of the noise amount may be calculated using appropriate parameters representing the volume / intensity of the sound, such as power and sound pressure, like the average value S of the utterance amount. (Integral value), maximum value / minimum value, etc. may be used.

  If it is determined in step 130 that the user's speech volume has reached an appropriate level, a speech recognition process is performed (step 140), and a speech recognition result is output (step 160). At this time, the dialogue control ECU 20 may output a system voice “Is it all right?” Via the speaker 40 as confirmation of voice recognition, if necessary. In this case, the voice dialogue will be continued, and if the user replies, for example, by saying “Yes”, the utterance related to the response will be continued from the processing of step 120 above. become. Note that the output of the system music started by the processing of step 110 is continued until a series of dialogs with the user is completed. For example, in the previous example, the dialogue control ECU 20 determines that the destination of the user is Toyota City Station, and when the series of dialogues is finished, stops the output of the system music (maintains the output of the audio music as it is). You can use the navigation system to execute route guidance to Toyota Station.

  On the other hand, if it is determined in step 130 that the user's speech volume has not reached an appropriate level, the process proceeds to step 160.

In step 160, a system voice (guidance voice) requesting a recurrent speech is output, and a process of controlling the volume of system music or audio music is performed in order to guide the user's voice volume to a desired level. That is, the dialogue control ECU 20 increases the volume of system music or audio music. At this time, the target volume L _target corresponding to an appropriate level (a fixed value or a variable value corresponding to the volume N of the noise) may be increased at a stroke, but the target volume L is gradually increased each time through the step 160. _It may be increased up to _target . In any case, the volume increase of the system music or audio music may be completed before the next user utters, or may be gradually increased during the user's utterance. It is good. Note that the volume increase of the system music or audio music is realized by controlling the gain of the amplifier 42.

  By increasing the background sound (system music or audio music) of the user who speaks in this way, the user can be guided to speak with a loud sound that does not lose the volume of the music. This expects the Lambard effect that people tend to speak loudly when they are noisy. Even if the system music or audio music is increased, the system music or audio music components are removed by the voice recognition processing as described above, so that the user's speech signal is thus increased in volume. It is not masked by the audio signal of music or audio music and cannot be recognized.

  In this step 160, from the same point of view, the volume of the system sound that requests re-speech may be increased in the same manner. The system voice requesting re-speech may be, for example, the general “Please speak again with a high voice level”, but if there is a recognition candidate with a certain degree of suitability, It may be "Is it true?" The increase in the volume of the system sound is realized by controlling the gain of the amplifier 42.

  In this way, when a re-utterance is requested and there is an utterance from the user again, the processing after step 120 is repeated.

  As described above, according to the present embodiment, when the system music or audio music is played in the surrounding environment (vehicle interior environment) that performs speech for speech recognition, and the volume of the user's speech is insufficient, the volume of the music By increasing the number, it is possible to form an environment in which a user who feels resistance to speaking in a loud voice can speak naturally in a loud voice. Thus, the user's speech volume can be effectively guided to a desired level, the voice recognition accuracy is improved, and the voice dialogue between the voice dialogue device and the user (information exchange associated therewith) is performed with high reliability. ) Can be realized.

  Similarly, for system response sounds (for example, system sounds that require recurrent speech), if the user's utterance volume is insufficient, the loudness can be increased by increasing the volume to speak loudly. It is possible to induce a user who feels resistance to speak effectively and with a loud voice.

  In this embodiment, in contrast to a normal system, from the viewpoint of improving recognition accuracy, background sounds such as audio music are not temporarily reduced in a state of waiting for an utterance from the user. In order to recognize that the user is waiting for an utterance, the background sound may be temporarily reduced. In this case, however, it is desirable to increase the background sound to an appropriate level as described above before the user utters.

  In this embodiment, when the user's speech volume is insufficient, the volume of system music or audio music and / or system sound is controlled so as to induce and increase the user's speech volume to a desired level. On the other hand, when the user's utterance volume is too high, the volume of the system music or audio music and / or the system voice is decreased in order to induce and decrease the user's utterance volume to a desired level. Also good.

  Further, in this embodiment, the dialogue control ECU 20 replaces or additionally displays on the display that the voice input through the in-vehicle microphone 30 is currently “speech recognition”. The system music may be output via the speaker 40. This allows the user to know that the system is currently “speech-recognizing” by listening to the predetermined system music, so it is not necessary to see the display and confirm that the system is currently “speech-recognizing”. . Again, the volume of system music that informs the system that it is currently “speech-recognizing” may be controlled to guide the user's utterance volume to a desired level. From the same point of view, different system music may be output via the speaker 40 according to the state of the voice interaction device 10, for example, a voice input waiting state, a recognition processing state, and the like.

  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, in the above-described embodiment, the present invention relates to the interactive application in the vehicle interior. However, the present invention is not particularly limited to this, and can be applied to the interactive application in other environments. is there.

1 is a system configuration diagram showing an embodiment of a voice interaction system in which a voice interaction device according to the present invention is incorporated. It is a flowchart which shows the main processes implement | achieved by the speech dialogue system of Example 1.

Explanation of symbols

10 Spoken Dialogue Device 20 Dialogue Control ECU
30 Car interior microphone 40 Speaker 42 Amplifier

Claims (6)

In a voice interaction apparatus that performs voice exchange of information with a user, and in which a voice recognition process is performed when the user's utterance volume has reached an appropriate level ,
Voice input means for inputting the utterance from the user;
Utterance volume calculation means for calculating a user's utterance volume based on utterance data input to the voice input means;
Music output means for outputting music;
Output control means for controlling the music output means,
The output control means starts the music output by the music output means or maintains the music output state, and if the user's utterance volume is lower than a predetermined volume according to the calculation result of the user's utterance volume, the music The music output by the music output means in such a manner that the volume of the music output by the output means is increased or the volume of the music output by the music output means is reduced when the user's utterance volume is higher than a predetermined volume. A voice dialogue apparatus characterized by adjusting the volume of the voice. Response sound output means for outputting a response sound for voice conversation with the user,
The output control means controls the music output means and the response sound output means,
The output control means further starts the output of music by the music output means or maintains the music output state, and further according to the calculation result of the user's utterance volume, the volume of the response sound output by the response sound output means The voice interaction device according to claim 1, wherein the voice interaction device is adjusted.   The voice interactive apparatus according to claim 2, wherein when the user's utterance volume is smaller than a predetermined volume, the volume of the response sound output by the response sound output means is increased.   The voice interactive apparatus according to claim 2, wherein when the user's utterance volume is higher than a predetermined volume, the volume of the response sound output by the response sound output unit is reduced.   5. The voice interactive apparatus according to claim 1, wherein the volume of music or response sound by the output control unit is adjusted during a user's speech. In the voice interactive apparatus according to any one of claims 1 to 5, which is used in a passenger compartment.
The music output means outputs, as the music, music recorded on a music CD, DVD, or the like set by the user on the in-vehicle audio device, or predetermined music prepared in advance through a speaker in the vehicle interior. Spoken dialogue device.

Images