JP2009122210A - Speech processing system, microcomputer for speech processing for implementing same, and electronic equipment with speech output function and speech input function - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform speech recognition, even when speech output is overlapped during a speech input period. <P>SOLUTION: This electronic equipment includes: a speech output processing section 3 for generating and outputting a speech output source waveform; a speaker 15 for outputting the speech output source waveform to outside by performing speech conversion on it; a microphone 17 for taking-in external speech; a correction processing section 5 for performing reverse phase conversion of the waveform similar to the speech output source waveform; and a feed-back processing section 7 for removing the speech output source waveform or its similar waveform, by using the waveform on which the reverse phase conversion is performed by the correction processing section 5, from the waveform of the external speech taken by the feed-back processing section 7. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a voice processing method, a microcomputer that executes this method, and an electronic device in which the microcomputer is mounted.

  In various products, as part of universal design (common design for all people), many products with guidance voice output function that prompts the user with voice guidance from the speaker and informs the current status are proposed and realized. Has been. On the other hand, there is also a product with a user voice input function that enables a required operation or the like by recognizing a microphone input of various voices (user voices) of the user or the like on the product side. In recent years, many products with voice input / output functions having the above-mentioned guidance voice output function and user voice input function have appeared. In this type of product with a voice input / output function, the guidance voice output timing and the user voice input timing do not overlap so that the guidance voice is not inputted as a user voice from a microphone. This will be described. If voice guidance flows while the microcomputer 1 is inputting voice (recognition), only the voice input from the user is normally accepted, but even the guidance voice is input to the inside through the microphone. become. Usually, to prevent this, audio output and audio input are processed alternately. When the voice is being output, no voice input is accepted, and when the voice is being input, no voice is output.

However, in such a product with a voice input / output function, it is necessary for the user side to input the user voice after waiting for the already understood guidance voice to end, and there is a problem of lack of speed. Therefore, in this product with voice input / output function, when the user voice can be accepted even during the guidance voice output, when the guidance voice is input from the microphone together with the user voice, the guidance voice is judged as noise from the product side, The problem is that the product side cannot correctly recognize the user voice due to the noise. For example, Patent Literature 1 can be cited as a technique for eliminating the above-described noise. However, the technique disclosed in Patent Literature 1 is a specific sound even if the noise can be eliminated. It was not possible to extract only the user voice.
JP-A-6-295188

  Therefore, the problem to be solved by the present invention is that voice input such as user voice can be received even during voice output such as guidance voice, and voice such as guidance voice is simultaneously input from a microphone together with voice such as user voice. However, voice processing is performed so that voice such as user voice can be correctly recognized.

  The sound processing method according to the first aspect of the present invention includes a first means for generating and outputting a sound output source waveform, a second means for converting the sound output source waveform into sound and releasing it to the outside, and a third means for capturing external sound. A fourth means for performing reverse phase conversion on the waveform approximated to the sound output source waveform, and the sound output source waveform using the waveform obtained by reverse phase conversion by the fourth means from the waveform of the external sound captured by the third means. And a fifth means for removing a waveform that approximates the sound processing method.

  The voice output source waveform is a voice waveform such as a guidance voice. The voices from the outside are not limited to voices such as user voices but also include voices such as guidance voices emitted from the second means to the outside. The second means is preferably a speaker, for example, and the third means is preferably a microphone, for example.

  In the first aspect of the present invention, the waveform of the external voice captured by the microphone, which is the third means, includes voice such as guidance voice and voice such as user voice. In the fifth means, the waveform approximated to the audio output source waveform is used to remove the waveform approximated to the audio output source waveform, so that the waveform approximated to the audio output source waveform is antiphase-transformed. The waveform approximated to is removed, and the user's voice and the like can be extracted. As a result, the user side can correctly recognize the user voice even when the already-understood guidance voice is input without the user waiting for the end of the guidance voice to improve the speed.

  In a first aspect of the present invention, the second means is a speaker.

  In the first aspect of the present invention, the third means is a microphone.

  In the first preferred embodiment of the present invention, the waveform approximated to the sound output source waveform is a sound output source waveform (internal sound output source waveform), and sound output that is emitted to the outside by the second means and captured by the third means. The approximation coefficient is calculated by comparing with the source waveform (external audio output source waveform), and the internal audio output source waveform is approximated with the approximation coefficient obtained by this calculation.

  An audio processing microcomputer according to a second aspect of the present invention is an audio processing microcomputer mounted on an electronic device including a speaker that converts an audio output source waveform into sound and emits it outside, and a microphone that captures sound from outside. The microcomputer generates a sound output source waveform, converts the sound from the speaker into sound, and performs output processing; a second step that performs reverse phase conversion processing on the waveform approximated to the sound output source waveform; and a microphone And a third step of removing a waveform approximating the audio output source waveform using the waveform obtained by reverse phase conversion from the audio waveform input to the audio waveform. is there.

  In a second preferred embodiment of the present invention, the second step is a voice output source waveform (internal voice output source waveform), which is a waveform approximated to the voice output source waveform, and a voice that is emitted from a speaker and is captured by a microphone. In this step, an approximation coefficient is calculated by comparing with an output source waveform (external audio output source waveform), and an internal audio output source waveform is approximated with an approximation coefficient obtained by this calculation.

  An electronic device according to a third aspect of the present invention includes an audio output source that outputs a guidance audio, a speaker that converts an audio output waveform from the audio output source into an external sound, a microphone that captures audio from the outside, an audio output A first means for performing antiphase conversion on a waveform approximated to a source waveform, and a waveform approximating to the sound output source waveform using a waveform obtained by performing antiphase conversion on the waveform of the external sound captured by the microphone; An electronic apparatus comprising: second means for processing; and third means for recognizing and processing the waveform of the voice subjected to the removal processing as a waveform of a user voice.

  According to a third aspect of the present invention, in the first means, the waveform approximated to the audio output source waveform is an audio output source waveform (internal audio output source waveform) and is emitted from a speaker to the outside and captured by a microphone. The approximation coefficient is calculated by comparing with the audio output source waveform (external audio output source waveform), and the internal audio output source waveform is approximated by the approximation coefficient obtained by this calculation.

  In the third aspect of the present invention, it is possible to set the adjustment mode and the normal mode by a user operation. In the adjustment mode, the first means can be executed, and in the normal mode, the second means can be executed. ing.

  In the third aspect of the present invention, the mode switching state by the user operation is stored in a register and set.

  In the voice processing method of the present invention, for example, even if a user voice is input while guiding the operation of the product with a guidance voice in a product with a voice guidance function and a user voice input function as a voice output source waveform, the user voice is correctly recognized. It becomes possible to cope with the user voice, and the guidance voice output process and the user voice input process can be processed at the same time, thereby realizing a highly immediate operation.

  Hereinafter, a voice processing method according to an embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 1 shows a hardware outline of the sound processing function of the sound processing microcomputer 1 particularly in an electronic device such as a microwave oven incorporating a sound processing microcomputer that implements the sound processing method according to the embodiment of the present invention. Shown in configuration. The electronic device 23 includes a sound processing microcomputer 1, a sound output interface 19, a speaker 15, a sound input interface 21, and a microphone 17. The voice processing microcomputer 1 can output guidance voice and input user voice.

  In the embodiment, if the electronic device is, for example, a microwave oven, the voice processing microcomputer 1 outputs a guidance voice corresponding to the cook start operation button or other button operation from the speaker 15 in response to a cook's cooking start operation button or other button operations. Then, while assisting cooking by the cook, the voice input processing unit 9 may process voice input related to cooking from the cook to the microphone 17 to control the operation of the microwave oven. The description and illustration of such control and other functional elements necessary for the control are omitted in the embodiments.

  The voice processing microcomputer 1 includes a voice output processing unit 3, a correction processing unit 5, a feedback processing unit 7, a voice input processing unit 9, an adjustment mode setting unit 11, a switching unit (MUX) in the voice processing. 13). In the embodiment, each functional element is given a name for convenience of explanation, but this name is given for convenience of explanation, and is not limited to these names.

  The voice output processing unit 3 converts the guidance voice waveform as the voice output source waveform from the speaker 15 through the voice output interface 19 into the guidance voice and outputs it to the user. For example, when the electronic device is a microwave oven, the guidance voice refers to a voice explaining the operation procedure of the microwave oven, the current cooking state, and the like to the user. Of course, the sound output source can be set appropriately according to other electronic devices, electronic devices, portable terminals, information devices, etc., not the microwave oven, and the sound output source waveform is limited to the guidance sound. It is not something.

  Here, in the guidance voice waveform, the guidance voice waveform that is directly input from the voice output processing unit 3 to the feedback processing unit 7 is the internal guidance voice waveform, and the guidance voice that is input to the feedback processing unit 7 through the speaker 15 and the microphone 17. This is called the external guidance voice waveform.

  The feedback processing unit 7 captures the external guidance speech waveform from the microphone 17 and compares the captured external guidance speech waveform with the internal guidance speech waveform input from the speech output processing unit 3 via the adjustment mode setting unit 11. Calculate approximate coefficients of both waveforms. The feedback processing unit 7 inputs the calculated waveform approximation coefficient to the correction processing unit 5.

  The correction processing unit 5 multiplies the guidance speech waveform from the speech output processing unit 3 by the waveform approximation coefficient from the feedback processing unit 7 to obtain an external guidance speech approximate waveform, and performs antiphase conversion processing on the external guidance speech approximate waveform. Then, it is output to the feedback processing unit 7 via the switching unit 13 as an external guidance voice waveform removal waveform.

  The feedback processing unit 7 removes the external guidance speech waveform from the synthesized speech waveform of the external guidance speech waveform input through the microphone 17 and the user speech, using the external guidance speech waveform removal waveform input from the correction processing unit 5. .

  Thereby, even if the external guidance voice and the user voice are simultaneously input into the voice processing microcomputer 1 via the microphone 17, only the user voice is taken into the voice input processing unit 9.

  From the above, it is possible to accept the input of the user voice even while the guidance voice is being output, and conversely, it is possible to play the guidance voice even when the user is inputting the voice.

  Note that, depending on the surrounding environment of the electronic device 23, the external guidance voice waveform input to the feedback processing unit 7 through the speaker 15 and the microphone 17 is different. Therefore, the feedback processing unit 7 uses the waveform approximation coefficient corresponding to the surrounding environment. Is calculated. Then, the calculated waveform approximation coefficient is stored in the correction processing unit 5, and the external guidance voice waveform is obtained from the waveform obtained by synthesizing the waveforms of the user voice and the external guidance voice input to the microphone 17 by the user using the waveform approximation coefficient. So that only the user-voice can be input to the voice input processing unit 9.

  Therefore, in this electronic device 23, setting of the adjustment mode for calculating the waveform approximation coefficient to adjust to the surrounding environment by the user operation, and after the setting is completed, the adjustment mode is canceled and the normal mode is set. To be able to set.

  The setting and cancellation of the adjustment mode will be mainly described with reference to FIG. The adjustment mode setting unit 11 includes a register. This register can change ON (bit 1), which is the adjustment mode setting, by user operation, and OFF (bit 0), which is the normal mode, when the adjustment mode setting is canceled by changing the specific bit.

  The adjustment mode is set by turning ON the register of the adjustment mode setting unit 11, the internal guidance voice waveform is input from the switching unit 13 to the feedback processing unit 7, and the external guidance voice waveform is input from the speaker 15 and the microphone 17 to the feedback processing unit 7. Entered. The feedback processing unit 7 performs a waveform comparison between the input internal guidance speech waveform and the external guidance speech waveform, and calculates a waveform approximation coefficient for approximating the internal guidance speech waveform to the external guidance speech waveform. This waveform approximation coefficient is fed back from the feedback processing unit 7 to the correction processing unit 5.

  When the adjustment mode is cancelled, the correction processing unit 5 and the feedback processing unit 7 are connected by the switching unit 13 to enter the normal mode. An output of the guidance voice is emitted from the speaker 15 from the voice output processing unit 3, and a synthesized voice waveform obtained by synthesizing the user voice waveform and the external guidance voice waveform is input from the microphone 17 to the feedback processing unit 7. The correction processing unit 5 multiplies the waveform approximation coefficient by the internal guidance speech waveform, and further generates an external guidance speech waveform removal waveform by anti-phase conversion, and sends the waveform to the feedback processing unit 7.

  The feedback processing unit 7 removes the external guidance speech waveform from the external guidance speech waveform removal waveform and the synthesized speech waveform, extracts only the user speech, and outputs it to the speech input processing unit 9. Thus, only the user voice is input to the voice input processing unit 9 regardless of the timing at which the user inputs voice to the microphone 17.

  A speech processing method according to another embodiment of the present invention will be described with reference to FIG. 3, parts corresponding to those in FIG. 1 are denoted by the same reference numerals. In the system of this embodiment, the feedback processing unit 7 captures the external guidance speech waveform from the microphone 17 and compares the captured external guidance speech waveform with the internal guidance speech waveform input from the speech output processing unit 3. Calculate approximate coefficients of both waveforms.

  The feedback processing unit 7 multiplies the guidance speech waveform from the speech output processing unit 3 by the above waveform approximation coefficient to obtain an external guidance speech approximate waveform, and performs external phase conversion processing on the external guidance speech approximate waveform to perform external phase conversion processing. Obtained as a waveform for removing the guidance voice waveform. Further, the feedback processing unit 7 may remove the external guidance voice waveform from the synthesized voice waveform of the external guidance voice waveform input through the microphone 17 and the user voice by using the waveform for removing the external guidance voice waveform. Good.

  The audio processing described above is not limited to a microwave oven, and can be applied to any electronic device having an audio output function and an audio input function.

  As described above, in the present embodiment, the user side does not wait for the already-understood guidance voice to end even if the user voice is input from the microphone 17 and the speed is improved. The electronic device can be made to recognize and perform a required operation.

FIG. 1 shows a hardware outline of the sound processing function of the sound processing microcomputer 1 in an electronic device such as a microwave oven incorporating a sound processing microcomputer for implementing the sound processing method according to the embodiment of the present invention. It is a figure shown by a structure. FIG. 2 is a diagram showing a setting flow in the adjustment mode and the normal mode. FIG. 3 is a diagram corresponding to FIG. 1 according to another embodiment of the present invention.

Explanation of symbols

1 Microcomputer for voice processing 1
3 Audio output processing unit 5 Correction processing unit 7 Feedback processing unit 9 Audio input processing unit

Claims (10)

A first means for generating and outputting an audio output source waveform;
A second means for converting the sound output source waveform into sound and releasing it to the outside;
A third means for capturing external audio;
A fourth means for anti-phase converting a waveform approximated to the audio output source waveform;
Fifth means for removing the sound output source waveform or a waveform approximate thereto using a waveform obtained by reverse phase conversion by the fourth means from the waveform of the external sound captured by the third means;
An audio processing method comprising:   The method according to claim 1, wherein the second means is a speaker.   The method according to claim 1 or 2, wherein the third means is a microphone.   The waveform approximated to the sound output source waveform in the fourth means is the sound output source waveform (internal sound output source waveform) and the sound output source waveform released to the outside by the second means and taken in by the third means (external sound). 4. The method according to claim 1, wherein an approximation coefficient is calculated by comparing with an output source waveform), and an internal audio output source waveform is approximated with an approximation coefficient obtained by the calculation. An audio processing microcomputer mounted on an electronic device including a speaker that converts an audio output source waveform into sound and emits the same to outside, and a microphone that captures sound from outside,
The microcomputer is
A first step of generating a sound output source waveform, converting the sound from a speaker to sound, and performing output processing;
A second step of performing anti-phase conversion processing on the waveform approximated to the audio output source waveform;
A third step of removing a waveform approximating to the audio output source waveform using the waveform obtained by antiphase conversion from the audio waveform input to the microphone;
It is possible to execute a microcomputer for voice processing.   In the second step, a waveform approximated to the sound output source waveform is divided into a sound output source waveform (internal sound output source waveform) and a sound output source waveform (external sound output source waveform) emitted from a speaker and captured by a microphone. 6. The sound processing microcomputer according to claim 5, wherein an approximation coefficient is calculated by comparing with (1) and an internal audio output source waveform is approximated with an approximation coefficient obtained by this calculation. An audio output source for outputting guidance audio;
A speaker that converts the sound output waveform from the sound output source into sound and emits it outside;
A microphone that captures audio from outside,
A first means for antiphase converting a waveform approximated to the audio output source waveform;
A second means for removing a waveform approximating the sound output source waveform using a waveform obtained by antiphase conversion by the first means from the waveform of the external sound captured by the microphone;
A third means for recognizing the waveform of the voice subjected to the removal processing as a waveform of the user voice;
An electronic apparatus comprising:   In the first means, the waveform approximated to the audio output source waveform includes an audio output source waveform (internal audio output source waveform), and an audio output source waveform (external audio output source) emitted from a speaker and captured by a microphone. The electronic device according to claim 7, wherein an approximation coefficient is calculated by comparing with the waveform), and the waveform of the internal audio output source waveform is approximated with the approximation coefficient obtained by the calculation.   The adjustment mode and the normal mode can be set by a user operation. The first means can be executed in the adjustment mode, and the second means can be executed in the normal mode. The electronic device described.   The electronic device according to claim 9, wherein the switching state of the mode by a user operation is set in a register.

Images