JP2007264327A - Bathroom apparatus and voice operation device used therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a bathroom apparatus which improves the recognition rate of voice recognition, even when a bathroom apparatus in operation generates noise, and to provide a voice operation device used therefor. <P>SOLUTION: The voice operation device B used for the bathroom apparatus A includes a noise removing unit 13 which removes noises from a voice signal A/D-converted by the an A/D conversion unit 12, a voice recognition unit 15 which performs recognition processing by performing pattern matching between a noise-removed voice and sound models stored in a sound model storage unit 14, and a voice operation unit 16 which operates corresponding bathroom apparatus, based on the recognition result of the voice recognition unit 15. The noise removing unit 13 is provided with a plurality of noise removing means 13a to 13n which remove the noises generated by corresponding apparatus in a bathroom corresponding to the bathroom apparatus in the bathroom, and the noise removing unit 13 removes noises from the voice signal input from the A/D converting unit 12, by using the noise removing means 13a to 13n, corresponding to the bathroom apparatus in operation detected by an operation state detecting unit 17. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a bathroom device and a voice operation device used therefor.

  2. Description of the Related Art Conventionally, a hot water controller for setting the hot water temperature of a water heater or the temperature and amount of hot water stored in a bathtub, or performing operations such as additional cooking and hot water is installed in the bathroom.

In recent years, in order to realize comfortable bathing, jet jet equipment that generates jets containing bubbles in the bathtub, ventilation heating and drying equipment that performs heating, ventilation, bathroom drying in the bathtub, video or music, etc. Audio-visual equipment for viewing the TV is installed in the bathroom, and a controller for operating these equipment is also installed as a bathroom device (see, for example, Patent Document 1).
Japanese Patent No. 2723881 (paragraph numbers [0011] to [0029] and FIG. 1)

  As described above, there are a plurality of controllers in the bathroom for controlling bathroom equipment such as jet jet equipment, ventilation heating and drying equipment, audiovisual equipment, etc. separately from the hot water controller. Increasingly, there was a problem of requiring a large installation space. Further, as the number of controllers increases, it is necessary to learn how to use each controller, and it takes time to learn how to use them, and it is difficult to use because it operates a plurality of controllers.

  Therefore, in order to control various bathroom devices, when a user issues a command by voice, a voice operation device that recognizes the voice and operates the bathroom device to be operated has been proposed. When a user utters a sound while equipment such as a water heater, jet jet device, and ventilation / heating / drying equipment is operating, the noise generated by the operating equipment is superimposed on the speech, so the recognition of speech recognition The rate declined and the intended operation could not be performed.

  The present invention has been made in view of the above problems, and its object is to provide a bathroom apparatus that improves the recognition rate of voice recognition even when the bathroom equipment is operating and generating noise, and to be used for the bathroom apparatus. It is to provide a voice operation device.

  In order to achieve the above object, a first aspect of the present invention is a noise removal device for removing noise from a plurality of bathroom devices provided in a bathroom, a voice detection unit provided in the bathroom, and a voice detected by the voice detection unit. A pattern matching between the acoustic model storage unit that stores the acoustic model representing the acoustic features of the predetermined voice command, the voice from which noise has been removed by the noise removal unit, and the acoustic model stored in the acoustic model storage unit A voice recognition unit for recognizing the content of the voice by performing, a voice operation unit for operating the corresponding bathroom device according to the content of the voice recognized by the voice recognition unit, and detecting a bathroom device in operation from a plurality of bathroom devices The noise removing unit is provided with a plurality of noise removing means for removing noise generated by the corresponding bathroom device, and the noise removing unit is operated. Using the noise removal means corresponding to the bathroom equipment detected by the state detection unit, and removing noise from speech detected by the speech detection section.

  According to a second aspect of the present invention, in the first aspect of the invention, each noise removing unit includes a noise storage unit in which information representing characteristics of noise generated by the corresponding bathroom device is registered in advance, and is stored in the noise storage unit. Based on the information obtained, noise is removed by taking the difference between the sound detected by the sound detection unit and the noise generated by the bathroom device.

  The invention of claim 3 is characterized in that, in the invention of claim 1, the plurality of noise removing means comprises a plurality of types of noise filters prepared in accordance with the noise of the corresponding bathroom equipment.

  According to a fourth aspect of the present invention, in the first aspect of the present invention, the noise removing unit individually sets the noise filter for removing noise from the input signal and the intensity of noise removal by the noise filter according to the noise of the corresponding bathroom device. Each noise removing means is configured by a noise filter in which the intensity of noise removal is individually set in accordance with the noise of the corresponding bathroom equipment by the filter strength setting means.

  According to a fifth aspect of the present invention, in any one of the first to fourth aspects of the present invention, the acoustic of the signal output when a voice command is input to the corresponding noise removing means corresponding to each of the plurality of noise removing means. An acoustic model representing the feature is prepared, and the speech recognition unit performs speech recognition using an acoustic model corresponding to the noise removal unit used by the noise removal unit.

  The invention of claim 6 is characterized in that, in any one of the inventions of claims 1 to 5, the noise removing means removes noise generated in a state where the corresponding bathroom device is operating alone.

  A seventh aspect of the invention is characterized in that, in any one of the first to fifth aspects of the invention, the noise removing means removes noise generated in a state where a plurality of corresponding bathroom devices are operating simultaneously.

  The invention of claim 8 includes a plurality of bathroom devices provided in the bathroom, a sound detection unit provided in the bathroom, an operating state detection unit that detects an operating bathroom device from the plurality of bathroom devices, and each bathroom device. An acoustic model storage unit that stores acoustic models representing acoustic features of voice commands including noise generated during operation of the corresponding bathroom device, and acoustics stored in the acoustic model storage unit A pattern of an acoustic model selection unit that selects an acoustic model corresponding to the bathroom device detected by the operating state detection unit from among the models, and a pattern of the voice detected by the voice detection unit and the acoustic model selected by the acoustic model selection unit A voice recognition unit that recognizes the content of the voice by performing matching, and a voice operation unit that operates the corresponding bathroom device according to the content of the voice recognized by the voice recognition unit Characterized by comprising Te.

  The invention of claim 9 is characterized in that, in the invention of claim 8, the acoustic model represents an acoustic feature of a voice command including noise generated in a state where the corresponding bathroom device is operating alone. .

  The invention of claim 10 is characterized in that, in the invention of claim 8, the acoustic model represents an acoustic feature of a voice command including noise generated when a plurality of corresponding bathroom devices are operating simultaneously. To do.

  The invention of claim 11 is a voice operation device used in the bathroom device according to any one of claims 1 to 7, wherein the voice detection unit, the noise removal unit, the acoustic model storage unit, the voice recognition unit, the voice An operation unit and an operating state detection unit are provided.

  The invention of claim 12 is a voice operation device used in the bathroom device according to any one of claims 8 to 10, wherein the voice detection unit, the operating state detection unit, the acoustic model storage unit, and the acoustic model selection unit. And a voice recognition unit and a voice operation unit.

  According to invention of Claim 1, the noise removal means which removes the noise which a corresponding bathroom apparatus generate | occur | produces is provided corresponding to each bathroom apparatus, and the bathroom detected by the operation state detection part in the noise removal part Since the noise removal means corresponding to the device is selected and noise is removed from the sound detected by the voice detection unit using the selected noise removal means, the sound signal generated by the bathroom device in operation is removed by the sound signal. Voice recognition can be performed, and the recognition rate of the voice recognition unit can be improved.

  According to the invention of claim 2, the noise removing means is operating by taking the difference between the sound detected by the sound detecting unit and the noise of the bathroom device based on the information stored in the noise storing means. The noise generated by the bathroom equipment can be reliably removed.

  According to invention of Claim 3, the noise which the bathroom apparatus in operation can be reliably removed by using the noise filter corresponding to the bathroom apparatus in operation.

  According to invention of Claim 4, since each noise removal means is comprised by the noise filter and the filter strength setting means which sets individually the intensity | strength of the noise removal by the said noise filter according to bathroom equipment, it is in operation The noise generated by the bathroom equipment can be reliably removed.

  By the way, when a voice command is input to the noise removal unit, the voice command output from the noise removal unit is also distorted, so that voice recognition may not be performed accurately when template matching with an acoustic model is performed. However, according to the invention of claim 5, an acoustic model representing the acoustic characteristics of the signal output when a voice command is input to the corresponding noise removing means is prepared corresponding to each noise removing means. However, since the speech recognition is performed using this acoustic model, the noise removal unit performs the process of removing the noise, so that the speech recognition can be correctly performed even if the voice command itself is distorted. There is.

  According to invention of Claim 6, the noise which a bathroom apparatus which is operating independently can remove noise reliably by a noise removal means.

  According to the seventh aspect of the present invention, noise generated by the plurality of bathroom devices can be reliably removed by the noise removing means while the plurality of bathroom devices are operating simultaneously.

  According to the invention of claim 8, the acoustic model storage unit stores an acoustic model representing acoustic features of a voice command including noise generated during operation of the corresponding bathroom device, and the acoustic model selection unit However, the acoustic model corresponding to the bathroom device detected by the operating state detection unit is selected from the acoustic models stored in the acoustic model storage unit, and the speech recognition unit performs speech recognition using the selected acoustic model. Since this is done, the recognition rate of the voice recognition unit can be improved even when there is noise generated by the bathroom equipment in operation.

  According to the invention of claim 9, the acoustic model stored in the acoustic model storage unit represents an acoustic feature of a voice command including noise generated in a state where the corresponding bathroom device is operating alone. Therefore, it is possible to reliably remove noise generated by bathroom equipment that is operating independently.

  According to the invention of claim 10, the acoustic model stored in the acoustic model storage unit represents an acoustic feature of a voice command including noise generated in a state where a plurality of corresponding bathroom devices are operating. Therefore, the noise which a bathroom apparatus emits in the state where the several bathroom apparatus is operating simultaneously can be removed reliably.

  According to an eleventh aspect of the present invention, the voice detection unit, the noise removal unit, the acoustic model storage unit, the voice recognition unit, the operating state detection unit, and the voice operation unit according to any one of the first to seventh aspects are provided. Thus, it is possible to realize a voice operation device that improves the recognition rate of voice recognition even in a state in which a bathroom device that is in operation generates noise.

  According to the twelfth aspect of the invention, the voice detection unit, the operation state detection unit, the acoustic model storage unit, the acoustic model selection unit, the voice recognition unit, and the voice operation unit according to any one of the eighth to tenth aspects are provided. By providing, it is possible to realize a voice operation device that improves the recognition rate of voice recognition even in a state where a bathroom device that is in operation emits noise.

  Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1)
Embodiment 1 of this invention is demonstrated based on FIGS. 1-3. FIG. 2 is a view schematically showing the configuration of the bathroom 1 to which the bathroom apparatus A according to the present invention is applied. The bathroom 1 is provided with a washing place 3 on the side close to the entrance 2 and on the side far from the entrance 2. There is a bathtub 4 installed. The bathtub 4 is provided with a jet jet generating device (not shown) that generates a jet jet containing bubbles in a state where hot water is filled in the bathtub 4. An air conditioner 6 for performing ventilation, heating or bathroom drying in the bathroom 1 and a lighting fixture 7 for illuminating the bathroom 1 are respectively installed on the ceiling 5 of the bathroom 1. Further, on the horizontal wall 8a of the bathroom 1, a shower facility 9 is provided on the washing room 3 side, and a bathroom TV 10 for viewing television and video is installed on the bathtub 4 side. A controller 11 having a voice recognition function of the present invention is installed on the horizontal wall 8b along the bathtub 4 in the bathroom 1. A hot water heater (not shown) for supplying hot water stored in the bathtub 4 or hot water discharged from the shower facility 9 is installed outside the bathroom 1.

  FIG. 3 is an external view of the controller 11, which includes a body 20 having a front surface formed in a rectangular shape, and is disposed on the horizontal wall 8 b with the front surface of the body 20 exposed on the surface of the horizontal wall 8 b. . On the front surface of the body 20, a liquid crystal monitor 21 that indicates an operating state of the water heater, a set value of the hot water temperature, and the like, a light emitting diode LED for operation display, a microphone MC, and a speaker SP are arranged. On the front side of the body 20, on the right side of the liquid crystal monitor 21, a menu button 22a for switching the operation mode to the setting mode, a confirmation button 22b for determining setting items and setting values, and the previous operation state A return button 22c that is operated to return to the position is arranged side by side, and characters indicating operation contents such as “menu”, “confirm”, and “return” are displayed on the front of each operation button 22a to 22c. Yes. Further, below the operation buttons 22a to 22c, there are arranged cursor buttons 23a to 23d for up / down / right / left and right for performing operations such as selection of setting items and change of setting values in the setting mode. In addition, on the front side of the body 20, on the left side of the liquid crystal monitor 21, for example, priority is given to the operation by the controller 11 on the bathroom side over the operation from a controller (not shown) placed separately in the kitchen or the like. A button 22d, a follow-up button 22e that is pushed to warm the hot water stored in the bathtub 4, a bath automatic button 22f that is pushed to hold a predetermined amount of hot water in the bathtub 4, and a kitchen Call buttons 22g that are operated when an interphone call is performed with a controller placed separately on the controller are arranged side by side. For example, "priority", "chase", " Characters indicating operation details such as “Buro automatic” and “call” are displayed.

  FIG. 1 is a block diagram of the main part of the controller 11 and shows only a configuration for performing voice command recognition processing and operating bathroom equipment. That is, FIG. 1 shows a schematic configuration of a voice operation device B applied to the bathroom device A, and the voice operation device B is provided in the bathroom 1 and converts a voice inputted into an electric signal into an electric signal, An A / D converter 12 that converts an electrical signal (analog signal) input from the microphone MC into a digital signal; and a noise remover 13 that removes noise included in the audio signal input from the A / D converter 12; An acoustic model storage unit 14 in which an acoustic model representing an acoustic feature of a predetermined voice command is registered in advance, an audio signal from which noise has been removed by the noise removal unit 13, and an acoustic model registered in the acoustic model storage unit 14 The voice recognition unit 15 for recognizing the content of the voice command by pattern matching, and the corresponding bathroom equipment (air conditioner 6, lighting fixture 7, bathroom tester) based on the recognition result of the voice recognition unit 15. A voice operation unit 16 for operating the bathroom 10, a water heater 18, a jet jet generator 19 and the like, and an operation state detection unit 17 for detecting a bathroom device currently in operation from a plurality of bathroom devices installed in the bathroom 1. I have.

  The microphone MC is used to convert the voice uttered by the user into an electrical signal and output it to the A / D converter 12. Such a microphone MC may be used in a usage state or usage environment such as a capacitor microphone or a piezoelectric microphone. An appropriate one can be used accordingly.

  The acoustic model storage unit 14 stores an acoustic model necessary for voice command recognition processing used to operate the bathroom equipment. As the acoustic model, for example, an acoustic unit of a small unit (phoneme) of human utterance is stored. A hidden Markov model (HMM) representing features is used.

  The speech recognition unit 15 performs pattern matching processing between the speech signal from which noise has been removed by the noise removal unit 13 and the acoustic model stored in advance in the acoustic model storage unit 14, thereby processing the content of the input speech signal. recognize.

  Here, a plurality of bathroom devices are installed in the bathroom 1, but basic bathrooms installed in the bathroom 1 (basic bathtubs such as the air conditioner 6, the water heater 18, and the jet jet generator 19). Specific equipment is used for the equipment and optional equipment such as the lighting fixture 7 and the bathroom television 10. Among these devices, the bathroom TV 10 and the audio device do not emit a fixed sound, but the bathroom devices such as the air conditioner 6, the lighting fixture 7, the water heater 18, and the jet jet generator 19 are in operation. Therefore, the noise generated by these devices can be measured in advance. Therefore, in the present embodiment, the noise removing unit 13 is provided with a plurality of noise removing means 13a, 13b... 13n for removing noise generated by the corresponding bathroom device, corresponding to each bathroom device. Then, based on the detection signal input from the operation state detection part 17, the noise removal means 13a ... corresponding to the bathroom apparatus in operation is selected. The noise removing means 13a selected at this time removes noise from the audio signal including noise input from the A / D converter 12, so that noise generated by the bathroom equipment currently in operation is reliably removed. Even when the bathroom equipment in operation is generating noise, the recognition rate of the voice recognition by the voice recognition unit 15 can be improved and a desired operation can be performed.

  Each of the noise removing means 13a may be configured to remove noise generated when the corresponding bathroom device is operating alone, but a plurality of corresponding bathroom devices are operating simultaneously. It may be configured to remove the noise generated in the case, the means for removing the noise generated when the bathroom device operates alone, and the noise generated when a plurality of bathroom devices operate simultaneously It is also preferable to have both means. During bathing, any one of the bathroom devices may be operated alone, or a plurality of bathroom devices may be used at the same time. Therefore, noise generated when the bathroom device operates alone as the noise removing means 13a. If it is provided with a means for removing noise and means for removing noise generated when a plurality of bathroom devices operate simultaneously, noise generated by the bathroom devices can be reliably removed and the recognition rate of voice recognition can be improved. Can do.

  In addition, assuming a plurality of scenes (scenes) in which a bather takes a bath, a combination of bathroom devices used for each scene is set in the controller 11, and when any scene is selected using the controller 11. It is considered that a plurality of bathroom devices corresponding to the selected scene are used in combination. For example, in a “relaxation scene” for making the bather feel comfortable, the air conditioner 6 is cooled to prevent the bather from climbing so that the jet jet flow generator 19 can be bathed for a long time. I have to. Therefore, noise removing means for removing noise generated while the jet jet flow generator 19 and the air conditioner 6 are simultaneously operated in the “relaxation scene” is prepared, and the controller 11 is used for “relaxation”. When the operation is performed in the “scene”, the noise removing unit 13 selects the noise removing unit corresponding to the “relaxation scene” based on the detection result of the operation state detecting unit 17, and the selected noise removing unit is used to select A By removing noise from the sound signal including noise input from the / D conversion unit 12, it is possible to reliably remove noise generated by the bathroom equipment currently in operation. Also, in other scenes, a noise removal means for removing noise of bathroom equipment operated in each scene is prepared, and when any scene is selected and operated using the controller 11, it is selected. The noise removing unit 13 selects the noise removing unit corresponding to the selected scene and removes the noise using the selected noise removing unit, so that the noise generated by the bathroom equipment in operation can be reliably removed. .

  Furthermore, if noise removal processing is performed on the input voice by the noise removing unit 13, distortion may occur in the processed voice, and pattern matching with the acoustic model may not be performed correctly. Therefore, acoustic models representing acoustic features of signals output when a predetermined voice command is input to each noise removing unit 13a ... are prepared in advance, and these acoustic models are associated with each noise removing unit 13a ... The acoustic model storage unit 14 stores the acoustic model corresponding to the bathroom device in operation from the acoustic model storage unit 14 based on the detection signal input from the operating state detection unit 17. It may be used for reading and voice recognition, and the recognition accuracy of voice recognition can be further improved.

(Embodiment 2)
A second embodiment of the present invention will be described with reference to FIG. In the present embodiment, in the bathroom device A and the voice operation device B used in the bathroom device A described in the first embodiment, the noise removal means 13a provided in the noise removal unit 13 is replaced with a noise removal means that removes noise using a spectrum subtraction method. It consists of. In addition, since structures other than the noise removal part 13 are the same as that of the bathroom apparatus A or the voice operation apparatus B demonstrated in Embodiment 1, the same code | symbol is attached | subjected to a common component and illustration and description are abbreviate | omitted.

  FIG. 4 is a block diagram showing the configuration of the noise removing unit 13. The noise removing unit 13 uses a plurality of noise removing units 13a... And a noise removing unit used based on detection signals input from the operating state detecting unit 17. And switching means 30 for switching 13a... Since each noise removing means 13a has the same configuration, the noise removing means 13a will be described as an example. The noise removal means 13a is a memory (noise storage means) 31 in which information representing the characteristics of noise generated by the corresponding bathroom equipment, that is, a noise power spectrum is registered in advance, and the A / D conversion section 12 via the switching means 30. The difference between the power spectrum obtained by Fourier transform of the input audio signal and the noise power spectrum stored in the memory 31 is calculated, and noise is removed by inverse Fourier transform of the signal after the difference is obtained. And an arithmetic unit 32 for restoring the voice signal.

  That is, the noise removing means 13a ... removes noise by the spectral subtraction method, and the spectral subtraction method is very effective for noise removal when stationary noise is added to speech. The noise removing unit 13 of the present embodiment removes the noise of the bathroom equipment installed in the bathroom 1, and what kind of equipment is used as the bathroom equipment and what kind of noise the equipment generates. Since it can be grasped in advance, noise removal means 13a... Corresponding to various devices expected to be used in the bathroom 1 are prepared, and noise removal using the spectral subtraction method is effective to effectively superimpose noise superimposed on speech. Can be removed.

(Embodiment 3)
Embodiment 3 of the present invention will be described with reference to FIG. In the present embodiment, in the bathroom device A and the voice operation device B used in the bathroom device A described in the first embodiment, the noise removing unit 13 is configured by a noise removing unit that removes noise using a noise filter such as a Kalman filter. . In addition, since structures other than the noise removal part 13 are the same as that of the bathroom apparatus A or the voice operation apparatus B demonstrated in Embodiment 1, the same code | symbol is attached | subjected to a common component and illustration and description are abbreviate | omitted.

  FIG. 5 is a principal block diagram showing the configuration of the noise removing unit 13. The noise removing unit 13 is used based on a plurality of noise removing means 13 a... And detection signals input from the operating state detecting unit 17. It is comprised with the switching means 30 which switches the noise removal means 13a .... Each of the plurality of noise removing means 13a is composed of a plurality of types of noise filters prepared in accordance with the noise of the corresponding bathroom equipment, and uses a plurality of types of Kalman filters 33 with parameters individually set as the noise filters. The Kalman filter 33 is a conventionally well-known filter, and a detailed description thereof will be omitted.

  In each noise removing means 13a ..., parameters representing noise and voice dispersion and correlation are set according to the noise of the corresponding bathroom equipment, and the coefficient of the Kalman filter is determined based on these parameters. By applying such a Kalman filter to the input signal, it is possible to remove the noise of the corresponding bathroom device and take out only the voice, so that the recognition rate of voice recognition can be improved.

(Embodiment 4)
A fourth embodiment of the present invention will be described with reference to FIG. In the third embodiment described above, the plurality of noise removing means 13a are configured by a plurality of types of noise filters prepared in accordance with the noise of the corresponding bathroom equipment, whereas in the present embodiment, the noise removing unit 13 includes one kind of Kalman filter 34 in which parameters representing noise and voice dispersion and correlation are set to predetermined values, and the intensity of noise removal by the Kalman filter 34 is set in advance according to the noise of the bathroom equipment in operation. The set strength is set. In addition, since structures other than the noise removal part 13 are the same as that of the above-mentioned Embodiment 1, the same code | symbol is attached | subjected to a common component and the description is abbreviate | omitted.

  FIG. 6 is a principal block diagram showing the configuration of the noise removing unit 13. The noise removing unit 13 has a set value indicating the intensity of noise removal for each bathroom device installed in the Kalman filter 34 and the bathroom 1. The filter strength setting unit 35 is configured in advance to set the noise removal strength by the Kalman filter 34 based on the detection signal input from the operating state detection unit 17 and the set value. Here, the noise removal means 13a described in the first embodiment is configured by a noise filter in which the noise removal strength is individually set in accordance with the noise of the corresponding bathroom device by the filter strength setting unit 35.

  By the way, although each bathroom apparatus installed in the bathroom 1 generates different noises depending on the type of the apparatus, the filter strength setting unit 35 has a noise removal by the Kalman filter 34 corresponding to each bathroom apparatus. A setting value indicating the intensity of the image is registered in advance. Thus, in the noise removing unit 13, when the detection signal is input from the operating state detecting unit 17, the filter strength setting unit 35 sets the noise removing strength by the Kalman filter 34 according to the bathroom equipment currently in operation. In addition, the Kalman filter 34 in which the intensity of noise removal is set according to the bathroom equipment in operation is applied to the input signal. Here, the Kalman filter 34 acts on the audio signal input from the A / D converter 12 and removes the noise of the corresponding bathroom device from the input signal. Since intensity | strength is switched to a predetermined setting value, the noise which the bathroom apparatus in operation can generate | occur | produce can be removed reliably.

(Embodiment 5)
Embodiment 5 of the present invention will be described with reference to FIG. The voice operation device B of the present embodiment is applied to the bathroom device A described in the first embodiment, and operates a bathroom device installed in the bathroom 1 by voice, and performs voice recognition processing of the voice operation device B. Since the parts other than the parts to be performed are the same as those in the first embodiment, common constituent elements are denoted by the same reference numerals and description thereof is omitted.

  In the first to fourth embodiments described above, noise removing means for removing noise generated by each bathroom device is prepared corresponding to each bathroom device, and noise is removed using the noise removing device corresponding to the bathroom device in operation. In the present embodiment, a plurality of acoustic models representing the acoustic features of voice commands including noise generated by the corresponding bathroom equipment are prepared and operated in this embodiment. Voice recognition is performed by selecting an acoustic model corresponding to the bathroom equipment inside.

  FIG. 7 is a block diagram of a main part of the controller 11 and shows only a configuration for performing voice command recognition processing and operating bathroom equipment. That is, FIG. 7 shows a schematic configuration of a voice operation device B applied to the bathroom device A, and the voice operation device B is provided in the bathroom 1 and converts a voice inputted into an electric signal into an electric signal, An A / D converter 12 that converts an electrical signal (analog signal) input from the microphone MC into a digital signal, and a plurality of bathroom devices (an air conditioner 6, a lighting fixture 7, a bathroom TV 10, installed in the bathroom 1) Including an operating state detection unit 17 that detects a bathroom device currently in operation from a water heater 18, a jet jet generator 19, and the like, and a noise provided by the corresponding bathroom device. Acoustic model storage units 14a, 14b,..., 14n for storing acoustic models representing the acoustic features of the voice commands, and acoustic models corresponding to the bathroom devices in operation based on the detection results of the operation state detection unit 17. Performing pattern matching between the acoustic model selection unit 36 for selecting the storage unit 14a, and the acoustic model registered in the selected acoustic model storage unit 14a and the audio signal input from the A / D conversion unit 12. The voice recognition unit 15 for recognizing the content of the voice command and the voice operation unit 16 for operating the corresponding bathroom device based on the recognition result of the voice recognition unit 15 are provided.

  Here, a plurality of bathroom devices are installed in the bathroom 1, but the bathroom devices such as the air conditioner 6, the lighting fixture 7, the water heater 18, and the jet jet generator 19 are almost constant during operation. Generates noise of the frequency or volume. Therefore, in the present embodiment, for bathroom equipment that emits stationary noise, acoustic models that represent the acoustic features of voice commands including noise generated by the bathroom equipment are created in advance, and these acoustic models are assigned to each bathroom equipment. The acoustic model storage units 14a, 14b,. Since the controller 11 controls the operation of each bathroom device, the operation state of each bathroom device can be grasped, and the acoustic model is adjusted in accordance with the bathroom device in operation detected by the operation state detection unit 17. The selection unit 36 can select the corresponding acoustic model storage unit 14a. Therefore, the voice recognition unit 15 can perform voice recognition using an acoustic model of a voice command including noise generated by an operating bathroom device. Even when bathroom device noise exists, the voice recognition recognition accuracy is high. Can be improved.

  The acoustic model storage units 14a, 14b,..., 14n may store acoustic models that represent the acoustic features of voice commands including noise that occurs when the corresponding bathroom device is operating alone. However, it may be possible to store an acoustic model representing the acoustic features of voice commands including noise generated when a plurality of corresponding bathroom devices are operating simultaneously, or when the bathroom device operates alone It is also preferable to prepare both an acoustic model of an audio signal including noise generated at the same time and an acoustic model of an audio signal including noise generated when a plurality of bathroom devices operate simultaneously. Occurs when bath equipment is operated alone as an acoustic model used for speech recognition, as some bathroom equipment may be operated independently during bathing, or multiple bathroom equipment may be used simultaneously. If you have both an acoustic model of an audio signal that includes noise and an acoustic model of an audio signal that includes noise that occurs when multiple bathroom devices operate simultaneously, the audio of the audio signal that includes noise Recognition can be performed reliably and the recognition rate of voice recognition can be improved.

  In addition, assuming a plurality of scenes (scenes) in which a bather takes a bath, a combination of bathroom devices used for each scene is set in the controller 11, and when any scene is selected using the controller 11. It is considered that a plurality of bathroom devices corresponding to the selected scene are used in combination. For example, in a “relaxation scene” for making the bather feel comfortable, the air conditioner 6 is cooled to prevent the bather from climbing so that the jet jet flow generator 19 can be bathed for a long time. I have to. Therefore, an acoustic model of an audio signal including noise generated when the jet jet generator 19 and the air conditioner 6 are simultaneously operated in the “relaxation scene” is prepared, and the controller 11 is used for “relaxation”. When the operation is performed in the “scene”, the acoustic model selection unit 36 selects an acoustic model corresponding to the “relaxation scene” based on the detection result of the operating state detection unit 17, and the speech recognition unit uses the selected acoustic model. When 15 performs voice recognition, the recognition accuracy of voice recognition can be improved. In other scenes, an acoustic model of an audio signal including noise of bathroom equipment operated in each scene is prepared, and when any scene is selected and operated using the controller 11, The acoustic model selection unit 36 selects an acoustic model corresponding to the selected scene, and the speech recognition unit 15 performs speech recognition using the selected acoustic model, so that the recognition rate of speech recognition can be improved. .

1 is a block diagram illustrating a system configuration of a voice operation device according to a first embodiment. It is an external view which shows the example of installation of the bathroom apparatus to which the voice operation apparatus same as the above is applied. It is a front view of the controller used for the same as the above. It is a principal part block diagram of the voice operating device of Embodiment 2. It is a principal part block diagram of the voice operating device of Embodiment 3. It is a principal part block diagram of the voice operating device of Embodiment 4. It is a principal part block diagram of the voice operating device of Embodiment 5.

Explanation of symbols

A bathroom device B voice operation device 12 A / D conversion unit 13 noise removal unit 13a to 13n noise removal unit 14 acoustic model storage unit 15 voice recognition unit 16 voice operation unit 17 operating state detection unit

Claims (12)

  A plurality of bathroom devices provided in the bathroom, a voice detection unit provided in the bathroom, a noise removal unit that removes noise from the voice detected by the voice detection unit, and an acoustic representing an acoustic feature of a predetermined voice command An acoustic model storage unit that stores a model; a voice recognition unit that recognizes the content of the voice by performing pattern matching between the voice from which noise has been removed by the noise removal unit and the acoustic model stored in the acoustic model storage unit; The noise removing unit, comprising: a voice operation unit that operates a corresponding bathroom device according to the content of the voice recognized by the voice recognition unit; and an operating state detection unit that detects an operating bathroom device from the plurality of bathroom devices. In addition, a plurality of noise removing means for removing noise generated by the corresponding bathroom device is provided corresponding to each bathroom device, and the noise removing unit is a bath detected by the operating state detecting unit. Using the noise removal means corresponding to the device, the bathroom apparatus characterized by removing noise from speech detected by the speech detection section.   Each noise removing unit includes a noise storage unit in which information representing characteristics of noise generated by a corresponding bathroom device is registered in advance. Based on the information stored in the noise storage unit, the voice detection unit The bathroom apparatus according to claim 1, wherein the noise is removed by taking a difference between the detected voice and the noise generated by the bathroom device.   The bathroom apparatus according to claim 1, wherein the plurality of noise removing means includes a plurality of types of noise filters prepared in accordance with a noise of a corresponding bathroom device.   The noise removing unit includes a noise filter that removes noise from an input signal, and a filter strength setting unit that individually sets the strength of noise removal by the noise filter in accordance with the noise of the corresponding bathroom device. 2. The bathroom apparatus according to claim 1, wherein each of the noise removing means is constituted by a noise filter in which the intensity of noise removal is individually set according to the noise of the corresponding bathroom equipment.   Corresponding to each of the plurality of noise removing means, an acoustic model representing an acoustic feature of a signal output when the voice command is input to the corresponding noise removing means is prepared, and the voice recognition unit The bathroom apparatus according to any one of claims 1 to 4, wherein speech recognition is performed using an acoustic model corresponding to the noise removing means used by the unit.   The bathroom apparatus according to any one of claims 1 to 5, wherein the noise removing means removes noise generated in a state where a corresponding bathroom device is operating alone.   The bathroom apparatus according to any one of claims 1 to 5, wherein the noise removing means removes noise generated in a state where a plurality of corresponding bathroom devices are operating simultaneously.   A plurality of bathroom devices provided in the bathroom, an audio detection unit provided in the bathroom, an operating state detection unit for detecting a bathroom device in operation from the plurality of bathroom devices, and corresponding to each bathroom device An acoustic model storage unit that stores an acoustic model representing an acoustic feature of a voice command including noise generated during operation of a corresponding bathroom device, and an acoustic model stored in the acoustic model storage unit Perform pattern matching between the acoustic model selection unit that selects the acoustic model corresponding to the bathroom device detected by the operating state detection unit, and the acoustic model selected by the acoustic model selection unit and the voice detected by the voice detection unit A voice recognition unit for recognizing the content of the voice by the voice recognition unit and a voice operation unit for operating the corresponding bathroom device according to the voice content recognized by the voice recognition unit. Bathroom equipment, characterized in that.   The bathroom apparatus according to claim 8, wherein the acoustic model represents an acoustic feature of a voice command including noise generated when a corresponding bathroom device is operating alone.   9. The bathroom apparatus according to claim 8, wherein the acoustic model represents an acoustic feature of a voice command including noise generated when a plurality of corresponding bathroom devices are operating simultaneously.   It is an audio | voice operating device used for the bathroom apparatus of any one of Claim 1 thru | or 7, Comprising: The said audio | voice detection part, the said noise removal part, the said acoustic model memory | storage part, the said voice recognition part, the said audio | voice operation part And a voice operating device comprising the operating state detecting unit. It is an audio | voice operating device used for the bathroom apparatus of any one of Claim 8 thru | or 10, Comprising: The said audio | voice detection part, the said operation state detection part, the said acoustic model memory | storage part, the said acoustic model selection part, the said audio | voice A voice operation device comprising a recognition unit and the voice operation unit.

Images