JP2005122042A - Speech recognizing device, speech recognizing method, speech synthesizer, speech synthesis method, speech recognition system, speech synthesis system, speech recognition and synthesizing device, navigation system, and moving body - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide speech recognition and synthesis service, capable of changing the speech output volume/tone quality of synthesized speech, in matching with a situation and timing on, based the assumption of generation of noise beforehand, and which is made smooth and effective to the user by changing the acoustic model to be used for the speech recognition. <P>SOLUTION: A control change-over part 101 which can acquire positional information and control a speech recognizing part 7 and a speech synthesis part 6 carries out speech recognition and speech synthesis. On the basis of the positional information and locational situation data acquired by the control change-over part 101, the speech recognizing part 7 selects an acoustic model suitable for the locational situation data and the positional information, from among a plurality of acoustic models and carries out speech recognition processing. Moreover, on the basis of the positional information and the locational situation data acquired by the control change-over part 101, the speech synthesizing part 6 and a utterance volume/utterance tone quality change-over part 107 carry out adjustment of the utterance volume of the synthesized speech and/or changeover of the utterance tone quality. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a speech recognition device, a speech recognition method, a speech synthesis device, a speech synthesis method, a speech recognition system, a speech synthesis system, a speech recognition synthesis device and a navigation system, and a mobile object. The present invention is suitable for application to a voice interactive communication navigation system.

  Conventionally, in-vehicle speech synthesis and speech recognition, those described in Patent Literature 1 and Patent Literature 2 are known.

  First, Patent Document 1 provides a volume control system that automatically adjusts a volume in a vehicle interior so that it can be easily heard in accordance with a noise level in a traveling vehicle interior in an audio device mounted on an automobile. As an object, in an audio apparatus having a sound field processing means that is installed in a vehicle interior and performs sound field control processing on an acoustic signal from an acoustic signal generation source, a microphone that is installed in the vehicle interior and detects a room sound in the vehicle interior; The sound signal processed by the sound field processing means is subtracted from the room sound detected by the microphone, and a noise signal extracting means for extracting a noise signal in the passenger compartment is extracted by the noise signal extracting means. A volume control system having volume control means for controlling the volume of an acoustic signal in accordance with a noise signal is described.

  Japanese Patent Laid-Open No. 2004-228561 estimates the noise component with high accuracy to improve the recognition rate, and for the purpose of making the user easy to use, when the user turns on the talk switch for voice input. In addition, there is described a voice recognition device that silences a car audio device and provides a noise estimation section for a predetermined time to estimate stationary noise.

Specifically, the speech recognition apparatus in Patent Document 2 outputs a notification sound of “beep” at the end of the noise estimation section to start the speech detection section, where the user inputs a command or destination by speech. . The estimated noise component is removed from the voice input signal in the voice detection section to obtain a voice signal. It is configured to perform voice recognition processing by performing notification even at the end of the voice detection section, and then perform talkback to cancel mute of the car audio device.
Japanese Patent Laid-Open No. 10-303669 JP 2000-322098 A Japanese Patent Laid-Open No. 9-90963

  However, the above-described conventional speech synthesis and speech recognition have the following problems.

  That is, in the sound volume adjustment system described in Patent Document 1, the sound volume is adjusted with respect to the vehicle interior noise level. Therefore, the sound volume can be adjusted after the noise is detected. . For this reason, there is a problem in that the timing for adjusting the volume is delayed from the generation of noise, resulting in a significant time lag.

In addition, the speech recognition apparatus described in Patent Document 2 is an apparatus that can estimate noise components with high accuracy, but in practice, a noise estimation section of a certain time is provided to estimate steady noise. doing. Therefore, there is a problem that the estimation of the noise component is inevitably performed after detecting noise or noise.

  According to the knowledge of the present inventor, the problems in these conventional techniques are caused by treating noise as an unknown element.

  Therefore, the object of the present invention is to predict the noise and noise generation situation in advance by predicting the noise and noise generation. Speech recognition device, speech recognition method, speech synthesizer, speech synthesis method, speech synthesis recognition that can prepare for recognition execution in recognition and can provide smooth and effective services to users who use it An apparatus, a navigation system, and a moving body are provided.

In order to achieve the above object, the first invention of the present invention provides:
A voice recognition means configured to be capable of voice recognition;
Control switching means configured to be capable of acquiring position information and capable of switching control; and
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from a plurality of acoustic models,
Based on the acquired location information and the location situation data included in the location information, the acoustic model control switching means selects an acoustic model suitable for the location situation data and the location information from a plurality of acoustic models, and selected The speech recognition apparatus is configured to be capable of executing speech recognition processing using an acoustic model.

  In the first aspect of the invention, typically, output signals output from a plurality of operating units are configured to be input, and noise corresponding to noise information specific to each operating unit based on the output signal for each operating unit. The noise suppression characteristic database that stores the suppression characteristic data is further provided. Information on the noise suppression characteristic database is obtained for an acoustic model selected from a plurality of acoustic models based on an output signal from an active operating unit. Are added to execute voice recognition processing.

The second invention of this invention is:
Speech recognition means configured to recognize speech;
Control switching means configured to be capable of acquiring position information and capable of switching control; and
A switching acoustic model database configured with a plurality of acoustic models;
The acoustic model control switching means for selecting one acoustic model from a plurality of acoustic models is configured to be able to transmit / receive data to / from each other,
Based on the acquired position information and the location situation data included in the location information, the acoustic model control switching means is configured to select an acoustic model suitable for the location situation data and the location information from a plurality of acoustic models. It is a voice recognition system characterized by

In the second aspect of the invention, typically, it is configured to be able to input output signals from a plurality of operating units, and based on the output signal for each operating unit, noise suppression characteristics for noise information specific to each operating unit For the acoustic model selected from a plurality of acoustic models based on the output signal from the active operating unit, further having a noise suppression characteristics database storing data.
The information of the noise suppression characteristic database is added and the speech recognition process is executed.

The third invention of the present invention is:
Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from the acoustic model database for switching;
Using the control switching means configured to be able to acquire the position information including the location status data and capable of switching the control,
The speech recognition method is characterized in that an acoustic model suitable for the location situation data is selected from a plurality of acoustic models by the acoustic model control switching means based on the location situation data.

  In the third aspect of the invention, typically, it is configured to be able to input output signals from a plurality of operating units, and based on the output signal for each operating unit, noise suppression characteristics for noise information specific to each operating unit It also has a noise suppression characteristics database that stores data, and adds information in the noise suppression characteristics database to an acoustic model selected from multiple acoustic models based on the output signal from the active operating part. Thus, the voice recognition process is executed.

The fourth invention of the present invention is:
A speech synthesizer configured to synthesize phonemes and output as speech data;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
And having a control switching means configured to be able to acquire position information and capable of switching control,
The speech synthesizer is configured to be capable of adjusting the utterance volume and / or switching the utterance sound quality based on the current position information and the place situation data included in the position information.

The fifth invention of the present invention is:
Including speech synthesis means configured to synthesize phonemes and output as speech data, speech volume sound quality switching means configured to adjust speech volume and / or switch speech quality, and location status data The control switching means configured to be able to acquire position information and to be able to execute control switching is configured to be able to transmit and receive data to and from each other,
The speech synthesis system is configured to be capable of adjusting the speech volume and / or switching the speech quality based on the acquired current position information and location status data.

The sixth invention of the present invention is:
A speech synthesis means configured to synthesize phonemes and output as speech, a speech volume sound quality switching means configured to execute speech volume adjustment and / or speech quality switching, and a position including location status data Using control switching means configured to be able to acquire information and configured to be able to perform control switching,
The speech synthesis method is characterized in that, when outputting speech, the speech volume is adjusted and / or the speech quality is switched based on the current position information and location status data.

The seventh invention of the present invention is:
Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from a plurality of acoustic models;
Speech synthesis means configured to synthesize phonemes and output as speech;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
It is configured to be able to acquire position information, and has a control switching unit configured to be able to perform control switching between the voice recognition unit and the voice synthesis unit,
When the voice recognition means is selected by the control switching means, the acoustic model control switching means, based on the acquired position information and the place situation data included in the position information, from the plurality of acoustic models, the location situation data and It is configured to select an acoustic model suitable for location information and perform speech recognition processing.
When the voice synthesizing unit is selected by the control switching unit, the speech volume is adjusted and / or the voice quality is switched based on the acquired position information and the location situation data included in the position information. This is a speech recognition / synthesis apparatus characterized by being configured to be able to output speech.

The eighth invention of the present invention is:
Route information providing means for providing position information regarding a route traveled by the mobile body using a predetermined communication system; and a route based on the information provided on the mobile body and provided by the route information providing means. The navigation means configured to be output based on the road map information, the voice recognition means configured to be able to recognize voice, the position information can be acquired, and the control can be switched. The data is input between the control switching means, the switching acoustic model database configured with a plurality of acoustic models, and the acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models. Configured to output,
Based on the acquired current position information and the location situation data included in the location information, the acoustic model control switching means selects an acoustic model suitable for the location situation data and location information from a plurality of acoustic models. It is a navigation system characterized by being configured.

The ninth aspect of the present invention is:
It is configured to be able to input output signals output from a plurality of operating units provided in the moving body, and stores noise suppression characteristic data for noise information specific to each operating unit based on the output signals for each operating unit. , Configured to be able to input and output data to and from the noise suppression characteristics database,
Based on the output signal from the operating unit, it is configured to add the information of the noise suppression characteristics database to the acoustic model selected from a plurality of acoustic models and execute the speech recognition process. It is a navigation system.

The tenth aspect of the present invention is:
Route information providing means for providing position information regarding a route traveled by the mobile body using a predetermined communication system; and a route based on the information provided on the mobile body and provided by the route information providing means. Navigation means configured to output based on road map information, speech synthesis means configured to synthesize phonemes and output as speech, and adjust speech volume and / or switch speech quality The utterance volume sound quality switching means is configured to be able to acquire position information and is configured to be able to input / output data between the control switching means configured to be able to perform control switching,
The navigation system is configured to be capable of adjusting the speech volume and / or switching the speech quality based on the current location information and the location situation data included in the location information.

The eleventh aspect of the present invention is:
Route information providing means for providing position information relating to a route along which the mobile body moves using a predetermined communication system;
Navigation means mounted on a mobile body and configured to be able to output a route based on the information provided by the route information providing means based on predetermined road map information;
Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from a plurality of acoustic models;
Speech synthesis means configured to synthesize phonemes and output as speech;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
The position change information is configured to be acquired, and the control switching unit configured to be able to execute control switching between the voice recognition unit and the voice synthesis unit is configured to be able to input and output data mutually.
When the voice recognition means is selected by the control switching means, the acoustic model control switching means, based on the acquired position information and the place situation data included in the position information, from the plurality of acoustic models, the location situation data and It is configured to select the acoustic model suitable for the location information and execute the speech recognition process,
When the speech synthesizer is selected by the control switching unit, based on the acquired position information and the place situation data included in the position information, the utterance volume is adjusted and / or the utterance sound quality is changed. The navigation system is configured to output synthesized speech.

The twelfth aspect of the present invention is
Route information providing means for providing position information regarding a moving route using a predetermined communication system, and a route based on the information provided by the route information providing means can be output based on predetermined road map information. Navigation means, voice recognition means configured to recognize voice, control switching means configured to be able to acquire position information and control switching, and a plurality of acoustic models. Configured to be able to input / output data between the acoustic model database for switching configured as described above and the acoustic model control switching means for selecting one acoustic model from a plurality of acoustic models. Based on the location status data included in the information, the acoustic model control switching means applies the location status data and the location information from a plurality of acoustic models. A moving body, comprising the navigation system configured to select an acoustic model.

  In the twelfth aspect of the invention, typically, the moving body has a plurality of operating units, is configured to be able to input output signals output from the plurality of operating units, and is based on an output signal for each operating unit. It is configured to be able to further input / output data to / from the noise suppression characteristic database storing noise suppression characteristic data for noise information specific to each active part of multiple active parts, and based on output signals from the active parts, multiple The information of the noise suppression characteristic database is added to the acoustic model selected from these acoustic models, and the speech recognition process is executed.

The thirteenth invention of the present invention is
Route information providing means for providing position information regarding a route traveled by the mobile body using a predetermined communication system; and a route based on the information provided on the mobile body and provided by the route information providing means. Navigation means configured to output based on road map information, speech synthesis means configured to synthesize phonemes and output as speech, and adjust speech volume and / or switch speech quality The utterance volume sound quality switching means is configured to be able to acquire position information and is configured to be able to input and output data between the control switching means configured to be able to perform control switching.
A mobile unit comprising a navigation system configured to be able to adjust the speech volume and / or switch the speech quality based on the current location information and the location status data included in the location information. is there.

The fourteenth aspect of the present invention is:
Route information providing means for providing position information about the route using a predetermined communication system;
A navigation means configured to be able to output a route based on the information provided by the route information providing means based on predetermined road map information;
Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from a plurality of acoustic models;
Speech synthesis means configured to synthesize phonemes and output as speech;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
It is configured to be able to acquire position information, and has a control switching unit configured to execute control switching between the voice recognition unit and the voice synthesis unit,
When the voice recognition means is selected by the control switching means, the acoustic model control switching means, based on the acquired position information and the place situation data included in the position information, from the plurality of acoustic models, the location situation data and It is configured to select an acoustic model suitable for location information and perform speech recognition processing.
When the speech synthesizer is selected by the control switching unit, based on the acquired position information and the place situation data included in the position information, the utterance volume is adjusted and / or the utterance sound quality is changed. A mobile object characterized by comprising a navigation system configured to be able to output synthesized speech.

  In the present invention, the case where the speech synthesis process is executed and the case where the speech recognition process is executed can be executed independently of each other, or can be executed concurrently. That is, the control switching unit can alternatively execute the voice recognition process by the voice recognition unit and the voice synthesis process by the voice synthesis unit, or can execute them simultaneously.

  According to the present invention configured as described above, based on the acquired position information and location situation data, an acoustic model suitable for the location situation data is selected from a plurality of acoustic models, and voice recognition processing or voice By executing the synthesis process, it is possible to execute an accurate voice recognition process or output a synthesized voice that is easy to hear for the user, without causing a time error.

  As described above, according to the present invention, location status data and location information are obtained from a plurality of acoustic models by the acoustic model control switching means based on the acquired location information and location status data included in the location information. By selecting an acoustic model that conforms to the above, it is assumed that such noise and noise will be generated before the noise and noise are generated. It is possible to prepare changes and changes to the acoustic model used in speech recognition, and to provide a smooth and effective speech synthesis service and speech recognition service to the user to use.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In all the drawings of the following embodiment, the same or corresponding parts are denoted by the same reference numerals. FIG. 1 shows an overall configuration of a system of a car navigation device installed in a vehicle according to this embodiment.

(Navigation system)
As shown in FIG. 1, in the vehicle-mounted navigation device 1 according to this embodiment, a microphone 2 capable of inputting sound, a speaker 3 capable of outputting sound, and a desired instruction can be input to the navigation device 1. It is configured to be connected to an input unit 4 as a simple input unit and a display unit 5 as an output unit for outputting data.

  The navigation device 1 includes a voice recognition unit 7 capable of voice recognition, a voice recognition unit 7 serving as an acoustic model control switching unit for selecting and switching an acoustic model, and a voice synthesis unit 6 serving as a voice synthesis unit capable of voice synthesis. have. In addition, a navigation ECU 8 is provided as a control unit of the navigation means for processing data input from the input unit 4, displaying desired data on the display unit 5, and executing various other navigations. It has been. In addition, it is connected to a GPS 9 having a sensor for measuring the position, and is configured so that the current position measurement unit 10 can uniquely measure the current position data of the vehicle.

  In addition, the server 20 and an information providing service provider (hereinafter referred to as ASP) 30 that provides useful information for navigation are connected to each other via a network 31 so that various traffic information data can be transmitted and received. Has been.

  The server 20 includes a voice communication processing unit 21 and a data communication processing unit 22. The voice communication processing unit 21 includes a voice synthesis processing unit 21a and a voice recognition processing unit 21b. In addition to these processing units, various processing units (not shown) for realizing other functions such as a voice / data synchronization control unit are provided.

  The traffic information server 30 that is an information providing ASP group is based on, for example, map information ASP that can provide route information based on a map information database 30a that is a database related to maps, or a traffic information database 30b that is a database related to traffic. It includes traffic information ASP that can provide road congestion and accident conditions, and event information ASP that has an event information database 30c that can provide information on various events.

  In this embodiment, the navigation apparatus uses the position information data supplied from the map information database 30a and the traffic information database 30b and the location status data included in the position information data. The travel route is navigated.

  The navigation device 1 is connected to the server 20 via the communication network 12 using the communication unit 11, and is configured to be able to execute transmission and reception of voice and data. The communication unit 11 may be configured to be detachable from the vehicle, and may be installed around the navigation device 1 at the time of boarding. Otherwise, the user may carry and use the communication unit 11.

(Speech recognition synthesizer)
Next, a voice recognition unit and a voice synthesis unit provided in the car navigation system according to the embodiment configured as described above will be described. FIG. 2 shows a car navigation system having a voice recognition unit and a voice synthesis unit according to this embodiment and a connection relationship with the vehicle.

  As shown in FIG. 2, the voice interactive car navigation system according to the embodiment includes a speech recognition / synthesis device 100. Note that only the navigation ECU 8 that performs information processing is shown for the configuration other than the speech recognition and synthesis apparatus in the car navigation system shown in FIG.

  In addition, in a vehicle (not shown), an air conditioner ECU 102 configured to be able to control an air conditioner, a wiper drive, a wiper drive unit 103 that outputs a wiper drive signal during this drive, and at least the rotational speed of the engine can be detected. The engine ECU 104 configured as described above, the power window opening / closing drive in the vehicle, and the power window driving unit 105 that outputs a control signal for opening / closing at this time, and the control signal at the time of blinking driving of the winker as the direction indicator A winker driving unit 106 for outputting is provided.

  In the speech recognition / synthesis apparatus 100, a speech-related speech control switching unit 101, a speech recognition unit 7 and a speech synthesis unit 6 are provided as control switching means.

  The voice control switching unit 101 is connected to the navigation ECU 8, the air conditioner ECU 102, the wiper driving unit 103, the engine ECU 104, the power window driving unit 105, and the winker driving unit 106. In addition, it is also possible to connect to operation parts and drives other than these as needed.

  In the speech recognition / synthesis apparatus 100, the speech control switching unit 101 is connected to the speech recognition unit 7 and the speech synthesis unit 6.

  When a voice uttered by the user is input for an instruction to the car navigation, the voice control switching unit 101 switches to the voice recognition unit 7 to execute a voice recognition process. On the other hand, when voice is output to the user for route guidance to the user, the voice control switching unit 101 switches to the voice synthesis unit 6 to execute voice synthesis processing.

  The voice recognition unit 7 is configured to be able to input and output data with the acoustic model database 121 stored and stored in, for example, a hard disk as a recording unit. The acoustic model database 121 includes an acoustic model switching database (acoustic model switching DB) 22 and an acoustic model addition database (acoustic model addition DB) 23.

  The acoustic model switching DB 122 is configured by storing data of a plurality of acoustic models corresponding to various situations in the speech recognition processing. In this embodiment, the plurality of acoustic models include a highway acoustic model 122a corresponding to a situation of traveling on a highway, a general road acoustic model 122b corresponding to a situation of traveling on a general road, and a situation of traveling on an uphill. A corresponding uphill acoustic model 122c and a mountain road acoustic model 122d corresponding to a situation of traveling on a mountain road are provided.

  Note that the situation corresponding to the acoustic model is not necessarily limited to these situations, and it is desirable to construct an acoustic model assuming various situations as much as possible and store it in the acoustic model switching DB 122. .

Specifically, for example, a road construction acoustic model corresponding to a situation where the road construction is being performed, a rain acoustic model corresponding to a running situation during rain, and a snowfall acoustic model corresponding to a running situation during snowfall , Tire chain mounted acoustic model corresponding to the situation where the tire chain is mounted, unpaved road acoustic model corresponding to the condition where the tire is traveling on the unpaved road, sandy beach acoustic model corresponding to the condition where the tire is driven on the sandy beach, shopping street The shopping street acoustic model corresponding to the situation of traveling can be listed.

  In this embodiment, the speech recognition unit 7 selects one acoustic model most suitable for the situation in which speech recognition processing is executed from the plurality of acoustic models stored in the acoustic model switching DB 122. It is used as an acoustic model in speech recognition processing.

  In addition, the acoustic model addition DB 123 is used when the acoustic recognition unit 7 selects and uses one acoustic model among the plurality of acoustic models stored in the acoustic model switching DB 122. It is a database in which data to be added as noise suppression characteristics is stored.

  The acoustic model addition database 123 is constructed from a plurality of noise suppression characteristics data configured to be able to reduce noise generated during driving in various mechanisms mounted on the vehicle. That is, for example, specific noise is generated when an engine or an air conditioner mounted on a vehicle is operated, and these noise characteristics are noises that can be recognized in advance.

  For example, the noise generated by driving the engine changes the frequency of the noise depending on the engine speed. This frequency change is a noise that depends on the engine speed, and is measured in advance to recognize its characteristics. Noise that can be kept.

  The noise generated when the air conditioner is operated varies depending on various factors such as the amount of air emitted from the air conditioner. This noise is also known so-called noise that can be measured in advance and its characteristics can be recognized.

  Similarly, noise generated when the wiper is driven, noise generated when the power window is operated to open / close the window, or noise when the winker is operating are also known noises.

  Since the noise suppression characteristics by these mechanisms are characteristics of noise generated by driving, drive signals output from the air conditioner ECU 102, the wiper driving unit 103, the engine ECU 104, the power window driving unit 105, and the winker driving unit 106 are input. Thus, it is possible to detect the operating state.

  In this embodiment, drive signals output from the air conditioner ECU 102, the wiper drive unit 103, the engine ECU 104, the power window drive unit 105, and the winker drive unit 106 are supplied to the voice control switching unit 101. Then, in accordance with this drive signal, the speech recognition unit 7 extracts the corresponding noise suppression characteristic data from the acoustic model addition DB 123 and adds it to the acoustic model used at that time, thereby adding noise related to known noise. Since it becomes possible to suppress, it becomes possible to perform noise suppression more effectively.

(Voice recognition processing)
Next, a voice recognition processing method in the car navigation system equipped with the voice recognition device according to the embodiment configured as described above will be described. FIG. 3 shows a speech recognition processing method according to this embodiment.

As shown in FIG. 3, in the speech recognition method according to this embodiment, first, in step ST1, the current vehicle position information and travel route information are transmitted via the in-vehicle LAN (local area network) to the voice control switching unit. 101.

  That is, the current position of the vehicle is acquired by the current position measurement unit 10 of the navigation system, and the vehicle travels after the current position included in the traffic information data from the traffic information database 30b of the traffic information server 30 through the communication unit 11. Road information, so-called road condition data is acquired. As a result, it is possible to obtain information necessary for selecting and searching for the acoustic model stored in the acoustic model switching DB 122.

  Next, the process proceeds to step ST <b> 2, and similarly, the operation information of the device driven in the current vehicle is supplied to the voice control switching unit 101 via the in-vehicle LAN. That is, drive signals are supplied to the audio control switching unit 101 from the air conditioner ECU 102, the wiper drive unit 103, the engine ECU 104, the power window drive unit 105, the winker drive unit 106, and the like.

  That is, for example, when the vehicle is running, the engine ECU 104 supplies a control signal including rotation speed data to each part of the engine. In this case, the audio control switching unit 101 is similarly supplied with a control signal including rotation speed data. Further, for example, when rain is encountered during traveling, the wiper is normally driven. Therefore, a drive signal is supplied from the wiper driving unit 103 to the wiper (not shown) and the audio control switching unit 101.

  Through the above-described steps ST1 and ST2, the current position of the vehicle is recognized, and travel path information on which the vehicle travels immediately afterward can be acquired. Note that the processing of step ST1 and step ST2 can be executed in reverse order.

  Next, the process proceeds to step ST3, and it is determined whether or not there is a change in the immediately following traveling path or traveling state, that is, the traveling state, during traveling of the vehicle. If the travel route is not changed immediately after that, the process proceeds to step ST5 described later. On the other hand, when the travel route is changed immediately after that, the process proceeds to step ST4.

  In step ST4, the most suitable acoustic model under the noise predicted to be generated is searched based on the travel route information acquired in step ST1, and the acoustic recognition unit 7 switches the acoustic model.

  Specifically, for example, when travel path information for traveling on a highway is acquired immediately after the vehicle is traveling on a general road in step ST1, the sound used in the speech recognition process in step ST3. The model is switched from the general road acoustic model 122b to the highway acoustic model 122a. The switching of the acoustic model is executed in accordance with the timing when the vehicle enters the highway from the general road.

  Thereafter, the process proceeds to step ST5, and the noise suppression characteristic data in the acoustic model addition DB 123 (in FIG. 2, engine noise noise suppression characteristic data 123a, air conditioner noise suppression is determined by the drive signal acquired by the speech recognition unit 7 in step ST2. From the characteristic data 123b, the wiper noise suppression characteristic data 123c, the window opening / closing noise suppression characteristic data 123d, and the winker noise suppression characteristic data 123e), the corresponding noise suppression characteristic data (for example, engine sound noise suppression characteristic data 123a, wiper noise suppression characteristic data). 123c) is retrieved.

The noise suppression characteristic data retrieved in this way is added with the engine sound noise suppression characteristic data 123a and the like to the predetermined acoustic model used at that stage, and the execution of the speech recognition process is continued. Specifically, for example, when switching from the general road acoustic model 122b, when the expressway acoustic model 122a is used, the engine noise suppression for suppressing noise according to the engine speed is used. The characteristic is added to the expressway acoustic model 122a, and the execution of the speech recognition process is continued.

  As described above, on the basis of the current position information and the information of the travel route that travels after that, the noise occurrence state based on the travel route information is assumed in advance, and the most appropriate for this noise An acoustic model is searched out, and an appropriate acoustic model is selected by the speech recognition unit 7 to execute speech recognition processing.

(Speech synthesis processing)
Next, speech synthesis processing will be described. FIG. 4 shows a flowchart of the speech synthesis process according to this embodiment.

  As shown in FIG. 4, in this embodiment, as in the voice recognition process, the current position information and the travel route information are acquired through the in-vehicle LAN in step ST11, and in step ST12, the air conditioner ECU 102 and the wiper drive are acquired. Drive signals are supplied to the voice control switching unit 101 from the unit 103, the engine ECU 104, the power window driving unit 105, the winker driving unit 106, and the like.

  Next, the process proceeds to step ST13, in which it is determined whether or not at least one of a situation in which a change in the driving situation occurs or an on / off state of the operation of the driving unit changes occurs.

  As a result of the determination, if there is no change in the travel path and no change in the operation of the drive unit, the speech synthesis process is executed with the setting state at that time. On the other hand, as in the case of the voice recognition process, when the traveling state is changed immediately or when the operation of various driving units changes, the process proceeds to step ST14.

  In step ST14, the utterance volume and / or the utterance sound quality are adjusted in accordance with the change of the change path in the location status data and / or the operation of each drive unit based on the drive signal.

  That is, when position information and location status data including a situation in which a travel route in a vehicle traveling on a general road immediately changes to a mountain road is supplied to the voice control switching unit 101, the travel route is from the general road to the mountain road. If it changes, noise often increases. Therefore, the signal from the voice control switching unit 101 is supplied to the voice synthesis unit 6, and the volume of sound output from the speaker 3 by the utterance volume / speech sound quality switching unit 107 is increased or decreased based on the control signal from the voice synthesis unit 6. Done. Note that, instead of increasing or decreasing the volume, it is also possible to switch the sound quality. For example, it is possible to switch from a male voice to a female voice.

  Further, for example, even when rain is encountered during driving and the wiper is driven, a wiper driving signal is output from the wiper driving unit 103 and supplied to the voice control switching unit 101. The drive signal supplied to the voice control switching unit 101 is supplied to the voice synthesis unit 6 during the voice synthesis process. The voice synthesis unit 6 that has acquired the drive signal supplies a control signal to the speech volume / speech sound quality switching unit 107. Then, the utterance volume / speech sound quality switching unit 107 adjusts the utterance volume. Note that it is possible to switch the speech quality instead of adjusting the speech volume accompanying the change of these drive units, but the change in the status of each drive unit occurs in a short time, so the speech It is desirable to employ volume adjustment.

  As described above, the speech recognition processing and speech synthesis processing according to the present invention are executed. These speech recognition processing and speech synthesis processing can be executed in parallel, or can be executed by switching timely.

  The embodiment of the present invention has been specifically described above, but the present invention is not limited to the above-described embodiment, and various modifications based on the technical idea of the present invention are possible.

  For example, the acoustic model and noise suppression characteristic data listed in the above-described embodiment are merely examples, and different acoustic models and noise suppression characteristic data may be used as necessary.

  For example, in the above-described embodiment, the speech synthesis / synthesis device 100 mounted on the vehicle is provided with the speech synthesis unit 6 and the speech recognition unit 7 to execute speech synthesis processing and speech recognition processing, respectively. 5, the speech synthesis processing unit 21 a and the speech recognition processing unit 21 b in the server 20 respectively execute speech synthesis processing and speech recognition processing according to the present invention, as shown in FIG. 5. It is possible.

  That is, the acoustic model switching database and the acoustic model addition database stored in the speech recognition / synthesis apparatus for navigation mounted on the vehicle in the above-described embodiment are stored in a predetermined storage unit (not shown) in the speech recognition processing unit 21b of the server 20. 2), the speech recognition processing unit 21b in the server 20 executes the speech recognition processing by the speech recognition unit 7 shown in FIG. 2, and the speech synthesis processing by the speech synthesis unit 6 is performed by the speech synthesis processing unit 21a. To make it run.

  Then, the input data used for the speech recognition process input from the microphone 2 and the output data subjected to the speech synthesis process output from the speaker 3 are transmitted to the server 20 via the communication unit 11 and the communication network 12. Can be configured to transmit and receive.

  In the case of such a distributed configuration, the vehicle only has a normal navigation device and a portable terminal or a mobile phone that can communicate with the outside, and the actual processing is executed in the server 20. . Therefore, it is possible to introduce operations using voice recognition processing, adopt driving guidance and event guidance using synthesized speech, etc. without significantly changing the operation of a normal mobile phone or navigation device. It becomes.

  In the above-described embodiment, the speech synthesizer and the speech recognizer are provided in one device. However, the speech synthesizer and the speech recognizer can be configured by separate hardware. In this case, the speech recognition device and the speech synthesizer are separately installed in the vehicle, or only one device is installed in the vehicle.

It is a figure which shows the structure of the communication type navigation system by one Embodiment of this invention. It is a block diagram which shows the speech recognition synthesizer by one Embodiment of this invention. It is a flowchart which shows the speech recognition processing method by one Embodiment of this invention. It is a flowchart which shows the speech synthesis processing method by one Embodiment of this invention. It is a figure which shows the structure of the communication type navigation system by the other example of one Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Navigation apparatus 2 Microphone 3 Speaker 4 Input part 5 Display part 6 Speech synthesizer part 7 Voice recognition part 8 Navigation ECU
9 GPS
DESCRIPTION OF SYMBOLS 10 Current position measuring part 11 Communication part 12 Communication network 20 Server 21 Voice communication process part 21a Speech synthesis process part 21b Speech recognition process part 22 Data communication process part 30 Traffic information server 30a Map information database 30b Traffic information database 30c Event information database 31 Network Network 100 Speech Recognition / Synthesizer 101 Voice Control Switching Unit 102 Air Conditioner ECU
103 Wiper drive unit 104 Engine ECU
105 power window driving unit 106 winker driving unit 107 utterance sound quality switching unit 121 acoustic model database 122 acoustic model switching database (acoustic model switching DB)
122a Expressway acoustic model 122b General road acoustic model 122c Uphill acoustic model 122d Mountain road acoustic model 123 Acoustic model addition database (acoustic model addition DB)
123a Engine sound noise suppression characteristic data 123b Air conditioner noise suppression characteristic data 123c Noise suppression characteristic data 123d Window opening / closing noise suppression characteristic data 123e Winker noise suppression characteristic data

Claims (18)

A voice recognition means configured to be capable of voice recognition;
Control switching means configured to be capable of acquiring position information and capable of switching control; and
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models,
Based on the acquired position information and the location situation data included in the location information, an acoustic model adapted to the location situation data and the location information is obtained from the plurality of acoustic models by the acoustic model control switching means. A speech recognition apparatus that is configured to be capable of performing speech recognition processing using the selected acoustic model. It is configured to be able to input output signals output from multiple operating parts,
Based on the output signal for each operating part, further comprising a noise suppression characteristic database storing noise suppression characteristic data for noise information specific to each operating part,
Based on an output signal from the operating unit that is in operation, the information of the noise suppression characteristic database is added to the acoustic model selected from the plurality of acoustic models, and the speech recognition process is executed. The speech recognition apparatus according to claim 1, wherein: Speech recognition means configured to recognize speech;
Control switching means configured to be capable of acquiring position information and capable of switching control; and
A switching acoustic model database configured with a plurality of acoustic models;
The acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models is configured to transmit and receive data to and from each other.
Based on the acquired position information and the location situation data included in the location information, an acoustic model suitable for the location situation data and the location information is obtained from the plurality of acoustic models by the acoustic model control switching means. A speech recognition system characterized by being configured to select. It is configured to be able to input output signals from multiple operating parts,
Based on the output signal for each operating part, further comprising a noise suppression characteristic database storing noise suppression characteristic data for noise information specific to each operating part,
Based on the output signal from the operating part in operation, the information of the noise suppression characteristic database is added to the acoustic model selected from the plurality of acoustic models, and the speech recognition process is executed. The speech recognition system according to claim 3, wherein the speech recognition system is configured. Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from the acoustic model database for switching;
Using the control switching means configured to be able to acquire the position information including the location status data and capable of switching the control,
A speech recognition method, wherein an acoustic model that matches the location status data is selected from the plurality of acoustic models by the acoustic model control switching means based on the location status data. It is configured to be able to input output signals from multiple operating parts,
Further comprising a noise suppression characteristic database storing noise suppression characteristic data for noise information specific to each of the operating units based on the output signal of each operating unit;
Based on the output signal from the operating section in operation, the information in the noise suppression characteristic database is added to the acoustic model selected from the plurality of acoustic models, and the speech recognition process is executed. The speech recognition method according to claim 5, wherein: A speech synthesizer configured to synthesize phonemes and output as speech data;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
And having a control switching means configured to be able to acquire position information and capable of switching control,
A speech synthesizer characterized in that the speech volume adjustment and / or speech quality switching can be performed based on the current location information and the location status data included in the location information. Including speech synthesis means configured to synthesize phonemes and output as speech data, speech volume sound quality switching means configured to adjust speech volume and / or switch speech quality, and location status data The control switching means configured to be able to acquire position information and to be able to execute control switching is configured to be able to transmit and receive data to and from each other,
A speech synthesizing system configured to be able to adjust the utterance volume and / or switch the utterance sound quality based on the acquired current position information and the location status data. A speech synthesis means configured to synthesize phonemes and output as speech, a speech volume sound quality switching means configured to execute speech volume adjustment and / or speech quality switching, and a position including location status data Using control switching means configured to be able to acquire information and configured to be able to perform control switching,
A speech synthesizing method characterized in that, when outputting speech, the speech volume is adjusted and / or the speech quality is switched based on the current position information and the location status data. Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models;
Speech synthesis means configured to synthesize phonemes and output as speech;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
And a control switching unit configured to be able to acquire position information, and configured to execute control switching between the voice recognition unit and the voice synthesis unit,
When a voice recognition unit is selected by the control switching unit, the acoustic model control switching unit selects from the plurality of acoustic models based on the acquired position information and the place situation data included in the position information. In addition, the acoustic model suitable for the location status data and the location information is selected and the voice recognition process can be executed.
When the speech synthesizer is selected by the control switching unit, the utterance volume adjustment and / or utterance sound quality switching is performed based on the acquired position information and the place situation data included in the position information. A speech recognition synthesizer that is configured to execute and output synthesized speech. Route information providing means for providing position information relating to a route along which the mobile body moves using a predetermined communication system, and the route based on the information mounted on the mobile body and provided by the route information providing means Navigation means configured to be output based on predetermined road map information, voice recognition means configured to be able to recognize voice, and the position information can be acquired, and control switching is performed Between the control switching means configured to be possible, the acoustic model database for switching configured with a plurality of acoustic models, and the acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models Is configured to be able to input and output data,
Based on the acquired current location information and the location status data included in the location information, the acoustic model control switching means uses the acoustic model control switching means to generate an acoustic suitable for the location status data and the location information. A navigation system that is configured to select a model. Noise suppression characteristic data for noise information specific to each operating unit, based on an output signal for each operating unit, configured to be able to input output signals output from a plurality of operating units provided in the moving body Is configured to be able to search further noise suppression characteristics data from the noise suppression characteristics database,
Based on the output signal from the operating unit, the information of the searched noise suppression characteristic database is added to the acoustic model selected from the plurality of acoustic models, and the speech recognition process is executed. It is comprised by these. The navigation system of Claim 11 characterized by the above-mentioned. Route information providing means for providing position information relating to a route along which the mobile body moves using a predetermined communication system, and the route based on the information mounted on the mobile body and provided by the route information providing means Navigation means configured to be output based on predetermined road map information, speech synthesis means configured to be able to output phonemes as speech, and adjustment of speech volume and / or switching of speech quality It is configured to enable input / output of data between the utterance volume sound quality switching means configured to be capable of acquiring position information and the control switching means configured to be able to perform control switching. ,
A navigation system configured to be capable of adjusting the utterance volume and / or switching the utterance sound quality on the basis of current position information and place situation data included in the position information. Route information providing means for providing position information relating to a route along which the mobile body moves using a predetermined communication system;
Navigation means mounted on the mobile body and configured to output the route based on the information provided by the route information providing unit based on predetermined road map information;
Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models;
Speech synthesis means configured to synthesize phonemes and output as speech;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
The position change information is configured to be acquired, and the control switching unit configured to be able to execute control switching between the voice recognition unit and the voice synthesis unit is configured to be able to input and output data to and from each other.
When a voice recognition unit is selected by the control switching unit, the acoustic model control switching unit selects from the plurality of acoustic models based on the acquired position information and the place situation data included in the position information. The voice recognition process is performed by selecting an acoustic model suitable for the location situation data and the position information, and
When the speech synthesizer is selected by the control switching unit, the utterance volume adjustment and / or utterance sound quality switching is performed based on the acquired position information and the place situation data included in the position information. A navigation system that is configured to execute and output synthesized speech. Route information providing means for providing position information about a moving route using a predetermined communication system, and the route based on the information provided by the route information providing means can be output based on predetermined road map information A configured navigation means; a voice recognition means configured to be able to recognize voice; a control switching means configured to be able to acquire the position information and capable of switching control; and a plurality of sound An acoustic model database for switching configured with a model and an acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models are configured to be able to input and output data between each other, and the acquired Based on the position information and the location situation data included in the position information, from the plurality of acoustic models by the acoustic model control switching means, Moving body, comprising the navigation system configured to select an acoustic model suitable for the serial location status data and the position information. Has multiple working parts,
It is configured to be able to input output signals output from the plurality of operating units, and stores noise suppression characteristic data for noise information specific to each operating unit of the plurality of operating units based on the output signal for each operating unit. It is configured to be able to input / output data to / from the noise suppression characteristics database,
Based on the output signal from the operating unit, the information of the noise suppression characteristic database is added to the acoustic model selected from the plurality of acoustic models, and the voice recognition process can be executed. The moving body according to claim 15, characterized in that: Route information providing means for providing position information relating to a route along which the mobile body moves using a predetermined communication system, and the route based on the information mounted on the mobile body and provided by the route information providing means Navigation means configured to be output based on predetermined road map information, speech synthesis means configured to be able to output phonemes as speech, and adjustment of speech volume and / or switching of speech quality It is configured to enable input / output of data between the utterance volume sound quality switching means configured to be able to acquire position information and the control switching means configured to be able to perform control switching. And
A navigation system configured to execute the adjustment of the utterance volume and / or the switching of the utterance sound quality based on the current position information and the location situation data included in the position information. body. Route information providing means for providing position information about the route using a predetermined communication system;
Navigation means configured to be able to output the route based on the information provided by the route information providing means based on predetermined road map information;
Speech recognition means configured to recognize speech;
A switching acoustic model database configured with a plurality of acoustic models;
Acoustic model control switching means for selecting one acoustic model from the plurality of acoustic models;
Speech synthesis means configured to synthesize phonemes and output as speech;
An utterance volume sound quality switching means configured to be able to adjust the utterance volume and / or switch the utterance sound quality;
Control information is configured to be able to acquire position information, and control switching means configured to be able to execute control switching between the voice recognition means and the voice synthesis means,
When a voice recognition unit is selected by the control switching unit, the acoustic model control switching unit selects from the plurality of acoustic models based on the acquired position information and the place situation data included in the position information. In addition, the acoustic model suitable for the location status data and the location information is selected and the voice recognition process can be executed.
When the speech synthesizer is selected by the control switching unit, the utterance volume adjustment and / or utterance sound quality switching is performed based on the acquired position information and the place situation data included in the position information. A moving body comprising a navigation system configured to execute and output synthesized speech.

Images