JP2003216177A - Speech recognition device for vehicle - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a speech recognition device for vehicle with a high recognition rate capable of properly selecting traffic information, an E-mail of a cellular telephone, news by the Internet connection, etc., by a speech instruction of a crew in the vehicle during traveling, etc. <P>SOLUTION: This invention is provided with a speech input means 2 for inputting the speech instruction and a manual execution means 3 capable of performing speech recognition of the speech instruction by comparing the speech instruction from the speech input means 2 with stored speech sample information and executing a request corresponding to the speech instruction in the speech recognition device for vehicle for executing an instruction corresponding to the request by the recognized speech instruction, when the speech recognition of the speech instruction from the speech input means 2 can not be performed, execution corresponding to the speech instruction which is not recognized can be performed by the manual execution means 3. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle voice recognition apparatus capable of appropriately selecting traffic information, E-mail of a mobile phone, news connected to the Internet, etc. by a voice command of an occupant in a vehicle while traveling. It is about.

In recent years, voice recognition technology has become widespread and has been partially put to practical use in various devices such as personal computers and households. When performing voice recognition on a device such as a personal computer, in order to make the device remember the features of the speaker, you can read sentences of a certain length, etc., and program the features of the speaker as a program (software). A method of improving the recognition rate by analyzing or extracting after analysis is used. In addition, some devices aim to improve the recognition rate by identifying the speaker and pronouncing necessary words in advance and registering them.

[0003]

Such a voice recognition technique has already been applied to the radio and car navigation system T.
In the case of adopting a vehicle such as a V in a vehicle, and a cellular phone, a personal computer, or the like, which is often brought in, in recent vehicles, there are the following problems.

(1) For example, when an unspecified number of people use it, such as a business car or rental car, even if it is a personal car, some families will Since the vehicle may be driven, a high recognition rate needs to be secured for at least a plurality of occupants.

(2) Further, when the vehicle is traveling or traveling at a high speed such as a highway, a considerably large amount of noise is generated.
There is. For this reason, even if you read a sentence etc. without noise and let the device remember the features of voice,
The recognition rate of voice recognition may not always increase.

The present invention has been made in view of such a conventional situation, and basically, even if the voice command of the occupant may not be recognized, the voice command intended by the occupant is surely dealt with. It is an object of the present invention to provide a device capable of executing the above request and easily responding to this unrecognized voice command from the next time.

[0007]

According to a first aspect of the present invention, a voice input means for inputting a voice command and a voice command from the voice input means are analyzed and compared with stored voice sample information. A voice recognition device for a vehicle, which recognizes a voice command by voice and executes a command corresponding to the request of the recognized voice command, has a manual execution means capable of executing the request corresponding to the voice command, and the voice input In the vehicle voice recognition device, when the voice command from the means cannot be recognized, the manual execution means executes the operation corresponding to the unrecognized voice command.

According to a second aspect of the present invention, the unrecognized voice command that could not be recognized is stored as voice sample information in a voice sample storage unit, and the voice command is the same as or similar to the next unrecognized voice command. The voice recognition device for a vehicle according to claim 1, wherein when a voice command is given, the voice command is voice-recognized by comparing it with the stored voice sample information.

According to a third aspect of the present invention, the voice input means is disposed in the vicinity of the first microphone for inputting the voice of the occupant and the first microphone, and the directional characteristic is substantially opposite to the first microphone. And a second microphone that mainly collects ambient environmental noise and that obtains a difference between the output of the first microphone and the output of the second microphone and outputs the difference as an occupant's voice command. Alternatively, the vehicle voice recognition device according to the above 2 is provided.

[0010]

1 is a block diagram showing an entire vehicle voice recognition apparatus according to an embodiment of the present invention. In the figure, 1 is a CP installed in an appropriate part of the target vehicle
A device main body (controller) including U, various memories, other electronic elements, parts, etc., 2 is voice input means such as a microphone, 3 is manual execution means including a switch, 4 is a radio that is carried in a vehicle or in a vehicle, Electronic devices such as a TV, a mobile phone, and a personal computer of a car navigation system, 5 is an audio output device including a speaker, and 6 is a power switch.

In the main body 1 of the apparatus, for example, a voice analysis unit 11 which mainly analyzes voice commands from the voice input means 2, for example.
0, voice temporary storage unit 12 for temporarily storing voice commands
0, a voice sample information storage unit 130 in which various voice commands are stored as a sample, a voice comparison determination unit 140 that compares and determines the input voice command of the occupant and voice sample information stored in advance, and is determined. Command control unit 150 for executing a request for a voice command, a forced recognition association unit 160 for forcibly associating an unrecognized voice command with a predetermined voice command by the manual execution means 3, various information of the electronic device 4, for example, “traffic” The information update units 171 to 173 for updating and saving "information", "E-mail", "news" and the like are incorporated. When any of the information update units 171 to 173 is selected, the corresponding update information is output from the voice output device 5 in the form of reading aloud.

In the case of a single microphone, the voice input means 2 may be installed, for example, on the steering column cover portion, the inside of the pillar beside the windshield, or the rearview mirror portion on the ceiling side. In this case, it is preferable to use an object having a strong directivity so that the sound of a specific occupant such as a driver can be picked up well. Further, as shown in FIG. 2, it may be a headphone type. In this type, the first microphone 2 facing the passenger side
1 and a second microphone arranged on the back surface (nearby side) of the first microphone 21 and having directivity in a direction substantially opposite to the first microphone 21 are attached to the tip of the arc-shaped head frame 23. In this voice input means 2, as shown in FIG. 3, the difference between the output of the first microphone 21 and the output of the second microphone 22 is obtained by the subtraction circuit 26, and this is input to the voice analysis unit 110 as the voice command of the occupant. Let As a result, noise during traveling is canceled to some extent. Note that 24 and 25 are amplifiers on the microphone side.

The switch of the manual execution means 3 may be attached to a portion easily reachable by the occupant, especially the driver side, for example, a steering wheel portion or an instrument panel portion. The structure is also not particularly limited, but for example, a structure as shown in FIG. 4 can be mentioned.
In this structure, the pressing knob 32 is incorporated in a part of the dome-shaped case 31, and preferably, a light emitting portion 33 for confirming the operation that emits light during pressing is provided in the vicinity thereof. As will be described later, with this knob 32, "traffic information" is output when pressed once and "E-mail" when pressed twice.
Is output, and when pressed twice, "news" is output.

In the voice recognition device having such a configuration, for example, as described above, in order to read aloud "traffic information", "E-mail", "news", etc., as shown in the flow chart of FIG. Just go.

First, the power switch 6 of the voice recognition device is turned on. Then, when the occupant issues a voice command, for example, "traffic information", the voice command is input from the voice input means 2 in step 201.
It is judged whether the analysis (analysis) has been completed with 0. For example, here, if the pronunciation of the "traffic information", which is the voice command of the occupant, is a standard pronunciation without a habit, the normal voice analysis unit 110
"Traffic information" is analyzed correctly and successfully. Also,
The voice characteristics of the occupants can also be captured. However, if the pronunciation of the occupant has a large habit and the pronunciation is different from the standard pronunciation, the "traffic information" cannot be analyzed normally and is regarded as a kind of code as an unrecognized voice command. . If there is no voice command input or if the voice command fails, an announcement such as "Please input again" is given.

Both the voice command normally analyzed and the unrecognized voice command are temporarily stored in the voice temporary storage unit 120 in step 202. Next, in step 203, the stored voice command or unrecognized voice command is compared with voice sample information registered in advance in the voice sample information storage unit 130, and in step 204,
The match is judged. Note that the voice sample information storage unit 130 stores, as voice sample information, for example, "traffic information" or a phrase of "traffic", "E-mail" or a phrase of "mail", a phrase of "news", and the like. Has been done.

Therefore, when the "traffic information" consisting of the normally analyzed voice command is input, the voice comparison / judgment unit 140 matches the "traffic information" of the voice sample information. so,
After making an announcement such as "I understand (or understand)", the process reaches step 206, and the "traffic information", which is a voice command request, is the command control unit 150, the information updating unit (traffic information). It is output in the form of reading aloud through 171 and the voice output device 5.

On the other hand, when there is a mismatch, that is, when there is an unrecognized voice command, for example, in step 210, after announcing, "Unrecognized. Manual recognition is required", the switch operation of the manual execution means 3 is guided. In step 211, it is judged whether or not the switch operation has been performed.
Judge whether or not it has been pressed twice. Here, in the case of once,
In step 213, the compulsory recognition association unit 160
Forcibly (hardware) unrecognized voice command "traffic information"
Associate with. As a result, although not recognized, "traffic information", which is a request for an occupant's voice command, is output in the form of aloud reading through the information updating unit (traffic information) 171 and the voice output device 5.

Further, in the forced recognition relating section 160, the voice sample information storage section 1 in a form in which an unrecognized voice command is associated with "traffic information" by the switch operation described above.
Registered at 30. Therefore, from the next time onward, even if the voice command of the occupant is still unrecognizable in terms of pronunciation and the like, since it is already registered as voice sample information, the voice comparison and determination unit 140 will match. Then, immediately, the "traffic information" is displayed in the command control unit 150,
Through the information updating unit (traffic information) 171, and the voice output device 5, the information is output in the form of reading aloud.

There are two types of flow for the voice command of the occupant (at the time of normal analysis and at the time of non-recognition). These are voice commands other than "traffic information", "traffic", "E-mail", The same applies to “mail” and “news”. However, when the voice command of the passenger's "E-mail" or "mail" is an unrecognized voice command, the switch operation is performed twice, so the process proceeds to step 214, and in step 215, the forced recognition association unit 160 may forcibly associate the unrecognized voice command with the E-mail. Further, when the occupant's “news” voice command is an unrecognized voice command, the switch operation is performed three times,
Proceeding to step 216, in step 217, the forced recognition association unit 160 may forcibly associate the unrecognized voice command with the news.

As a result, similar to the above, "E-mail" and "news" are sent in accordance with the voice command request of the passenger.
It is output in the form of reading aloud through the information updating unit (E-mail) 172, the information updating unit (news) 173, and the voice output device 5. Of course, also in this case, these unrecognized voice commands are registered in the voice sample information storage unit 130 in a form associated with “E-mail” or “news”. Therefore, from the next time, even if the voice command of the occupant has a pronunciation problem, the desired "E-mail" or "news" is immediately output in the form of reading aloud.

As described above, in the voice recognition apparatus of the present invention, even if the voice command of the first occupant becomes an unrecognized voice command, if the manual execution means 3 is used, the unrecognized voice command is not recognized. Since the voice command can be associated as the voice command intended by the occupant, a high recognition rate is obtained as a result. Therefore, as the use is repeated, it becomes possible to deal with more passengers relatively easily.

If a headphone type as shown in FIG. 2 is used as the voice input means 2 in the voice recognition apparatus of the present invention, noise during traveling is canceled (eliminated) to some extent. It is also possible to improve the recognition rate.

Although the switch knob 32 in the manual execution means 3 of the above embodiment distinguishes the voice command of the occupant by the number of times of pressing, the present invention is not limited to this. For example, it is possible to have a structure in which a plurality of switches are arranged. Further, the number of requests by the voice command is not limited to three kinds as in the present embodiment, but may be four kinds or more, or may be one kind.

[0025]

As is apparent from the above description, according to the voice recognition device of the present invention, the following excellent effects can be obtained.

In appropriately selecting traffic information, E-mail of a mobile phone, news connected to the Internet, etc. by a voice command of an occupant in a vehicle while traveling,
If the voice command of the occupant is a standard pronunciation, the request corresponding to the target voice command is quickly realized. That is, "traffic information", "E-mail", "news", etc. as desired are provided by reading aloud.

Further, even when the voice command of the occupant is unrecognized due to a habit, the manual execution means and the compulsory recognition association unit are forcibly associated with the request intended by the occupant. The desired "traffic information" or "E-
"Mail", "news", etc. are provided by reading aloud. Moreover, from the next time onward, the unrecognized voice command is registered in the voice sample information storage unit by associating the operation by the first-time manual execution means. Will be realized.

The voice input means is composed of a first microphone for inputting the voice of the occupant and a second microphone in the vicinity of this microphone whose directional characteristics are in opposite directions, and the outputs of these two microphones are provided. When a voice command of the occupant is obtained by calculating the difference, the high recognition rate can be obtained even in the case where there is running noise in a special object such as a vehicle.

[Brief description of drawings]

FIG. 1 is a block diagram showing an entire voice recognition device according to an embodiment of the present invention.

FIG. 2 is a perspective view showing an example of voice input means used in the present invention.

3 is a block diagram of a circuit used in the voice input means of FIG.

FIG. 4 is a perspective view showing an example of a manual execution means used in the present invention.

FIG. 5 is a flowchart of a voice recognition device according to the present invention.

[Explanation of symbols]

1 Device body (controller) 2 Voice input means 3 Manual execution means 4 electronic devices 5 Audio output equipment 110 Voice analysis unit 120 Voice temporary storage 130 Voice sample information storage 140 voice comparison judgment unit 150 Command controller 160 Compulsory recognition related department 171 to 173 Information update unit

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Tomio Murai             1-12-1 Osaki, Shinagawa-ku, Tokyo Stock market             In company Altia (72) Inventor Toshihiro Kimura             1-12-1 Osaki, Shinagawa-ku, Tokyo Stock market             In company Altia (72) Inventor Hiroshi Taguchi             1-12-1 Osaki, Shinagawa-ku, Tokyo Stock market             In company Altia F-term (reference) 5D015 EE05 GG03 KK01 LL07

Claims (3)

[Claims] 1. A voice input means for inputting a voice command, and a voice command is voice-recognized by analyzing the voice command from the voice input means and comparing it with stored voice sample information, and the recognized voice is recognized. In a vehicle voice recognition device for executing a command corresponding to a command request, the vehicle voice recognition device has a manual execution means capable of executing the request corresponding to the voice command, and when the voice command from the voice input means cannot be recognized by voice, A voice recognition device for a vehicle, wherein the manual execution means performs an execution corresponding to a voice command for recognition. 2. The unrecognized voice command that could not be recognized is stored in the voice sample storage unit as voice sample information, and when the next voice command is the same as or similar to the unrecognized voice command, 2. The vehicle voice recognition device according to claim 1, wherein the voice command is voice-recognized by comparing with the stored voice sample information. 3. The voice input means is arranged near the first microphone for inputting the voice of the occupant and the first microphone, and has a directional characteristic in a direction substantially opposite to the first microphone, and mainly has a surrounding environment. It consists of a second microphone that collects noise,
The vehicle voice recognition device according to claim 1 or 2, wherein a difference between the output of the first microphone and the output of the second microphone is obtained and used as a voice command of an occupant.