JPS5961890A - Voice recognition system for vehicle - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a voice recognition method for a vehicle, and particularly to a voice recognition method for a vehicle that identifies a speaker's voice control command and controls vehicle equipment.

In recent years, many types of in-vehicle devices such as air conditioners and audio systems have been installed in vehicles such as automobiles, and with this, it has become possible to operate these various in-vehicle devices easily and reliably without disturbing the driver's concentration. Method development is required. For this reason, for example, an operation system has been developed and put into practical use that provides a keypad near the front panel of the vehicle and centrally operates multiple in-vehicle devices by operating various keys provided on the keypad. In such an operating system, when a driver operates a key while driving, the driver must keep his/her line of sight constantly moving, which hinders the driver's concentration on driving, even if only temporarily, which poses a problem in terms of safe driving.

Therefore, in order to easily and reliably operate in-vehicle devices while solving such problems regarding safe driving, efforts are being made to develop and put into practical use systems that use voice recognition methods for operating in-vehicle devices.

FIG. 1 shows an apparatus for carrying out a method of voice recognition for a vehicle.

In FIG. 1, a voice recognition circuit 12 is connected to a microphone 10 that inputs a speaker's voice control command 100, and the voice recognition circuit 12 includes a radio 14-1, an air conditioner 14-2,
In-vehicle equipment 14 such as the above is connected. The voice control command 100 is converted into an electrical signal by the microphone 10 and then supplied to the voice recognition circuit 12, and the voice uR circuit 12 recognizes the voice control command 100 and determines the control content corresponding to the voice control command 100. A signal 102 indicating the control content is output to the on-vehicle device 14. Therefore, this signal 102 causes the desired in-vehicle device 14 to be operated. for example,
When the speaker inputs the voice control command 100 "radio on" into the microphone 10, the voice recognition circuit 12 inputs the voice control command 1
00 is recognized and a signal 102 is outputted, whereby the lasso 14-1 operation is performed.

Note that FIG. 2 shows a flowchart of the vehicle voice recognition method. In FIG. 2, first, a voice control command is output, the voice control command is input, and the amplitude is normalized. Thereafter, the distance to the registered word is calculated, the distance is normalized, and a determination is made. In this way, the control content corresponding to the input voice control command is determined.

As described above, according to the voice recognition method for vehicles, the driver can control various in-vehicle devices without taking his/her line of sight from the direction of travel of the vehicle, and even if the driver operates the in-vehicle devices while driving. You can drive safely without disturbing your concentration on driving.

However, in conventional voice recognition methods for vehicles, such as "from the moth", the recognition is greatly affected by surrounding noise, such as the sound of other vehicles or own troops, audio sounds, the sound of the heater or air conditioner blower, and the chatter of passengers. However, there is a problem in that the speech recognition rate decreases due to surrounding noise, and in the worst case, the noise may cause υ erroneous recognition or malfunction. For example, when inputting a voice control command 100 such as "radio on", the average correct recognition rate is 80 and the false recognition rate is 20%. Therefore, there is a 20% probability that other in-vehicle There was a drawback that the device 14 was manipulated.

In addition, as a conventional improved vehicle voice recognition method, the voice control command 100 is input twice, and only when the control contents of the first command 100 and the control contents of the second command 1000 match, the signal 102 is A method has been proposed for outputting , but such a method has the advantage of significantly reducing the false recognition rate, but has the disadvantage that the correct recognition rate also decreases. For example, one voice control command 1
When inputting 00, if the correct recognition rate is 80% and the false recognition rate is 20, if the voice control command Zoo is input twice, the false recognition rate is 0.2 x 0.2 = 0.04 ( 4
%), but the correct recognition rate is 0.8X 0.8=0.
There was a drawback that the ratio decreased to 64 (64%).

The present invention has been made in view of the above-mentioned conventional problems, and an object thereof is to provide a voice recognition method for a vehicle that can improve the correct recognition rate and reduce the false recognition rate.

In order to achieve the above object, the present invention captures at least three or more preset odd number voice control commands from a plurality of repeated voice control commands by a speaker, and recognizes each voice control command. The control content corresponding to the voice fli control command is determined, the control content is classified by frequency, and a signal of the control content with high frequency is output.

Hereinafter, preferred embodiments of the present invention will be described based on the drawings.

The present invention allows a speaker to repeatedly input a voice control command 100 a plurality of times, and from the plurality of commands 100, at least three
The voice control commands 100 of a preset odd number of times or more are received, the control contents corresponding to the voice control commands 100 of each time are determined, the control contents are classified by frequency, and the control contents with high frequency are determined. It is characterized by outputting a signal 102 of. Since the signal 102 of the control content with high frequency is output, it is possible to improve the correct recognition rate and reduce the false recognition rate.

FIG. 3 shows a flowchart of a voice recognition method for a vehicle according to an embodiment of the present invention. In the embodiment shown in FIG. 3, the voice control command 100 is received three times from the voice control command 100 repeated nine times by the speaker.

In FIG. 3, the voice control command 100 is received three times, and the control content corresponding to the voice control command 100 is determined each time based on the chart 70 in FIG. 2. After the input and control content determination for the third voice control command ioo is completed, the three control content is classified by frequency, and it is examined whether the same control content occurs more than once. If the number of times or more is exceeded, the signal 102 of the control content with high frequency is output, and in other cases, the Rejecti signal 102 is output.
Display +.

Regarding the flowchart in Fig. 3, Radio 14-1
This will be explained in more detail by taking as an example the case of operating .

The speaker repeatedly inputs the voice control command 100 "radio on" multiple times, and the voice control commands 100 to 3
Take in the voice control command Zoo. Then, the control content corresponding to each voice control command 100 is determined, and it is determined that the control content corresponding to the first and second voice control commands 100 is "radio on", and the third voice control The control content corresponding to command 100 is mistakenly recognized as "air conditioner on" due to the noise. Therefore, when these control contents are classified by frequency, since the control contents "radio on" have a high frequency, a signal 102 of the control contents "radio on" having a high frequency is output. In addition to this, the desired radio 14
-1 operation will be performed 1The control content corresponding to the first voice control command 100 is "radio on", and the control content corresponding to the second voice control command 100 is [air conditioner on J]. and the third voice control command 10
If the control content corresponding to 0 is "Y-GON", the same control content does not occur more than once, so the control content signal 102 cannot be output, and "Reject" is displayed.

Next, it will be explained that according to the embodiment of the present invention, the correct recognition iii& rate is improved and the false recognition rate is reduced.

For example, when 1000 voice control commands are input one time, the correct recognition rate is 80chi and the false recognition rate is 81'& zo%.If 100 voice control commands are input three times, The correct M recognition rate and false recognition rate are as follows. That is, correct recognition rate = 0.8 x O,8X 0.2+ 0,8
x O, 2x O, 8X O, 2X O, 8X O, 8
+ 0,8 X O, l'l X Q, 8=0.896
(About 90 inches) Error fj + XA11t rate = 0.2x0.2x0.8+0.
2x0.8xO, 24-0, 8x0.2x0.2+0.
2X0.2X0,2=o, 1o4 (approximately 10 chi).

Therefore, it is understood that according to the embodiment of the present invention, the correct recognition rate is improved from 80% to 90%, and the false recognition rate is decreased from 20% to well over 10%.

In addition, large 'l'i,j''Reject with a recognition rate of 10chi,
As a result, there is almost no chance of accidental operation of in-vehicle equipment.

As described above, according to the present invention, even under noisy conditions, it is possible to improve the correct recognition rate and reduce the false recognition rate. Therefore, when controlling the in-vehicle device using the speaker's voice control command, the in-vehicle device can be operated easily and reliably.

Further, the present invention can be applied as is to equipment currently in use without requiring any special equipment. Furthermore, the present invention has the advantage that it does not impose an extreme load such as turning on a voice capture switch every time a user utters voice during voice recognition.

[Brief explanation of drawings]

FIG. 1 is a block circuit diagram of an apparatus for carrying out the vehicle voice recognition method, FIG. 2 is a flowchart of the vehicle voice recognition method, and FIG. 3 is a flowchart of the vehicle voice recognition method according to an embodiment of the present invention. It is a diagram. 100...Voice control command, 102...Signal. Agent: Patent attorney Yoshi 1) Kenji (and 1 other person) Figure 1 Figure 2

Claims (1)

[Claims] (Accepts at least 3 or more preset odd number voice control commands from multiple repeated voice control commands by 11 speakers, recognizes each voice control command, and responds to the voice control command.) A voice recognition method for a vehicle, characterized by determining control contents, classifying the control contents by frequency, and outputting a signal of the control contents with high frequency.