JPS5975300A - Voice recognition equipment 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 device for a vehicle, and more particularly to a voice recognition device 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 equipment have been installed in vehicles such as automobiles, and with this, it has become necessary to easily and reliably operate these various in-vehicle devices without disturbing the driver's concentration. It is necessary to develop a device that can do this. For this reason, an operation system has been developed and put into practical use, in which a keyboard is provided near the front panel of the vehicle, and a plurality of in-vehicle devices are centrally operated by operating various keys provided on the keyboard.

However, with this type of operation system, when the driver operates the keys while driving, the driver must move his/her line of sight to the keyboard, which may interfere with the driver's concentration on driving, even if only temporarily, and pose problems for safe driving. Ta.

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 devices for operating in-vehicle devices. When controlling in-vehicle devices using this voice recognition device, the driver can command the operation of a predetermined in-vehicle device by simply speaking into the microphone, for example, ``radio on''. For example, an operation to turn on a radio or the like is performed. This allows the driver to control various in-vehicle devices without taking his/her line of sight away from the direction in which the vehicle is traveling, and it is safe because even if the driver operates the in-vehicle devices while driving, his or her concentration on driving is not disturbed. Able to drive.

However, voice recognition devices for vehicles are greatly affected by noise such as running sounds and radio sounds, and the voice recognition rate decreases due to surrounding noise, and in the worst case, the noise can cause erroneous recognition and candy behavior. There was a problem.

Therefore, in conventional improved vehicle voice recognition devices, a threshold level for voice cutout is set, and the voice control signal is output only when the voice control signal from the speaker exceeds the threshold level. Intake,
A device has been proposed that controls predetermined in-vehicle equipment, and attempts to improve the recognition rate by capturing the voice control signal only when the voice control signal exceeds a threshold level.

However, in such a conventional device, since the threshold level is fixed, there is a problem that when the noise is large, a silent section cannot be detected and the recognition rate decreases.

In addition, as another conventional device, there is a device that lowers the gain of the amplifier circuit for the audio control signal with the threshold level fixed at l~, thereby relatively reducing the noise. The device also has the drawback of narrowing the dynamic range of the voice and lowering the recognition rate.

Furthermore, there is another conventional device that detects voice control commands and noise using a single speaker microphone, and changes the threshold level depending on the noise level. However, when noise having a pattern similar to a pre-stored registered voice pattern is captured, the threshold level changes and the recognition rate decreases.

The present invention has been made in view of the above-mentioned conventional problems, and its purpose is to provide a voice recognition device for a vehicle that can improve the recognition rate by changing the threshold level for all possible noises. Our goal is to provide the following.

In order to achieve the above object, the present invention includes a voice input section into which a speaker's voice control command is input, and when a voice control signal from the voice input section exceeds a threshold level for voice cutting. Only in a vehicle voice recognition device in which predetermined in-vehicle equipment is controlled by a voice control signal, a noise input section is provided separately from the voice input section, and the threshold value is set based on the noise signal from the noise input section. It is characterized by changing levels.

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

FIG. 1 shows a vehicle speech recognition device according to an embodiment of the present invention.

In FIG. 1, a speaker microphone 10 serving as a voice input section into which a speaker's voice control command 100 is input includes an amplifier circuit 1.
2. Multiplexer 16 via filter bank 14
is connected to the multiplexer 16, and a voice recognition circuit 20 is connected to the multiplexer 16 via an AD conversion circuit 18. The voice recognition circuit 20 recognizes the voice control signal 102 and outputs a control signal 104 corresponding to the voice control signal 102, thereby controlling in-vehicle equipment (not shown) such as a radio and an air conditioner. Become.

The voice recognition circuit 20 sets a threshold level V'I;
1, and only when the voice control signal 102 is equal to or higher than the threshold level Vth, the voice control signal 1 is
Process 02. Therefore, the predetermined in-vehicle equipment is controlled only when the voice control signal 102 exceeds the threshold level Vth.

FIG. 2 shows the relationship between the voice extraction threshold level (Vih) and the recognition rate (Q).

In order to effectively utilize this relationship between the threshold level and the recognition rate, the present invention (c) changes the threshold level according to the noise signal without reducing the gain of the amplifier circuit for the audio control signal. , which aims to improve the recognition rate. That is, in FIG. 1, the noise microphone 22 as a noise input section is provided at a position where the speaker's voice hardly enters and the noise is easily captured, and the noise signal 106 from the noise microphone 22 is
The multiplex 1/1 through the amplifier circuit 24 and the smoothing circuit 26
It is supplied to the chain 16. And the voice recognition circuit 20
changes the threshold level Vth based on the noise signal 106 from the noise microphone 22.

As shown in FIG. 2F, according to the embodiment of the present invention, the threshold level vthl; -r, the recognition rate can be improved and erroneous recognition and malfunction can be prevented.

FIG. 3 shows a flowchart according to an embodiment of the invention.

First, starting at step 200, step 2
At 02, the voice and threshold level Vth
In step 204, the voice and threshold levels Vt and h are compared. If the voice threshold level Vth is lower than the voice threshold level Vth, the counter A and the threshold level Vth are reset in step 206, and the voice and threshold level Vth are captured again in step 202.

If the audio is at or above the threshold level Vth in step 204, counter A is started in step 208 and the threshold level Vth is reset.In step 210, the count value of counter A exceeds the equivalent of 0.2 seconds. Find out.

Here, the voice control command is normally 0.2 seconds or more and 1.2 seconds or less. Therefore, if the counter A does not exceed the equivalent of 0.2 seconds in step 210, the captured voice is not a voice control command, so again,
Returning to step 202, the audio and threshold level Vth are captured. Further, in step 210, if the counter A exceeds the equivalent of 0.2 seconds,
The captured voice is a voice control command, so step 2
In step 12, import the audio again and proceed to step 2.
In step 14, the voice is compared with the threshold level Vth.

If the audio is equal to or higher than the threshold level Vth in step 214, audio processing is performed in state 216, and counter B is set in step 218.
Reset. Thereafter, in step 220, it is checked whether the counter A has exceeded the equivalent of 1.2 seconds. Here, as mentioned above, the voice control command is normally 0.2
It is at least 1.2 seconds and at most 1.2 seconds. Therefore, if the counter A does not exceed the equivalent of 1.2 seconds, the captured voice is a voice control command, so the process returns to step 212 and continues capturing the voice, and in step 214 the voice and threshold level Vth are Compare the results.

Further, if the audio is lower than the threshold level Vth in step 214, since audio capture has finished, counter B is started in step 222, and then in step 224, counter B is set to 0.2. Check whether the time has exceeded the equivalent of seconds.

Here, counter B is a counter that determines whether or not the voice control command is completely completed. For example, if the voice control command is "radio", there is no silent period, but
When the voice control command is "air conditioner", there is a silent section because "air" and "con" are pronounced separately.

Therefore, counter B is provided in order to prevent a malfunction in which the intake ends when only "air" is taken in. That is, the silent period of the voice control command is normally 0.2 seconds or less, and step 224
If counter B does not exceed the equivalent of 0.2 seconds, the process returns to step 212 because the audio capture has not been completed completely, and if counter B exceeds the equivalent of 0.2 seconds, the process returns to step 212. In this case, since there is no silent section and the audio capture has been completely completed, the process proceeds to step 226 and the audio capture is completed. If the audio processing is completed in step 228, counters A and B are reset in step 230, the threshold level Vth is reset in step 232, and the result of the audio processing is output in step 234. For example, the radio or air conditioner is operated.

In step 220, if the counter A exceeds the equivalent of 1.2 seconds, the captured audio is not a voice control command but noise, so the audio processing is stopped in step 236, and "Reject" is displayed in step 238. Then, in step 240, counter A is reset, and in step 242, the threshold level Vth is reset.In this case, the noise is mistakenly recognized as a voice control command, and this causes the threshold level Vth to be lowered. Therefore, in the present invention, the threshold level Vth is set high as described above in the next voice capture, and thereby the noise is determined to be a voice control command. Misrecognition can be prevented.

As mentioned above, using the flowchart in Figure 3, αυ 106..., Gua Province 8/#↓1 Therefore, only voice control commands can be accurately recognized.

As described above, according to the present invention, a noise t'f input section is provided, and the threshold level can be changed based on the noise signal from the noise input section, thereby preventing erroneous recognition by the speech recognition device. Malfunctions can be prevented. Therefore, when controlling the in-vehicle device using the speaker's voice control command, the in-vehicle device can be operated easily and reliably.

[Brief explanation of the drawing]

Fig. 1 is a block circuit diagram of a vehicle speech recognition device according to an embodiment of the present invention, Fig. 2 is a graph showing the relationship between the threshold level of speech extraction and recognition rate, and Fig. 3 is an embodiment of the present invention. FIG. 10...Mic for speaker, 22...Microphone for noise, 100...Voice control command, 102...Voice control signal, Mutsu agent Patent attorney Yoshi 1) Kenji (ha7ri s 1 person) )

Claims (1)

[Claims] (1) It has a voice input section that inputs the speaker's voice control command, and only when the voice control signal from the voice input section exceeds the voice cutting threshold level, the voice control signal is used to A voice recognition device for a vehicle that controls equipment is characterized in that a noise input section is provided separately from the voice input section, and the threshold level is changed based on a noise signal from the noise input section. Voice recognition device for vehicles.