JPS5870284A - 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 that prevents voice registration under high noise levels and reduces erroneous voice recognition.

Recently, a voice N111i device has been developed that distinguishes human commands by voice and determines their contents, and is being applied to various industrial fields60. It is necessary to memorize the characteristics of the voice. FIG. 1 is a four-dimensional diagram of a conventional speech recognition device of this type, where l is a microphone that inputs superimposed sound information such as voice commands and noise, and 2 is a sound information signal output from microphone 1 ( 6 a level detector which temporally averages the level of 100); 3 a threshold setter which sets a threshold value as a reference for determining whether or not it is a voice; 4 a signal from the level detector 2 (200); and threshold (300)
Comparing with
The judgment unit b outputs a prohibition signal (400) when the threshold value (300) is exceeded over a period of more than mm ). A display section for displaying certain things, 6
is the voice H111 part, 7 in which the voice is not registered or recognized when the prohibition signal (400) from the judgment unit 4 is input.
6 is a command switch for switching the operation of the voice recognition section 6 to the registration mode or the recognition mode.

In the above voice recognition device, K for voice registration is a command switch, and when 7 is switched to the registration side and the voice 11w14 unit 6 is set to the registration mode and a command to be registered is inputted by voice, the voice is superimposed with the noise and the sound information is transmitted. as microphone p phone I
It will be registered with K manpower. However, the six-tone information signal (100) outputted from the microphone siphon 1 becomes a temporally averaged signal (200) by the level detector 2, and is compared with a fixed threshold value (300) in the determining section 4.

As a result, the signal (200) is
If the threshold value (300) is exceeded for more than 2 seconds, the determination unit 4 determines that the microphone input is noise and outputs a prohibition signal (400) indicating that registration is not possible, and the sound 11!
! The audio is not registered by the display section 5.j section 6. A message indicating that registration is not possible is displayed.

Next, when performing voice recognition, the command switch 4 notch is switched to the recognition side, the voice recognition section 6 is pressed to the recognition mode, and the desired command voice among the previously registered commands is input. In this case, when a lot of noise is input to the miter, the operation of the judgment unit 4 is similar to the operation described above when determining whether or not to register, and when the prohibition signal (400) from the judgment unit 4 is output, The voice aii is not performed by the voice command unit 6, and the display unit 5 displays a message indicating that the recognition is not possible.

In such speech recognition devices, the level of the input sound information signal is compared with the same threshold value during both registration and recognition, and this threshold value can be used for speech recognition even when the noise level is relatively high, such as when driving. Because the noise level is set relatively large so that the noise level is relatively large, registration is possible even when the noise is relatively loud, resulting in a lower recognition rate such as erroneous ams.

The present invention has been made in consideration of the above points, and in order to reduce the msIm of the voice by setting K so that voice cannot be registered under a large noise level6, the reference level for determining the noise level is set at the time of voice registration. This is made smaller than during speech recognition.

The present invention will be explained below based on the drawings.

FIG. 2 is a Plutz diagram showing an embodiment of the vehicle voice recognition device according to the present invention. In the figure, the same reference numerals as in Figure 1 indicate the same components, and the six points in which this embodiment differs from the conventional example shown in Figure 1 are the six threshold settings I! The point is that 6 threshold value settings lI8 having two threshold values of different sizes are provided in place of #3. This threshold setter 8
is designed to be switched in conjunction with the operation of the command switch 7, and when the command switch 7 is turned to the recognition side, a large threshold is output, and when it is turned to the registration side, a small threshold is output.

Next, the operation of this embodiment will be explained.

First, when performing a voice edict, turn command switch 7 to
Lllllll! signal from command switch 7 (
700), the voice recognition s6 is set to recognition mode, and a desired command among the previously registered commands is inputted by voice. At this time, the threshold *a constant-8 is the same as the conventional example from when the 0 sound information that outputs a large threshold (goo) in conjunction with the operation of the command switch 7 is input to the microphone l until it is manually input to the judgment unit 4. When the signal (200) from the level' detection unit 2 exceeds the threshold (SOO) for a period of time or more, the determination unit 4 determines that it is noise, and the determination unit 4 outputs a prohibition signal (400). As a result, the voice recognition unit 6 does not glIll the voice, and the display unit 5
11111Il is not allowed.

Next, when registering the voice, the command switch 7 is switched to the registration side, the voice 11114 section 6 is placed in the registration mode based on the signal (7θ0) from the command switch 7, and the command to be registered is manually input to the microphone 1 by voice. At this time, the threshold value setter 8 outputs a small threshold value (80σ) in conjunction with the operation of the command switch 7. Note that from judgment s4 onwards, the signal from the level detection unit 2 [[
(200) for an amount of time or more than the threshold (80
0'), the determining unit 4 determines that it is noise, and the determining unit 4 outputs a prohibition signal (400).

As a result, the voice recognition unit 6 does not register the voice, and the display unit 5 displays a message indicating that registration is not possible. In this way, when registering a voice, the threshold value that is compared with the level of the sound information signal (100) in the judgment unit 4 is made smaller than when introducing a voice, so that not only a single loud sound but even the slightest noise cannot be registered. .

As explained above, in the present invention, when the level of input sound information is higher than a predetermined reference level, in the audio Ba1l device N which does not perform audio registration or audio edict, when registering audio, it is possible to Since the reference level is made smaller, recognition is possible even when the noise level is high, and registration is possible only when the noise level is low. B can reduce recognition.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a conventional vehicle audio lIII1m device, and FIG. 2 is a diagram of a vehicle audio introduction system according to the present invention.
FIG. 2 is a block diagram of the device. l・°・Microphone 4 phones, 2...Repé” Examination B Department,
s * s...Threshold value setter, 4...Judgment unit, 5.-
Display unit, 6... 88 audio units, 7... Command switch. Patent applicant Hiroo Tsuchi Suzuki, Patent Attorney, Nissan Motor Co., Ltd.

Claims (1)

[Claims] A voice recognition device that does not perform voice registration or voice recognition when the level of sound information input to a microphone is higher than a predetermined reference level is provided with means for setting the reference level lower during voice registration than during voice recognition. Characteristic vehicle audio 111m1 device.