JPH0497400A - Voice recognition device - Google Patents

Abstract

PURPOSE:To discriminate between commands from a driver's seat and a assistant' seat and to improve the reliability by providing a command effectiveness judgment part which judges the effectiveness of the commands. CONSTITUTION:The command effectiveness judgment part 6 judges the effectiveness of a command according to command contents from a voice recognition part and the seat discrimination result of the command from a command seat discrimination part to control various equipments. Namely, two microphones 1 are arranged symmetrically at the driver's seat, so an in-phase component detection part 3 detects the command of a driver's voice while it is superposed upon as in-phase components from the microphones. A phase-shifted in-phase component detection part 4 shifts the phases of respective microphones 1 so that the commands of a assistant's voice are superposed by the microphones since microphones 1 are arranged symmetrically at the assistant's seat. In this case, a command seal discrimination part 5 compares the output levels of the in-phase component detection part 3 and a phase-shifted in-phase component detection part 4. Consequently, a driver's command regarding vehicle control is discriminated as a command from the assistant's seat and the reliability of the voice discrimination device is improved.

Description

[Detailed Description of the Invention] [Summary] The present invention includes an in-phase component detection section that detects an in-phase component of an audio signal from microphones arranged almost symmetrically in the driver's seat, and a phase shifter for shifting the phase of the audio signal from the microphone by a certain amount. A phase shift in-phase component detection section that detects an in-phase component using a command, a command seat identification section that compares the output levels of the in-phase component detection section and a phase shift in-phase component detection section, and a command content and command seat identification section from the voice identification section. and a command validity determining unit that determines the validity of the command based on the seat identification result of the command from the driver's seat, the voice recognition device comprising: a command validity determining unit that determines the validity of the command based on the seat identification result of the command from the driver's seat, the voice recognition device comprising: This improves the reliability of the voice identification device.

FIELD OF INDUSTRIAL APPLICATION] The present invention relates to a voice recognition device for automobiles that substitutes for the hand and foot movements of a driver.

In particular, the present invention refers to a voice recognition device that improves reliability by identifying commands from a driver and a passenger seat.

1 Conventional technology] FIG. 10 is a diagram showing a conventional speech recognition device.

Note that similar components are represented by the same symbols or symbols throughout the drawings. Explain the configuration. This figure shows a microphone 1, a voice recognition unit 2 that recognizes voice by comparing an unknown pattern, which is a digital electrical signal obtained by converting a command voice signal from microphone 1, with a registered pattern, and The device 100 includes various devices 100 such as audio devices that are controlled according to commands recognized by the unit 2.

A detailed explanation of the speech recognition unit 2 will be given in the embodiment section.

Next, the operation will be explained. The driver in the driver's seat of the car can turn 1 audio on for microphone 1, or
When you command FF, the voice recognition unit 2 recognizes the command ``audio ON: or ni audio 0FF-'' and turns various devices such as audio on or off.
F works. In this way, the driver can operate various devices such as the audio system without taking his hands off the wheel while driving, and without taking his eyes off the direction of travel. Furthermore, such voice recognition devices could also be operated by commands from the passenger seat.

[Problem to be solved by the invention]

However, with conventional voice recognition devices, vehicle controls such as wiper mirrors can be operated by commands from the passenger seat, and there is a risk that operations important to driving may be performed against the driver's will due to commands from the passenger seat. There was a problem.

SUMMARY OF THE INVENTION In view of the above-mentioned problems, it is therefore an object of the present invention to provide a voice recognition device that identifies commands from a driver and an assistant to improve reliability of control of devices installed in an automobile.

Melt.

[Means to solve the problem]

FIG. 1 is a diagram showing the basic configuration of the present invention. In order to solve the above-mentioned problem, this figure shows that an in-phase component detection section 3, a phase shift in-phase component detection section 4, a command seat identification section 5, and a command validity judgment section 6 are installed in the speech recognition device. have The in-phase component detection section detects the in-phase component of the audio signal from the microphones arranged symmetrically to the driver's seat.

The phase soft in-phase component detection section 4 shifts the phase of the audio signal from the microphone 1 by a certain amount to detect an in-phase component.

The command seat identification section compares the output levels of the in-phase component detection section and the phase shift in-phase component detection section.

The command validity determining unit controls the various devices by determining the validity of the command based on the command content from the voice identifying unit and the command seat identification result from the command seat identifying unit. [Function] In FIG. 1, according to the voice recognition device of the present invention, the two microphones 1 are arranged symmetrically in the driver's seat in the in-phase component detecting unit 3, so that the driver's voice command is detected as the in-phase component of each microphone. It is detected as a superimposed image. On the other hand, in the phase shift in-phase component detection unit 4, since the microphone 1 is asymmetrically arranged on the passenger seat, the phase of each microphone 1 is shifted so that the voice command from the passenger seat is superimposed on each microphone. The command seat identification section 5 compares the output levels of the in-phase component detection section 3 and the phase shift in-phase component detection section 4. In other words, when a command occurs in the driver's seat, the in-phase component detection unit 3 detects the two microphones 1.
However, in the phase shift in-phase component detection unit 4, the in-phase components of the electric signals of the two microphones 1 are out of phase, so the in-phase components are not superimposed. In this case, since the output level of the in-phase component detection section 3 is higher, the command seat identification section 5 can distinguish the command from the driver's seat.

Conversely, when a command occurs in the passenger seat, the in-phase component detection unit 3
In this case, there is no superposition of the in-phase component, but the in-phase component is superimposed on the phase shift in-phase component detection section 4. In this case, since the output level of the phase shift in-phase component detection section 4 is higher, the seat identification section 5 of the command detects the passenger seat. Can be identified as a command. Based on the command content from the voice recognition unit 2 and the command seat recognition result from the command seat recognition unit 5, the command validity determination unit 6 determines whether there is a command related to vehicle control to be performed by the driver from the passenger seat. Eliminating this command ultimately improves the reliability of the speech recognition device.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings. FIG. 2 is a diagram showing a voice recognition device according to a first embodiment of the present invention. The configuration of this figure will be explained.

This diagram shows the vehicle interior 10 and the driver's seat 11 inside the vehicle interior 10.
, a passenger seat 12 in the vehicle interior 10, two microphones 1 arranged symmetrically with the driver's seat 11, and converting voice commands from the driver's seat 11 or the passenger seat into electrical signals; A/D (Analog-t) converts analog signals into digital signals.

Digital Converter) 7 and the A/
Speech recognition unit 2 that recognizes speech from the signal from the D converter 7
, an in-phase component detection unit 3 that detects in-phase components of the two microphones from the A/D converter 7, and a phase shifter that shifts the phase of the audio signals of the two microphones 1 by a certain amount to detect the in-phase component. In-phase component detection section 4, the in-phase component detection section 3 and the phase shift in-phase component detection section 4
a command seat identification unit 5 that compares the output level of the command, and a command validity that determines the validity of the command based on the command content from the voice identification unit 2 and the command seat identification result from the command seat identification unit 5. Based on the determination unit 6 and the determination of the validity determination unit 6 of the command, various devices that are installed in the automobile, such as the audio 100, near conditioning 101, telephone 102, navigation 103, and vehicle control equipment 104, are controlled. Control unit 8
, a speech synthesis section 9 for reporting the control contents of the various control sections 8, and a digital signal of the speech synthesis section 9 is converted into an analog signal by a D/A (Digital-to-Analog) converter (not shown). and a speaker 13 for reproducing sound from the power amplification stage. Note that the voice recognition unit 2, in-phase component detection unit 3, phase shift in-phase component detection unit 4, command seat identification unit 5, command validity determination unit 6, etc. are implemented using a DSP (D
(Signal Processor).

The speech recognition unit 2 will be explained. FIG. 3 is a diagram showing the speech recognition section of FIG. 2. This diagram shows a bandpass filter BP.
F20-1, 20-2, -, 20-n and each of the band pass filters BPF20-1, 20-2, -
, 20-n, an absolute value conversion unit 21 converts the electric signals of 20-n into absolute values.
-1, 21-2, . . . 21-n, and the unknown pattern which is the frequency distribution from the absolute value conversion unit 21-1, 21-2, . . . , 21-n is compared with the registered pattern. a recognition unit 22 that recognizes the command as the registered pattern as an unknown pattern;
The storage unit 23 includes a storage unit 23 that stores registered patterns necessary for recognition by the recognition unit 22. In the recognition unit 22, the unknown pattern and the registered pattern are compared using linear matching D P
(Dynamic Programming) Pattern matching such as munching is performed sequentially. The storage unit 23 stores audio 0N10 as a registered pattern.
Commands such as FF, near conditioning 0N10FF, telephone 0N10FF, navigation 0N10FF, and vehicle control such as set/resume of wipers, mirrors, etc. are stored.

Next, the in-phase component detection section 3 will be explained. FIG. 4 is a diagram showing the in-phase component detection section of FIG. 2. This figure includes an adder 30 that adds electrical signals from two microphones 1 located symmetrically with respect to the driver's seat, and a multiplier 31 that adjusts the output level of the adder 30. When a command generated at the driver's seat 11 is received by each microphone 1, the outputs of each microphone are superimposed at the output of the adder 30 because these electrical signals are in phase due to the arrangement. Conversely, for commands generated in the passenger seat 12, there is a phase shift in the electrical signals of the microphones 1, so the outputs of the respective microphones are not superimposed in the output of the adder.

The phase shift in-phase component detection section 4 will be explained.

FIG. 5 is a diagram showing the phase shift in-phase component detection section in FIG. 2. This figure shows a phase shifter 40-1 that shifts the phase of the electrical signal from each microphone 1.

40-2, and an in-phase component detection section 41 that detects in-phase components of the phase shift sections 40-1 and 40-2.

The phase shift sections 40-1 and 40-2 will be explained.

FIG. 6 is a diagram showing the phase shift section of FIG. 5. This figure shows a unit delay unit 400 that delays one sample and is connected in series.
-1, 400-2, 400-3, -, 4
00-(n-1), 400-n, and a multiplier 401-1 connected to the input side or output side of each unit delay device.
401-2. 401-3. 401-4-. 401
-(n-1) , 401-n, 401-(n
+1), an adder 402- that sequentially adds the outputs of each multiplier
1. 402-2゜402-3 , - , 402-(
n-2), 402-(n-1), 402-
n. The amount of phase shift can be determined by adjusting the multiplication coefficient of each multiplier. Note that the configuration of the in-phase component detection section 41 is the same as that of the in-phase component detection section 3, so a description thereof will be omitted.

Here, the two microphones 1 are symmetrical with respect to the driver's seat 11, but are asymmetrical with respect to the passenger seat 12. Therefore, since the command generated in the passenger seat 12 is out of phase with the electrical signals of the two microphones 1, the phase shift sections 40-1 and 40-2 adjust the multiplication coefficients of each multiplier to correct the phase shift. so that it disappears. In this way, the phase amount of the microphone 1 is adjusted by shifting by a certain amount, so that the in-phase component detection unit 41 of the phase shift in-phase component detection unit 4 outputs the sounds of the two microphones 1 in response to a command from the passenger seat 12. They begin to overlap.

Note that in response to a command generated from the passenger seat 12, the in-phase component detection unit 3 does not superimpose the voices of the two microphones at the output level. This is because there is a phase shift in each electrical signal.

Next, the command seat identification section 5 will be explained.

The seat identification section 5 of the command is composed of a comparator.

If the output signal of the in-phase component detector 3 is Va and the output signal of the phase shift in-phase component detector 4 is vb, for example, Va≧
When the signal is Va<Vb,
When the "L (Low)" signal is output. When the command seat identification section 5 outputs the "°H" signal, Va≧
Vb, the command is identified as being generated from the driver's seat 11 because the output level of the electrical signal via the in-phase component detection section 3 is high.

In the opposite case, it is determined that the noise is coming from the passenger seat 12.

Next, the command validity determining section 6 will be explained. 7th
The figure is a diagram showing the determination of the validity of the command based on the voice recognition result in FIG. 2. The command validity determining unit 6 determines that the content of the recognized command from the voice recognition unit 2 is audio 0N10FF, near conditioning 0N10FF, navigation 0N10FF, and vehicle control set/resume, and that the seat identification unit 5 of the command indicates that the command is not operated. If the command is identified as a command issued from the seat 11, the command is determined to be valid. Also, the commands are audio 0N10FF, near conditioning 0N10FF, navigation 0N10F.
If the command is F and is issued from the passenger seat, the command is determined to be valid. However, if the command is vehicle control set/resume and is uttered from the passenger seat, it is determined to be invalid. In this way, in terms of operation control,
For matters that should be decided by the driver, the voice from the passenger seat can be prevented from interfering with the driver's decision, and for other matters, various services can be provided in response to commands from the passenger seat. becomes possible.

Next, a series of operations of this embodiment will be explained. FIG. 8 is a series of operation flowcharts according to this embodiment. In this figure, as a command, an analog electrical signal from a microphone 1 is converted into a digital signal (step 1). The voice recognition unit 2 recognizes whether the unknown pattern of the digital signal of the command is a registered pattern (step 2). If the unknown pattern that is a command is a registered pattern,
The in-phase component detection unit 3 determines whether the command was issued by the driver's seat operator 1 (step 3). If the command is from the driver's seat 11, the corresponding 0N10FF control of the audio 100, near conditioning 101, telephone 102, navigation 103 is performed via the various control units 8.
Vehicle control 104 cents/resium control is executed (
Step 4). Furthermore, the control contents are converted into speech by the speech synthesis section 9 via the various control sections 8, and the D/
A (Digital-t.

Analog Converter), a power amplifier, etc., the signal is reproduced into audio by the speaker 13, and is reported to the driver's seat 11 (Step 5). If it is determined in step 3 that the command is from the passenger seat 12, the content of the command is determined to be a vehicle control command (step 6), and if it is not vehicle control but audio 0N10FF, the process proceeds to step 4.
That control is executed. If the content of the command is a command for vehicle control, the control will not be executed and a report will be made that the control cannot be executed (step 5).
.

FIG. 9 is a diagram showing a voice recognition device according to a second embodiment of the present invention. The different components in FIG. 2 are rear dust 13 and 14 in the vehicle interior 10, a microphone 1-1 installed near the rear dust 14 of the driver's seat 11, and a phase shift in-phase component detection unit 4 for the passenger seat 12. -1, a phase shift in-phase component detection unit 4-2 for the rear dust 13 and a rear dust 14
The output level of the in-phase component detection unit 3 and the phase shift in-phase component detection unit 4-3
1.4-2.4-3 The command seat identification section 5' compares the output level with the maximum output level. Phase shift of the rear dust 13 The in-phase component detection unit 4-2 takes into account the asymmetry between the rear dust 13 and the microphone 1 and adjusts the phases of the two microphones 1 to match. The shift in-phase component detection unit 4-3 of the rear dust 14 detects the difference between the rear dust 14 and the microphone 1.
Since the commands are symmetrical to the above, there is a risk of confusion with the command from the driver's seat 11. Therefore, the asymmetry between one microphone 1 and the microphone 11 is taken into account and the phases of the two microphones 1 are adjusted so as to match. Therefore, the output level of the rear dust 14 command can be increased in the in-phase component detection section 3 and the phase shift in-phase component detection section 4-3, and can be distinguished from the driver's seat 11 command. In the command seat al separate section 5', the output level of the in-phase component detection section 3 is phase-shifted.In-phase component detection section 4-1.4-2.4-3
If it is larger than that of Lenomo, it is identified as a command from the driver's seat 11, and on the other hand, the output level of the in-phase component detection unit 3 is phase-shifted. If it is smaller than , it is identified as a command from a source other than the driver's seat 11 .

〔Effect of the invention〕

As explained above, according to the present invention, there is provided an in-phase component detection section that detects an in-phase component of an audio signal from a microphone disposed symmetrically to the driver's seat, and an in-phase component detection section that detects an in-phase component of an audio signal from a microphone arranged symmetrically to the driver's seat. Compare the output levels of the phase shift in-phase component detection unit that detects the phase shift in-phase component detection unit and the output level of the in-phase component detection unit and the phase shift in-phase component detection unit. Since a command validity judgment unit is provided to judge the validity of the command based on the seat identification result of the voice recognition device, commands related to the driver's vehicle control can be identified as commands from the passenger seat. Confidence improves.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the principle configuration of the present invention, FIG. 2 is a diagram showing a voice recognition device according to a first embodiment of the present invention, FIG. 3 is a diagram showing the voice recognition unit of FIG. 4 is a diagram showing the in-phase component detection section in FIG. 2, FIG. 5 is a diagram showing the phase shift in-phase component detection section in FIG. 2, FIG. 6 is a diagram showing the phase shift section in FIG. The figure is the second
FIG. 8 is a flowchart of a series of operations according to this embodiment; FIG. 9 is a diagram showing a voice recognition device according to a second embodiment of the present invention. FIG. 10 is a diagram showing a conventional speech recognition device. In the figure, 1...microphone, 2...speech recognition unit, 3...
- In-phase component detection section, 4... Phase shift in-phase component detection section, 5... Command seat identification section, 6... Command validity judgment section.

Claims (1)

[Claims] 1. At least a microphone installed in a car (
1) and a voice recognition unit (2) that recognizes voice commands from the microphone (1) and controls various devices installed in the vehicle, the voice recognition device is approximately symmetrical to the driver's seat. an in-phase component detection unit (3) that detects an in-phase component of an audio signal of a microphone (1) placed in the microphone (1); a component detection unit (4); a command seat identification unit (5) that compares the output levels of the in-phase component detection unit (3) and the phase shift in-phase component detection unit (4); and the voice identification unit (2). a command validity determining unit (6) for controlling the various devices by determining the validity of the command based on the command content from the command and the command seat identification result from the command seat identifying unit (5); A voice recognition device comprising: