JPH05231922A - Display device of sound pressure level for sound recognizing apparatus - Google Patents

Abstract

PURPOSE:To surely improve the recognizing efficiency of sounds of a sound recognizing apparatus. CONSTITUTION:This display device has a sound pressure level ratio calculating means 5 for obtaining the ratio of the sound pressure level input to a first sound detector 1 which receives sound signals and the sound pressure level input to a second sound detector 2, and a display means 6 for displaying the ratio of the sound pressure level.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sound pressure level indicator for a voice recognition device, which is extremely suitable when the voice recognition device is used in an environment with noise or other noise.

[0002]

2. Description of the Related Art When using a voice recognition device, it is necessary that the input sound pressure level is appropriate in order to improve the recognition performance of the device. If the input sound pressure level is too low, it becomes difficult for the device to detect the voice from the ambient noise, resulting in an increase in false recognition due to a detection error in the input voice section, or no response to the input voice. It becomes the state of. On the other hand, if the input sound pressure level is too high,
Since the signal is distorted in the sound receiver or the amplifier after the sound receiver, the feature amount of the sound is deformed, and as a result, the possibility of erroneous recognition by another sound increases. In order to avoid the above inconveniences, some conventional voice recognition devices have a means for displaying the sound pressure input to the sound receiver for one voice input sound receiver. Be done. As a display method, there are a meter with an analog needle display, a method in which a plurality of light emitting diodes are arranged, and the like.

FIG. 4 shows a conventional example of a sound pressure level indicator using a plurality of light emitting diodes. In this conventional example,
If there is no voice input and only ambient noise with a low sound pressure level, one or two green diodes from the left end light up to react to ambient noise, and if there is an input voice, the left will be left according to the sound pressure. The third or fourth red diode lights up. Since there are individual differences in the vocal sound pressure, a variable resistor is usually attached so that the display sensitivity can be controlled, and the user adjusts the variable resistor volume according to his or her sound pressure level. For example, the red diode is 1
Use them by lighting them individually.

[0004]

In the conventional sound pressure level display, when the ambient noise level input to the sound receiver is increased, the display responds to noise and even when no sound is input. , Display as if voice is being input. That is, in this conventional example, the red diode is turned on. Further, there is a method of adding an automatic gain control function to an amplifier as a means for reducing signal distortion due to an excessive input sound pressure level. Even if this method is used, in the case where there is only one sound receiver. Is not an essential solution because it works the same for both noise and speech. Such a situation means that the speaker loses a proper guideline for displaying the vocal sound pressure level, which causes confusion. Further, generally, the rate of increase in the vocal sound pressure level is small with respect to the rate of increase in the ambient noise level, so when the noise level increases, the voice signal-to-noise signal level (SN) of the acoustic signal input to the voice recognition device is increased. There is a problem that the ratio becomes small, and as a result, the voice recognition performance deteriorates.

Therefore, an object of the present invention is to keep the state of sound pressure level display constant even in an environment in which the ambient noise level fluctuates greatly, and the speaker can see the display and have a substantially constant voice signal-to-noise level. An object of the present invention is to provide a sound pressure level indicator for a voice recognition device, which can improve voice recognition performance by uttering while maintaining a signal (SN) ratio.

[0006]

In order to solve the above problems, the present invention provides a first sound receiver for receiving a voice signal and a first sound receiver for receiving a noise signal in the vicinity of the first sound receiver. No. 2 sound receiver, sound pressure level ratio calculating means for obtaining a ratio between the sound pressure level input to the first sound receiver and the sound pressure level input to the second sound receiver, and the sound pressure The gist is to have display means for displaying the ratio of the sound pressure levels obtained by the level ratio calculation means.

[0007]

In the above construction, even in an environment in which the ambient noise level fluctuates greatly, the display means does not depend on the noise level itself, and the SN ratio level at the time of voice input is visualized and displayed. .. Therefore, the guideline about the appropriate utterance level is always shown to the user, and the user can surely improve the voice recognition performance by uttering so that the display state is kept almost constant. Is possible.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS.

First, the structure of the sound pressure level indicator for the voice recognition device will be described. In FIG. 1, reference numeral 1 is a first sound receiver for the purpose of receiving a voice signal, and the directivity main axis (reference axis) is set so that the sensitivity is maximum with respect to the input voice. .. Reference numeral 2 is a second sound receiver for the purpose of receiving noise in the vicinity of the first sound receiver 1, and it is desirable that a noise signal having a characteristic as close as possible to that of the first sound receiver 1 can be received. On the other hand, the reference axis is set so that the sensitivity to the audio signal is as low as possible.
Reference numerals 3 and 4 are amplifiers for input signals to the first and second sound receivers 1 and 2, respectively. Reference numeral 5 is a sound pressure level calculation unit as a sound pressure level ratio calculation means, and here, the respective output signals of the amplifiers 3 and 4 are input, and input to the first sound receiver 1 and the second sound receiver 2. The sound pressure level of the signal and both receivers 1,
The ratio of the input sound pressure level to 2 is calculated for each short time. Reference numeral 6 denotes a display unit as a display unit, which visualizes and displays the sound pressure level ratio output from the sound pressure level calculation unit 5 as an input signal. The display method is the example shown in FIG. Etc. can be applied.
Reference numeral 7 is a voice recognition unit, which recognizes the voice signal by using the output signal of the amplifier 3 as an input. However, when noise removal is performed in the voice recognition unit 7, the output signal of the amplifier 4 may be input at the same time. Reference numeral 8 denotes a gain control unit, which is a sound pressure level calculation unit 5
The input sound pressure level to the first sound receiver 1 calculated in step 1 is used as an input, and a gain control signal is sent to the amplifiers 3 and 4 according to this value.

FIG. 2 shows a sound pressure level calculating section 5 and a display section 6.
2 shows the details of the internal configuration of the. First, 51 in the sound pressure level calculation unit 5 is a first sound pressure level calculation unit that calculates the output sound pressure level of the amplifier 3 every short time, and 52 is a first sound pressure level calculation unit that calculates the output sound pressure level of the amplifier 4 every short time. The two sound pressure level calculation unit, 53 is a sound pressure level ratio calculation unit that calculates the ratio between the first sound pressure level and the second sound pressure level. Further, the display unit 6
61 is a switch, 62 is a reference voltage storage unit, 63
Is a display circuit drive unit, and 64, 65, 66 and 67 are light emitting diode display units.

Next, the operation of the sound pressure level indicator for the voice recognition device constructed as described above will be described.

First, the setting of the reference sound pressure on the display unit 6 will be described.

The ratio of the sound pressure levels input to the first and second sound receivers 1 and 2 in the state where there is no voice input and only noise signals are received depends on the installation conditions of both sound receivers 1 and 2. Dependent. Therefore, before using the voice recognition device, it is necessary to store the value of the sound pressure level ratio when there is no voice input and adjust the display unit 6 using this value as the reference sound pressure. In order to properly carry out this, the display unit 6 has the structure shown in FIG.

The adjusting method will be described below. Now, assuming that only the noise signal is input, the sound pressure level ratio calculation unit 53 calculates the sound pressure level ratio Vref in this state. At this time, the switch 61 is assumed to be connected to the reference voltage storage unit 62 side, and the calculated sound pressure level ratio V
The ref is input to the display circuit drive unit 63 via the reference voltage storage unit 62. The display circuit drive section 63 is connected to the light emitting diode display sections 64, 65, 66, and 67,
As shown in Table 1, the light emitting diode 64 receives the input voltage Vd
There lights if V ₆₄ or more, below the light emitting diode 67 and similarly if are driven if the input voltage Vd is V ₆₇ or more, for example V ₆₄ ≦ Vref <V ₆₅ and so as to display circuit driving unit By adjusting the resistance value of 63 and the like, in a state in which no sound is input thereafter, the light emitting diode 64 is generally lit even if the ambient noise level changes. This is because, as described above, the first and second sound receivers 1 and 2 always receive the noise signals having substantially the same characteristics, so that the absolute values of the noise levels input to the sound receivers 1 and 2 are the absolute values. This is because it is considered that the fluctuation of Vd, which is the relative level ratio of both inputs, is very small even when f. By setting V ₆₄ and V ₆₅ in this way, it is possible to set V ₆₆ and V ₆₇ by assuming the gain width so that they are sequentially lit when Vd becomes a certain value.

[0015]

[Table 1] Further, when the ambient noise level is very low, the influence of the sensitivity difference between the first and second sound receivers 1 and 2 and the respective amplifiers 3 and 4 becomes large, and as a result, the sound pressure level ratio Vd is calculated. It is considered that the error increases and affects the display state. In order to prevent this, the sound pressure level ratio calculation unit 53 constantly monitors the sound pressure level to the first sound receiver 1 which is input from the first sound pressure level calculation unit 51, and the value is a constant value. In the following cases, the sound pressure level ratio calculation unit 53 may output a constant value (for example, the same value as Vref).

Next, the operation after setting the reference sound pressure will be described.

When the setting of the reference sound pressure is completed by the above method, the switch 61 is switched to change the sound pressure level ratio calculating section 53.
By directly connecting the display circuit driving unit 63 with the display circuit driving unit 63, the input sound pressure level can be displayed during the operation of the voice recognition device. In the middle, for example, two sound receivers 1, 2
When there is a change in the acoustic environment of the two sound receivers 1 and 2, such as when the arrangement is changed, the reference sound pressure may be set again.

In the operating state after the setting, the first sound pressure level calculation section 51 always calculates the input sound pressure V ₁ to the first sound receiver 1, and the value is output to the gain control section 8. .. In the gain control unit 8, with respect to the input sound pressure V _1, by setting the gain control signal A ₁ and A ₂ into amplifier 3 and 4 as shown in FIG. 3, the first receiving a speech input It is possible to reduce the distortion of the signal when there is an excessive input to the sound device 1. Further, when the input sound pressure V ₁ has a very small value, it is considered that the ambient noise level is very small, and in that case, as in the case of setting the reference sound pressure,
The sound pressure level ratio calculation unit 53 may output a constant value.

Now, V ₁ is the sound pressure level of the signal input to the first sound receiver, V ₂ is the sound pressure level of the signal input to the second sound receiver, S _{1 is} the first sound reception level. Sound pressure level of voice signal component input to sound device S ₂ : Sound pressure level of voice signal component input to second sound receiver N ₁ : Noise signal component input to first sound receiver sound pressure level N _2: the sound pressure level of the noise signal component is input to the second sound receiving unit, if _{S 2 «N 2, N 1 ≒} N 2, V 1 = S 1 + N 1 V 2 ≒ Since S ₂ + N ₂ ≈N ₂ ≈N ₁ , the sound pressure ratio Vd between V ₁ and V ₂ is Vd = V ₁ / V ₂ ≈ (S ₁ / N ₁ ) +1. That is, Vd represents the level ratio (SN ratio) between the voice and the noise input to the first sound receiver 1, and therefore the operation of the display does not depend on the noise level itself, and when the voice is input, The SN ratio level will be displayed.

In the conventional method, since the display displays the value of V ₁ itself, it reacts even when no voice is input when the ambient noise level is high. Therefore, there is a fear that the user is confused. On the other hand, in this embodiment, since the SN ratio of the input voice is visualized, the user is always provided with a guideline regarding an appropriate utterance level, and a psychological effect on the user can be expected. ..

On the other hand, from the viewpoint of voice recognition performance, it has been reported that the maximum recognition performance is obtained when the SN ratio of the voice input to the voice recognition device is the same during voice registration and during recognition ( Reference "Kitamura, Hayahara," Noisy word recognition using dynamic features of speech ", Proceedings of ASJ, p.
p. 59-60, (October 1988) "and many others). Therefore, if the user of the voice recognition device uses the method of this embodiment to speak so as to keep the state of the display constant, the SN ratio of the voice signal input to the voice recognition device approaches the desired value, and As a result, a reliable improvement in the voice recognition performance of the device can be expected.

[0022]

As described above, according to the present invention,
Since the ratio of the sound pressure level input to the first sound receiver for receiving the audio signal and the sound pressure level input to the second sound receiver for receiving the noise signal is displayed on the display means, Even in an environment in which the ambient noise level fluctuates greatly, the display means visualizes and displays the SN ratio level when a voice signal is input, without depending on the noise level itself. Therefore, the user can surely improve the voice recognition performance by uttering so as to keep the display state substantially constant.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a sound pressure level indicator for a voice recognition device according to the present invention.

FIG. 2 is a block diagram showing an internal configuration of a sound pressure level calculation unit and a display unit in FIG.

FIG. 3 shows a first input sound pressure level V ₁ in the present embodiment.
It is a figure which shows the relationship of the gain of the amplifier with respect to.

FIG. 4 is a diagram showing a conventional sound pressure level indicator using a plurality of light emitting diodes.

[Explanation of symbols]

 1 1st sound receiver 2 2nd sound receiver 5 Sound pressure level calculation part (sound pressure level ratio calculation means) 6 Display part (display means) 7 Speech recognition part 63 Display circuit drive part 64, 65, 66, 67 Light emitting diode display

Claims (1)

[Claims] 1. A first sound receiver for receiving an audio signal, a second sound receiver for receiving a noise signal in the vicinity of the first sound receiver, and an input to the first sound receiver. And a sound pressure level ratio calculating means for obtaining a ratio between the sound pressure level to be input and the sound pressure level input to the second sound receiver, and a ratio of the sound pressure levels obtained by the sound pressure level ratio calculating means. A sound pressure level indicator for a voice recognition device, comprising: a display means.