JPS63118197A - Voice detector - Google Patents

Abstract

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

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a voice detection device for accurately detecting a voice signal section from a signal mixed with noise in a voice transmission device such as a telephone.

BACKGROUND OF THE INVENTION In recent years, speakerphones and telecombustion systems have become widespread, and voice switches are used in these devices to prevent nodding and eliminate indoor noise. Digitalization of voice transmission is also progressing, and research and development is being conducted to improve transmission efficiency by distinguishing between non-voice sections and voice sections, and encoding and transmitting signals only for the voice sections.

Furthermore, voice recognition technology for controlling various devices using human voice is rapidly progressing.

In order to improve the performance of these devices, a voice detection device that does not malfunction due to noise is strongly desired and has been developed.

Hereinafter, a conventional voice detection device will be explained using FIGS. 3 to 5 with reference to the drawings.

In FIG. 3, 31 is a signal input terminal, 32 is a rectifying and smoothing circuit for rectifying and smoothing this signal to obtain an input signal amplitude value, and 33 is a first dip hold for obtaining a minimum noise amplitude value from this input signal amplitude value. 34 is a peak hold circuit that calculates the envelope of the amplitude value peak from the input signal amplitude value, 35 is a second dip hold circuit that calculates the maximum noise amplitude value from the envelope of this amplitude value peak, and 36 is the maximum noise amplitude A threshold calculation circuit 37 calculates a threshold value from the value and the minimum noise amplitude value output from the first dip hold circuit 33;
a determination circuit that compares the input signal amplitude value outputted from the input signal amplitude value and determines that an audio signal is present when the input signal amplitude value exceeds a threshold; 38 is an output terminal for this determination signal; be.

FIG. 4 shows signal waveforms at various parts of the conventional example shown in FIG. 3, and the operation of the conventional example shown in FIG. 3 will be explained using this diagram.

In FIG. 4, 41 is the waveform of the input signal at point A in FIG. 3, 42 is the waveform of the input signal amplitude value at point B in FIG. 3, 43 is the waveform of the minimum noise amplitude value at point 0, and 44 is the waveform of D 45 is the waveform of the maximum noise amplitude value at point E, 46 is the threshold waveform at point F, and 47 is the waveform of the determination signal at point 0.

First, the input signal 41 shown in FIG. 4 is input to the input terminal 31 shown in FIG. This input signal 41 is in a noise section up to time To-TI and does not contain any audio signal. The audio signal section is from time T1' to time T2. Time T2-T
3 is a noise section.

As described above, the noise contained in the voice signal of a telephone or the like includes electrical noise of equipment and lines, and noise caused by room noise when voice is collected. Noise noise is usually louder than electrical noise. Noise can be generated from air conditioning equipment, ventilation fans, automobiles, factory equipment, etc. The difference between noise caused by these types of noise and electrical noise is that its amplitude fluctuates greatly over time. . The waveform of the input signal 41 is mainly mixed with noise due to this noise, and the amplitude of the noise varies greatly over time.

This input signal 41 is rectified and smoothed by the rectification and smoothing circuit 32 shown in FIG. 3, resulting in an input signal amplitude value 42 shown in FIG. The input signal 41 is the result of adding stationary noise with a substantially constant amplitude, non-stationary noise with a varying amplitude, and an audio signal. The minimum value of this input signal amplitude value 42 indicates the amplitude value of stationary noise. Therefore, the amplitude value 42 is processed by the first dip-hold circuit 33 to obtain the minimum noise amplitude value 43 which is the stationary noise amplitude. On the other hand, the input signal amplitude value 42 is also input to a peak hold circuit 34 whose attenuation time constant is set relatively short, and an envelope 44 of the amplitude value peak is obtained. If there is a relatively long noise section from the envelope 44 of this amplitude value peak to time To, such as TI and time T2-T3,
It can be seen that this value is sufficiently lower than the peak value of the voice and traces the maximum value of the noise amplitude. In real conversations, there are always pauses in the voice. Therefore, the minimum value of the envelope of this amplitude value peak is set to the second dip hold circuit 36.
The maximum noise amplitude value 45 can be obtained.

Next, the maximum noise amplitude value 45 and the minimum noise amplitude value 43 are input to a threshold calculation circuit 36, and a threshold value 46 is calculated. This threshold calculation circuit 36 calculates the threshold using the following equation.

Threshold value = 2 x maximum noise amplitude value - minimum noise amplitude value (1) In this way, when the noise fluctuation is large, the minimum noise amplitude value and the maximum noise amplitude value are large. In contrast, the threshold value is set to be significantly larger than the maximum noise amplitude value, that is, to a threshold value with a large margin.

When the noise fluctuation is small, a threshold value close to the maximum amplitude value with a small margin is automatically set. The threshold value 46 is always set to a value higher than the input signal amplitude value 42 from time TO to T1 and from time T2 to T3.

The determination circuit 37 compares this threshold value 46 with the input signal amplitude value 42 and determines that the input signal amplitude value 42 is equal to the threshold value 46.
If it exceeds , it is determined that an audio signal is present, and a determination signal 47 is output from the output terminal 38 to the external device. This determination signal 47 is ON between times T1 and T2, that is, it is determined that a voice signal is present, and it can be seen that voice detection is performed accurately even in the presence of noise.

Problems to be Solved by the Invention However, in the voice detection device having the conventional configuration described above, there is a problem in that false detection occurs when detecting voice in a signal switched by a voice switch, such as an incoming signal of a hands-free telephone. It had a point. That is, in detecting voice in a received signal, the received signal is switched by the voice switch of the other party, so the voice detection device determines that the noise level has changed significantly. The influence of this audio switch will be explained using the signal waveform diagram in FIG.

In FIG. 6, 61 is the waveform of the switched receiving signal input to the input terminal 31 in FIG. 3 at the human point in FIG. 3, and 62 is the input signal amplitude value obtained by rectifying and smoothing this receiving signal at point B. 53 shows the waveform of the threshold value at point F. As shown in the threshold value 53 in FIG.
At the time of switch 08 of 2 to T3, the noise level becomes large, and a threshold value corresponding to this is set. At time T1 to T2 when the switch is off, it is naturally assumed that the noise level has become very low, and a low threshold value is set. In the conventional voice detection device, the threshold follows quickly when the noise level decreases, and slowly when the noise level increases, as can be seen from the comparison between the threshold 53 and the input signal amplitude value 62 in FIG. However, there was a drawback that noise was mistakenly detected as voice immediately after switching from off to on.

In view of the above-mentioned problems, the present invention prevents the threshold value from lowering when the audio switch is turned on, and always makes a determination between noise and audio based on the threshold value when the audio switch is ON, so that the signal switched by the audio switch is The purpose of the present invention is to provide a voice detection device that can accurately detect voice even when the user is using the computer.

Means for Solving the Problems To achieve this objective, the voice detection device of the present invention comprises:
A rectifying and smoothing circuit that calculates an input signal amplitude value from an audio signal mixed with noise, a first dip-hold circuit that calculates a minimum noise amplitude from this input signal amplitude value, and an envelope of an amplitude value peak from the input signal amplitude value. , a first peak hold circuit, a second dip hold circuit that calculates the maximum noise amplitude value from the envelope of this amplitude value peak, and an instantaneous threshold value calculated from the maximum noise amplitude value and the minimum noise amplitude value. A second peak hold circuit calculates the maximum threshold value from this instantaneous threshold value, and compares this maximum threshold value with the input signal amplitude value to determine the input signal amplitude value. and a determination circuit that determines that an audio signal is present when the maximum threshold value is exceeded.

Effect: With this configuration, the speed at which the threshold value decreases when the audio switch is turned 077 can be delayed, and even when the audio switch is OFF, it is possible to maintain the threshold value when the audio switch is turned ON immediately before turning OFF. As a result, it is possible to realize a voice detection device that does not erroneously detect noise as voice, even from signals switched by the voice switch, as in the conventional example.

EXAMPLE An example of the present invention will be described below with reference to the drawings.

FIG. 1 shows a block diagram of a voice detection device according to an embodiment of the present invention.

In FIG. 1, 11 is a signal input terminal, 12 is a rectifying and smoothing circuit for rectifying and smoothing this signal to obtain an input signal amplitude value, and 13 is a first dip hold for obtaining a minimum noise amplitude value from this input signal amplitude value. 14 is a first peak hold circuit that calculates the envelope of the amplitude peak from the input signal amplitude value; 15 is a second dip hold circuit that calculates the maximum noise amplitude from the envelope of the amplitude peak; 16 is this circuit; An instantaneous threshold calculation circuit 17 calculates an instantaneous threshold from the maximum noise amplitude value and the minimum noise amplitude value output from the first dip-hold circuit 13; The second peak hold circuit to seek,
18 compares this maximum threshold value with the input signal amplitude value output from the rectifying and smoothing circuit 12, and determines that an audio signal exists if the input signal amplitude value exceeds the maximum threshold value. A determination circuit 19 is an output terminal of this determination signal.

The difference between the embodiment of the present invention shown in FIG. 1 and the conventional example shown in FIG. They have the same configuration and operation.

That is, the input terminal 11 in FIG. 1 is the input terminal 3 in FIG.
1, the rectifying and smoothing circuit 12 is connected to the rectifying and smoothing circuit 32, the first dip hold circuit 13 is connected to the first dip hold circuit 33, the first peak hold circuit 14 is connected to the peak hold circuit 34, and the second dip hold circuit The circuit 15 is a second dip hold circuit, and the instantaneous threshold calculation circuit 1
6 corresponds to the threshold calculation circuit 36, the determination circuit 18 corresponds to the determination circuit 37, and the output terminal 19 corresponds to the output terminal 38.

FIG. 2 shows signal waveforms at various parts of the embodiment shown in FIG. 1, and the operation of this embodiment will be explained using this diagram.

In FIG. 2, 21 is the waveform of the input signal at point M in FIG. 1, 22 is the waveform of the input signal amplitude value at point B in FIG. 26 shows the waveform of the maximum threshold value at a point, and the waveform of the determination signal at two points.

First, the input signal 21 shown in FIG. 2 is input to the input terminal 11 shown in FIG. This input signal 21 is the rectifying and smoothing circuit 1 in FIG.
2 to obtain the input signal amplitude value 22 shown in FIG.

Next, this input signal amplitude value 22 is input to the first dip hold circuit 13, the peak hold circuit 14, the second dip hold circuit 16, and the instantaneous threshold calculation circuit 16, and the instantaneous threshold value 23 is calculated. .

The determination circuit 18 compares this maximum threshold value 24 with the input signal amplitude value 22, and determines that an audio signal is present when the input signal amplitude value 22 exceeds the maximum threshold value 24, The determination signal 25 is outputted from the output terminal 39 to an external device.

As described above, according to this embodiment, by providing the second peak hold circuit 17 that calculates the maximum threshold value from the instantaneous threshold value, the speed at which the threshold value decreases when the audio switch is OFF can be slowed down. Can be done. Therefore, even when the audio switch is in the OFF state, the threshold value at the ON time immediately before the OFF state can be maintained. As a result, it is possible to realize a voice detection device that does not erroneously detect noise as voice, even from signals switched by the voice switch, as in the conventional example.

Effects of the Invention The present invention provides a second method for determining the maximum threshold value from the instantaneous threshold value.
By providing the peak hold circuit, it is possible to realize a voice detection device that can accurately distinguish between voice and noise even from signals switched by the voice switch. Furthermore, it also has a function to set an optimal threshold value in accordance with the nature of the noise, making it possible to detect speech with high sensitivity even in a signal mixed with noise of any nature.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a voice detection device in an embodiment of the present invention, FIG. 2 is a waveform diagram of signals in each part of FIG. 1,
FIG. 3 is a block diagram of a conventional voice detection device, FIG. 4 is a waveform diagram of signals in each part of FIG. 3, and FIG. 5 is a waveform diagram of each part of FIG. 3. 12... Rectifier smoothing circuit, 13... First
dip hold circuit, 14... first peak hold circuit 1, 15... second dip hold circuit, 16... instantaneous threshold calculation circuit,
17...Second peak hold circuit, 18...
...Judgment circuit. Name of agent: Patent attorney Toshio Nakao and 1 other person No. 4
figure time

Claims (1)

[Claims] a rectifying and smoothing circuit that calculates an input signal amplitude value from an audio signal mixed with noise; a first dip-hold circuit that calculates a minimum noise amplitude value from the input signal amplitude value; and an envelope of an amplitude value peak from the input signal amplitude value. A first peak hold circuit calculates the maximum noise amplitude value from the envelope of this amplitude value peak, and a second dip hold circuit calculates the maximum noise amplitude value from the envelope of this amplitude value peak, and an instantaneous threshold value is determined from the maximum noise amplitude value and the minimum noise amplitude value. A second peak hold circuit calculates a maximum threshold value from this instantaneous threshold value, and compares this maximum threshold value with the input signal amplitude value to determine the input signal amplitude value. A voice detection device comprising: a determination circuit that determines that a voice signal is present when a signal amplitude value exceeds the maximum threshold value.