JPS5873217A - Signal input device - Google Patents

Abstract

PURPOSE:To correctly input sound without being affected with surrounding noise, by automatically adjusting the gain of a sound signal including the surrounding noise and keeping the result to an optimum value, through the signal level at a frequency region other than frequency components of the sound signal. CONSTITUTION:A sound signal detected at a sound detection microphone 1 is amplified at a preamplifier 2, and the sound frequency region only is inputted to an amplifying circuit 5 at a band-pass filter. A noise signal detected at a microhone 1' for noise detection is amplified at a preamplifier 2' and components other than the sound frequency region are inputted to a smoothing circuit 4 at a band block filter 3'. The amplification factor of the circuit 5 is adjusted in response to the output level of the circuit 4. An output L5 of the circuit 5 is compared with a threshold level LT at a signal processing circuit 6, and when L5>LT, the component is decided as that in the sound region. The output of the circuit 6 is inputted to a filter bank 7 to form an output pattern at each frequency, and decision is done through the pattern matching with the sound pattern registrated at a deciding circuit 8.

Description

[Detailed Description of the Invention] The present invention relates to a signal input device, which controls an audio signal amplification circuit according to ambient conditions such as noise, and extracts and recognizes the speaker's voice. .

When attempting to detect an audio signal, if the noise level around the speaker is high, noise may be input along with the audio through the microphone for audio input, making it difficult to extract the audio area. Even if a loud voice is uttered to increase the level ratio between voice and noise, if the surrounding noise level is extremely high, the voice region may not be extracted.

In addition, in the method of placing a microphone near the speaker's mouth and lowering the amplification factor of the amplifier to extract the voice region by speaking loudly, it is necessary to consider how to hold the microphone.
This has the disadvantage that the speaker is forced to bear the burden of holding the microphone, resulting in inconvenience in use.

The present invention eliminates these drawbacks by providing a detection circuit for detecting the ambient noise level with a circuit that weights frequency components without affecting audio area detection. ), the amplification factor of the speaker's voice signal including ambient noise is automatically controlled, thereby making it possible to discriminate the voice region based on a constant threshold level, regardless of the noise level.

One embodiment will be described below with reference to the drawings.

In Fig. 1, 1 is a microphone for voice detection, 1'
is a microphone for noise detection; 2 is a preamplifier that amplifies the audio signal (including ambient noise signals) detected by microphone 1; 2' is a preamplifier that amplifies the noise signal detected by microphone 1'; is a circuit that weights frequency components that affect audio region detection, and is composed of, for example, a band-pass filter. Reference numeral 3' denotes a circuit for extracting a signal mainly consisting of a noise portion, which weights frequency components other than the frequency region that affects voice region detection, and uses, for example, a band-cut filter. 4 is circuit 3'
This is a smoothing circuit that smoothes the output and levels it.

Reference numeral 5 denotes an amplifier circuit whose amplification factor changes according to the output of the smoothing circuit 4, and the output signal of the bandpass filter 3 is amplified with an amplification factor corresponding to the output level of the smoothing circuit 4. That is, when the output level L4 of the smoothing circuit 4 is high, the amplification circuit 5 suppresses the amplification factor of the voice band signal because the ambient noise level is high, and when L4 is small, the amplification factor is increased. It is a circuit. Reference numeral 6 denotes a signal processing circuit that processes the signal L5 amplified by the amplifier circuit 5, and compares it with a certain threshold level LT and determines that the signal is in the audio region if L5>LT. This threshold 9 level LT may be constant. Ji [If the sound level is low, the audio area (section 201 in Figure 2,
It is natural that the section 301 in the figure) can be determined, but even when the ambient noise level is high, the output of the amplifier circuit 50 is controlled by the noise level of the smoothing circuit 4, and the output of the amplifier circuit 50 is controlled by the noise level of the smoothing circuit 4. The level will be lowered as shown in Figure 3.

FIG. 2 shows an output waveform of an amplifier circuit in a conventional audio input device. In conventional devices, the amplification factor of the amplifier circuit was constant, so when determining the audio region, if the threshold level L was fixed, it was difficult to determine the audio region (2010 section) when the ambient noise l/bell was low. was correct, but when the noise level was high, the audio region was in section 202, but it was mistaken for section 203.

On the other hand, in the present invention, since the increase rate of the amplifier circuit is changed according to the noise level, the output waveform of the amplifier circuit becomes as shown in FIG.
As shown in item 2, the voice area can be determined fairly accurately.

Regardless of the magnitude of the ambient noise level, the audio signal determined to be in the audio area by the signal processing circuit 6 in FIG.
For example, the output R for each frequency is output through the filter bank 7 of an analog frequency spectrum analysis using a group of bandpass filters.
A turn is created and determined by performing, for example, turn matching with the voice A turn previously registered in the determination circuit 8.

9 is the determination result.

As described above, according to the present invention, the r-in of an audio signal including ambient noise is automatically adjusted based on the signal level of a frequency domain other than the frequency components of the audio signal, and is kept at an optimal value. It is possible to input voice correctly without changing the distorted threshold.

Furthermore, it has the advantage that the microphone for voice input does not have to be brought close to the mouth, so there is no inconvenience in using it. In order to improve TJ performance, the microphones 1 and 1' and the preamplifier circuits 2 and 2' do not need to be separated, and the bandpass and lA:I filters 3 can be omitted. It is possible. These processes can be performed in both analog and digital ways.

The present invention is particularly effective for voice recognition in noisy places.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a signal input device according to an embodiment of the present invention, FIG. 2 is an output waveform diagram of the amplifier of the conventional signal input device, and FIG. 3 is an output waveform diagram of the amplifier of the device of the present invention. 1.1'...microphone, 2.2'...preamplifier,
3.Band pass filter, 3'...Band cut filter,
4. Smoothing circuit, 5... Amplifier, 6. Signal processing circuit, 7
...Filter puncture, 8. Judgment circuit, 9. Judgment result.

Claims (1)

[Claims] an amplifier with a variable amplification factor that amplifies audio signals including noise;
It has a control circuit that detects the level of noise outside the audio range and changes the amplification factor of the amplifier according to this noise level, and a circuit that extracts the output of the amplifier above a certain level, and the output of this circuit. A signal input device characterized in that it uses as an audio input signal.