JPH048480Y2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention relates to an audio processing circuit, particularly AD conversion for digital audio processing.

Speech signals usually have different amplitudes depending on the vocalization method and phoneme, and in order to efficiently convert them into AD, an AGC (automatic gain control) circuit is used.
These fluctuations were prevented.

In addition, after passing through a circuit with frequency characteristic functions such as pre-emphasis as pre-processing in the subsequent stage,
AD conversion is also often performed. The present invention is an audio processing circuit for AD converting an audio signal.
A pre-emphasis circuit for making the amplitudes of the first to third formants substantially constant; and an AGC circuit for making the level of the audio signal substantially constant after the first to third formants have been made to have substantially constant amplitudes by the pre-emphasis circuit. and the first output from the AGC circuit.
This audio processing circuit is characterized in that it includes, in the above-mentioned order, an AD conversion circuit that AD converts an audio signal in which the third formant has a substantially constant amplitude and the level has been made substantially constant. By doing this, the present invention can significantly reduce the number of quantizations in AD conversion compared to conventional circuits, and is particularly effective in speech recognition, achieving stable recognition against level fluctuations. be able to.

The present invention will be explained in detail below using specific examples.

Microphones generally used for voice recognition etc.
It is usually a close-talk type microphone such as a headset or a unidirectional microphone, and is used at a distance of 10 to 50 cm from the mouth. Headset allows for a constant distance between the mouth and the microphone, but this is difficult with other methods. or,
There is also a fluctuation of about 10 dB in vocalization.

Therefore, assuming that the fluctuation of the audio signal obtained from the microphone is about 20 dB, and designing the AGC circuit to keep the level close to constant, the dynamic range can be reduced and AD conversion can be performed effectively.

On the other hand, looking at the frequency distribution of long-term audio,
6 per octave from frequency range above 600Hz
It is known that the sound pressure is reduced by ~9dB.
Therefore, in order to clearly extract the information of the first to third formants, it is better to pass the pre-emphasis circuit of 6 to 9 dB per octave so that the first to third formants have approximately constant amplitude. The number of quantizations during AD conversion can be reduced.

FIG. 1 shows a conventional circuit configuration. 1 is the microphone,
2 is an amplifier with an AGC circuit, 3 is a pre-emphasis circuit, and 4 is an AD conversion circuit. Conventionally, the input from microphone 1 was passed through amplifier 2 with AGC circuit and then input to pre-emphasis circuit 3, so the input signal to AD conversion circuit 4 varied in amplitude depending on the phoneme, requiring a large quantization number. .

In the present invention, in order to improve the above-mentioned drawbacks, as shown in FIG. 2, the amplifier 2 with an AGC circuit and the pre-emphasis circuit 3 are reversely replaced. In addition,
Reference numeral 5 denotes an amplifier at the front stage of the pre-emphasis circuit 3. As explained above, the present invention converts audio signals into
An audio processing circuit for AD conversion includes a pre-emphasis circuit for making the first to third formants of an input audio signal approximately constant amplitude; and the pre-emphasis circuit causes the first to third formants to have approximately constant amplitude. an AGC circuit for making the level of the audio signal substantially constant after the signal has been adjusted, and an AGC circuit for making the first to third formants output from the AGC circuit have a substantially constant amplitude; The audio processing circuit is characterized in that it includes an AD conversion circuit that performs AD conversion of a signal in the above-mentioned order. By adopting this configuration, the present invention
Before entering the AGC circuit, a pre-emphasis circuit is used to make the amplitude of the first to third formants almost constant, and then the AGC circuit makes the level of the audio signal almost constant, so the number of quantization during AD conversion is reduced. It has the remarkable effect of being able to. This makes it possible to provide an audio processing circuit that can easily reduce the number of AD conversion bits without sacrificing audio information for recognition using a pre-emphasis circuit with a simple circuit configuration.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a conventional circuit configuration, and FIG. 2 is a block diagram showing a circuit configuration in an embodiment of the present invention. 1...Microphone, 2...Amplifier with AGC circuit, 3
...Pre-emphasis circuit, 4...AD conversion circuit, 5...Amplifier.

Claims (1)

[Claims for Utility Model Registration] An audio processing circuit for AD converting an audio signal, comprising: a pre-emphasis circuit for making the first to third formants of an input audio signal approximately constant in amplitude; 1st to 3rd
an AGC circuit for making the level of the audio signal substantially constant after the formants are made to have a substantially constant amplitude, and the first to third formants output from the AGC circuit are made to have a substantially constant amplitude, and An audio processing circuit comprising: an AD conversion circuit that AD converts a constant audio signal; and an AD conversion circuit that performs AD conversion on a constant audio signal, in the above order.