JPH02164129A - Voice signal processing circuit - Google Patents

Abstract

PURPOSE:To correct an offset voltage of an analog element such as an operational amplifier automatically by providing an adder, a comparator, a counter and a D/A converter so as to give an adjust signal externally. CONSTITUTION:A latch circuit 12 allows a selection circuit 11 to select a reference voltage signal till an enable signal being an output of a comparator 9 is finished from the input of an adjust signal. The adjust signal is used to input the reference voltage to a voice signal processing section 3 and to clear a counter 9. In this case, an offset voltage of a voice signal processing section appears at the output of the voice signal processing section 3. At that time, the comparator 9 enables the counter 6. During the enable period the counter 6 counts up synchronously with the clock signal, the output of a D/A converter 7 is gradually increased and when the output of an adder 4 exceeds a reference voltage. The output of the comparator 9 changes, the counter 6 is disabled and a signal from a microphone is inputted to the voice signal processing section 3. Moreover, the count of the counter 6 at this time is used to correct the offset of the voice signal processing section 3.

Description

[Detailed description of the invention] 1 dish partial pressure The present invention relates to an audio signal processing circuit, and more particularly to an audio signal processing circuit.

The present invention relates to a technology for canceling the offset voltage of an operational amplifier in an analog signal processing circuit, and is applied, for example, to voice processing such as voice recognition and voice synthesis.

In a voice processing circuit made up of analog circuits, the offset voltage of analog elements such as operational amplifiers affects the conditions for voice section detection and voice feature patterns. cause undesirable phenomena.

If this offset voltage is constant among individual products, correction is easy, but in reality there is variation. This offset voltage variation is caused by variations between individual elements, differences in power supply voltage values, and the like. In particular, voice recognition boards that do not have their own power supply are affected by the difference in the WX voltage value, and furthermore, this offset voltage is
Even in one circuit, it changes due to fluctuations in power supply voltage and changes in ambient temperature.

In order to solve this problem, conventionally, individual elements such as operational amplifiers are used with as small an offset voltage as possible. Or adjust individual offset voltages.

・Install a constant voltage circuit to prevent differences in the voltage of the supplied power.

Some ideas have been considered and implemented.

・It is technically difficult to eliminate the offset voltage of individual elements. Also, the method of individually:JJu increases production costs.

- Addition of a constant voltage circuit increases the circuit M model and production cost.

It had the following drawbacks and was not an appropriate solution.

The present invention was made in view of the above-mentioned circumstances.
In the audio processing circuit that is made up of analog circuits, the conditions for vocal section rJJ detection and the offset voltage of analog elements such as operational amplifiers that affect the audio feature pattern are determined using adders, comparators, counters, and D/A. By simply adding a simple converter (which may be implanted with an op amp and a resistor), by instructing the user to unplug the microphone jack, or by providing an external Asyas 548 This was done for the purpose of making it possible to correct it automatically.

In order to achieve the above object, the present invention includes a switch that switches between an audio signal and a reference voltage signal, and an analog circuit that processes the audio signal or the reference voltage signal that is the output of the switch. an audio signal processing section that adds the output of the D/A converter and the output of the audio signal processing section, or an adder that calculates 2I8, and compares the output of the adder with a reference voltage signal.

A comparator that sends an enable signal to a counter based on the comparison result, a counter that is initialized with an adjustment signal that indicates an adjustment instruction and counts up or down only during the period when the enable signal is input, and a counter that outputs an output from the counter. A D/A converter that performs D/A conversion and sends it to the adder, and has a function of correcting the offset voltage of the audio signal processing section, or a selection of switching between the audio signal and the jA reference voltage signal. a circuit, an audio signal processing section that is configured of an analog circuit and processes the audio signal or reference voltage signal that is the output of the switch, and adds the output of the D/A converter and the output of the audio signal processing section, or , an adder for subtraction, the output of the adder and a reference voltage signal are compared, and the comparator sends an enable (ff signal) to the counter based on the comparison result, and the enable is initialized with an adjustment signal indicating an adjustment instruction. A counter that counts up or counts down by 1 only while a signal is being input;
and a D/A converter which D/A converts the output of the counter and sends it to the adder.
causing the selection circuit to select the reference voltage signal during a period from the start of the H signal until the end of the enable signal;
The present invention is characterized in that it has a function of automatically correcting the offset voltage of the audio signal processing section. An explanation will be given below based on one embodiment of the present invention.

FIG. 1 is a configuration diagram for explaining an embodiment of the invention described in claim 1 of δa, in which l is a microphone jack socket, 2 is a reference voltage, 3 is an audio signal processing section, and 4 is a Adder, 5 is A/D converter.

6 is a counter, 7 is a D/A converter, 8 is a reference voltage,
9 is a comparator, and microphone jack 1 is.

Switch between the audio signal from the microphone and the reference voltage signal (ground) 2.

The audio signal processing section 3 is constructed with an analog circuit, and processes (No. 3) from the microphone jack socket 1.

The adder 4 adds the output of the D/A converter 7 and the output of the audio signal processing section 3.

Comparator 9 compares the output of adder 4 with reference voltage signal 8, and sends an enable signal to counter 6 based on the comparison result.

The counter 6 is initialized with an adjustment signal indicating an adjustment instruction, and counts up in synchronization with the clock signal only while the enable signal is input.

The D/A converter 7 converts the output of the counter 6 into a D/A converter.
It is converted and sent to the adder 4 mentioned above.

The A/D converter 5 A/D converts the output of the adder and sends it to a subsequent digital circuit (not shown).

Next, the operation of the circuit shown in FIG. 1 to correct the offset voltage will be explained.

When the adjustment signal arrives, it is assumed that the microphone jack is not inserted into the microphone jack socket (the reference voltage signal is input to the audio processing section), or the user is instructed to remove the microphone jack from the microphone jack socket. give instructions. At this time, the offset voltage of the audio signal processing section appears in the output of the audio signal processing section 3. The counter 6 is reset by the adjustment signal. In the D/A converter 7, when the counter 6 is in a cleared state, the value obtained by adding the offset voltage 1 to the voltage of the audio G signal processing unit 3 and the output of the D/A converter 7 by the adder 4 is always negative. It is preset to be . Further, the voltage corresponding to the LSB of the D/A converter 7 is the LSB of the A/D converter 5.
The voltage is set below the corresponding voltage.

When the output of adder 4 is negative, the output of comparator 9 becomes H.
i, which enables counter 6.

During the enable period, the counter 6 counts up in synchronization with the clock signal, and the output of the D/A converter 7 gradually increases toward the positive side.

Counter 6 advances, and at a certain point the output of adder 4! When the l voltage value is exceeded, the output of comparator 9 changes and counter 6 is disabled.

The value of the counter 6 at this time is a value for correcting the offset of the audio processing section 3, and is held until the first adjustment signal arrives.

If the user is first instructed to unplug the microphone jack, the user is instructed to connect the microphone jack after sufficient time has elapsed for this series of procedures to be completed.

As is clear from the above description, according to the embodiment shown in FIG. The offset voltage of the element is calculated using an adder, a comparator, a counter, and a D/
A converter (may be composed of an operational amplifier and a resistor)
By simply adding this simple circuit, it is possible to automatically correct this by instructing the user to unplug the microphone jack.

FIG. 2 is a block diagram for explaining in detail the invention set forth in claim 2. In FIG. 1, 11 is a selection circuit, 12 is a latch circuit, and the selection circuit 11 receives the audio signal from the microphone. and the reference voltage signal (ground). In addition, the latch @ path 12 operates from the time when the adjuster I signal is input until the enable signal, which is the output of the comparator, ends.
The selection circuit selects the reference voltage signal. In addition, the same reference numerals as in the case of FIG. 1 are given to the other parts having the same functions as in the embodiment shown in FIG. 1, and individual explanations are omitted.

Next, the operation of the circuit shown in FIG. 2 to correct the offset voltage will be explained.

First, a reference voltage is input to the audio signal processing unit 3 via the selection circuit 11 as an adjustment signal, and the counter 6 is also cleared. At this time, the offset voltage of the audio signal processing section appears in the output of the audio signal processing section 3. The D/A converter 7 is configured such that when the counter 6 is cleared, the value obtained by adding the offset voltage of the audio signal processing section 3 and the output of the D/A converter 7 by the adder 4 is always negative. It is set in advance. Further, the voltage corresponding to the LSB of the D/A converter 7 is set to be lower than the voltage corresponding to the LSB of the A/D converter 7.

When the output of adder 4 is negative, the output of comparator 9 becomes H.
i, which enables counter 6. During the enable period, the counter 6 counts up in synchronization with the clock signal, and the output of the D/A converter 7 gradually increases toward the positive side. When the counter 6 advances and at some point the output of the adder 4 exceeds the reference voltage value, the output of the comparator 9 changes and the counter 6 is disabled.

In addition, (6) from the microphone is input to the audio A processing unit 3. The value of the counter 6 at this time is a value for correcting the offset of the audio borrowing/processing unit 3, and the next adjustment signal will be held until

As is clear from the above explanation, according to the embodiment shown in FIG. Add the offset voltage of the element by $7. equipment, comparators, counters, and
By simply adding a simple circuit called a D/A converter (which may be composed of an operational amplifier and a resistor), automatic correction can be performed by applying an external adjustment signal.

Figure 3 shows the audio signal processing [1] shown in Figures 1 and 2.
In the lli circuit 10, the adjuster l-(i) includes a system reset signal. In this way, in an audio processing circuit composed of an analog circuit, the conditions for detecting a voice section and the operational amplifier, etc. that affect the voice feature pattern can be controlled. The offset voltage of the analog elements of It can be corrected.

FIG. 4 shows a feature extracting section of a speech recognition device configured with the above-mentioned speech signal processing circuit. In the figure, 21 is a microphone, 22 is a feature extraction section, 23 is a recognition section, and 24 is a speech standard pattern memory. The feature extraction unit 22 is configured with the above-mentioned audio signal processing circuit, and extracts the feature amount of the audio input from the microphone 21. The audio standard pattern memory 24 is
Contains standard patterns of speech to be recognized.

The recognition unit 23 detects a voice section from the voice feature extracted by the feature extraction unit 27, and creates a pattern using the voice feature between the voice features and the voice standard pattern stored in the voice standard pattern memory 24. Performs matching and uses the most similar standard pattern as the recognition result.

Therefore, according to the embodiment shown in FIG. 4, in a speech recognition device, especially in a speech recognition board that does not have its own power supply section, the conditions for speech section detection and the speech feature pattern are affected. The offset voltage of an analog circuit caused by variations between individual elements and differences in power supply voltage values, which cause an undesirable phenomenon of a reduction in efficiency, can be automatically corrected at a low cost.

FIG. 5 shows a speech recognition device that includes an audio signal processing circuit and automatically applies an adjustment signal to the audio signal processing circuit during a pause section of input audio, and FIG. 6 shows its operation. The feature extraction unit 22 is configured with the above-mentioned audio signal processing circuit, and extracts the feature amount of the audio input from the microphone 21. The speech standard pattern memory 24 stores standard speech patterns to be recognized. The recognition unit 23 extracts a voice section (sixth
Figures (b) and T) are detected, and pattern matching is performed using the voice features between them and the voice standard pattern stored in the voice standard pattern memory 24, and the most similar standard pattern is recognized as the result. shall be. Further, when the voice is input and the recognition section detects the end of the voice, it sends an adjustment signal (S in FIG. 6(c)) to the special Vl amount extraction section.

Therefore, according to this embodiment, in the speech recognition device,
In particular, in voice recognition boards that do not have their own power supply, the conditions for detecting voice zone hill and the differences between individual elements that affect the voice feature pattern and cause an undesirable phenomenon such as a decrease in recognition rate. Automatically corrects analog circuit offset voltage caused by variations, differences in power supply voltage values, power supply fluctuations, and changes in ambient temperature at a low cost, and always performs recognition in the best condition for each audio input. I can do it.

EFFECT-11 BUTTON 1 As is clear from the above description, according to the invention of claim 1, in a speech processing circuit constituted by an analog circuit, the conditions for speech section detection and the speech feature pattern are affected. By simply adding a simple circuit such as an adder, a comparator, a counter, and a D/A converter (which may also be composed of an operational amplifier and a resistor), the offset voltage of an analog device such as an operational amplifier can be adjusted by adding a microphone jack to the user. It is now possible to automatically correct this by giving an instruction to remove it.

Further, according to the invention of claim 2, in the audio processing path a composed of analog circuits, an adder calculates the offset voltage of an analog element such as an operational amplifier that affects the conditions for detecting a voice section and the voice feature pattern. , comparators, counters, and.

Just add a simple circuit called a D/A converter (which may be composed of an operational amplifier and a resistor).

Automatic correction is now possible by applying an external adjustment signal.

[Brief explanation of the drawing]

1 to 5 are sweep diagrams for explaining the present invention in detail, respectively, and FIG. 6 is a timing chart of the embodiment shown in FIG. 5. 1... Microphone jack socket, 2... Reference voltage,
3...Audio signal processing section, 4...Adder, 5...A
/D converter, 6...counter, 7...D/A converter. 8...M quasi voltage, 9...Comparator, 10...
Audio (W processing circuit, 11...selection rjjJ @,
l 2...Latch circuit. 13...Adjust switch, 14...Order, 2
1...Microphone, 22...Feature extraction unit, 23
. . . recognition unit, 24 . . . speech standard pattern memory. Figure 2 Patent agent Ricoh Co., Ltd.

Claims (1)

[Claims] 1. A switch for switching between an audio signal and a reference voltage signal, an audio signal processing section that is configured of an analog circuit and processes the audio signal or the reference voltage signal that is the output of the switch, and a D/ an adder that adds or subtracts the output of the A converter and the output of the audio signal processing section; a comparator that compares the output of the adder with a reference voltage signal and sends an enable signal to the counter based on the comparison result; , a counter that is initialized with an adjustment signal indicating an adjustment instruction and counts up or down only during the period when the enable signal is input; and D that converts the output of the counter from D/A and sends it to the adder. 1. An audio signal processing circuit comprising: a /A converter, and having a function of correcting an offset voltage of the audio signal processing section. 2. A selection circuit that switches between the audio signal and the reference voltage signal, an audio signal processing section that is composed of an analog circuit and processes the audio signal or the reference voltage signal that is the output of the switch, and the output of the D/A converter. an adder that adds or subtracts the output of the audio signal processing section; a comparator that compares the output of the adder with a reference voltage signal; and sends an enable signal to a counter based on the comparison result; a counter that is initialized with an adjustment signal shown and counts up or down only during the period when the enable signal is input; and a D/A converter that converts the output of the counter into a D/A and sends it to the adder. a function of causing the selection circuit to select the reference voltage signal and automatically correcting the offset voltage of the audio signal processing section during a period from the start of the adjustment signal until the end of the enable signal; An audio signal processing circuit comprising: