JPH08321786A - Sound discrimination circuit - Google Patents

Abstract

PURPOSE: To attain transmission processing by a transmission system and a reception system with a simple configuration when the transmission characteristic is improved through the application of the principle of distributed antenna. CONSTITUTION: The circuit is provided with an A/D converter circuit 11 converting a received sound signal into a digital signal at internals of a prescribed time, an absolute value processing circuit 12 converting an output signal of the A/D converter into an absolute value, and a smoothing means 10 eliminating an AC component of the output signal from the absolute value processing circuit 12. Furthermore, the smoothing means 10 has a digital filter means having an adder means in a signal processing path from an input to an output and a delay means in a signal feedback path from the input to the other input of the adder means and has a lower value limit circuit 13 inserted to the signal processing path or the signal feedback path to limit a lower limit value of the processed signal and also has a comparator circuit 15 comparing respective output signals of a reference value generating circuit 16 and the smoothing means 10 and using the result as a discrimination output.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice determination circuit, and more particularly to a voice determination circuit based on digital signal processing.

[0002]

2. Description of the Related Art Conventionally, this type of sound determination circuit has been used for automatic transmission / reception switching of a wireless communication device, automatic recording control of a recording device, and the like. For example, FIG. 7 is a block diagram showing the configuration of a conventional sound determination circuit.

Referring to FIG. 7, this conventional sound determining circuit removes an AC component contained in an output signal of the rectifier circuit 71 for rectifying a voice signal containing an AC signal in an audible frequency band. A smoothing circuit 72, a comparison circuit 73 that compares the output of the smoothing circuit 72 with a preset reference value, and outputs the comparison result as a determination output signal, and a reference value generation circuit 74 that generates a preset reference value. It consists of

Next, referring to FIG. 7 and FIGS. 8A to 8D, which are waveform charts showing an example of the operation thereof, the operation of the conventional sound determining circuit will be described. In this conventional sound determination circuit, when an audio signal as shown in FIG. 8A is input to the rectifier circuit 61, the output of the rectifier circuit 71 is shown in FIG.
A pulsating flow signal as shown in FIG. 8B is generated, and a smoothed signal as shown in FIG. 8C appears at the output of the smoothing circuit 72. The output of the smoothing circuit 72 is compared with a preset reference value in the comparison circuit 73. The output of the comparison circuit 73 is
As the determination output signal shown in FIG. 8D, a logical value "1" is output when the output of the smoothing circuit 72 exceeds the reference value, and conversely, when the output of the smoothing circuit 72 is below the reference value. Outputs a logical value "0".

As described above, the conventional voice determination circuit shown in FIG. 7 outputs a logical value "1" when a voice signal having an amplitude of a certain level or more continues to be input for a certain period of time, and outputs a voice signal. It acts to notify that there is something.

[0006]

The main fields of application of the speech determination circuit are automatic switching of transmission / reception of a wireless communication device and automatic recording of a recording device. Also, in fields such as mobile phones, where it is required to make long-term calls with a small battery, it is necessary to stop transmission during silent periods in order to minimize power consumption. , A voice determination circuit is required. In either case, the target voice signal for voice determination is also the target of transmission or recording. Therefore, if the determination delay of the voice determination circuit is large, the delay from the start of voice input to the start of transmission or recording becomes large. As a result, the beginning of the voice signal transmitted or recorded is missing, and there is a problem that the conversation is hindered especially when the beginning is a consonant.

Therefore, there is little delay from the start of voice input to the determination of the voice in the voice determination circuit.
In addition, it is required to prevent erroneous determination due to unnecessary noise other than voice.

However, in the above-described conventional voice decision circuit, the smoothing circuit is inherently accompanied by a delay, which causes a delay from the input of voice until the decision output becomes the logical value "1". Has become.

In order to shorten the judgment delay, it is necessary to reduce the time constant of the smoothing circuit or set the judgment level supplied to the comparison circuit low. On the other hand, the judgment level needs to be set high in order to relax the allowable level for noise, and the time constant of the smoothing circuit needs to be increased in order to prevent erroneous judgment due to pulsed noise. In other words, the reduction of the determination time and the relaxation of the allowable level for noise are incompatible problems.

Further, if a delay circuit is inserted in the voice transmission system instead of allowing the voice decision delay, and the voice transmission is relatively delayed compared to the voice decision, there is a problem that the consonant at the beginning of the word is missing. Will be resolved. However, since this method increases the delay of the audio signal to be transmitted, it is effective in the case of recording and unidirectional communication, but bidirectional communication,
Especially in the case of a telephone, there is a drawback that it causes unnatural reverberation and deteriorates the call quality. Therefore, the method of inserting the delay circuit in the voice transmission system is not suitable for the application to the telephone, and the fundamental problem cannot be solved unless the delay in determining the voice is itself shortened.

As described above, in the conventional voice determination circuit, the voice determination time is short without delay of the signal itself,
Moreover, it is very difficult to realize a voiced decision circuit having a sufficient allowance for noise.

Therefore, an object of the present invention is to maintain a sufficient permissible amount of noise in the voiced decision circuit and shorten the voiced decision time.

[0013]

Therefore, the voiced sound judging circuit according to the present invention comprises a rectifying means for rectifying an input audio signal and a smoothing means for removing and smoothing an AC component of an output signal of the rectifying means. In a voiced decision circuit having a comparison means for comparing the output signal of the smoothing means with a preset reference value and using the result as a decision output signal, the voice signal is converted into a digital signal at regular time intervals. AD conversion means for converting is provided, the rectification means is made up of absolute value conversion means for converting the output signal of the AD conversion means into an absolute value, and the smoothing means is added to a signal processing path from its input terminal to its output terminal. Means and digital filter means having delay means in a signal feedback path from the output terminal to the other input of the adding means, and is inserted in the signal processing path or the signal feedback path. And a lower limit value limiting means for limiting the lower limit value of the processed signal.

[0014]

Next, the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a first embodiment of a sound determining circuit of the present invention.

Referring to FIG. 1, the sound determination circuit of the present embodiment includes a smoothing means 10 including an AD conversion circuit 11, an absolute value conversion circuit 12, a lower limit value limiting circuit 13 and a low pass filter 14, a comparison circuit 15, and a reference. And a value generating circuit 16.

The AD conversion circuit 11 converts the voice signal input to the voice determination circuit into a digital signal at regular time intervals and outputs the digital signal.

The absolute value conversion circuit 12 converts the output signal of the AD conversion circuit 11 into an absolute value and outputs it to the smoothing means 10.

The smoothing means 10 is a digital filter means for removing the AC component of the output signal of the absolute value circuit 12 and smoothing it. In this embodiment, the smoothing means 10 is constituted by a lower limit value limiting circuit 13 and a low-pass filter 14 connected in cascade. ing.

The comparison circuit 15 compares the output signal of the smoothing means 10 with a preset reference value and uses the result as a judgment output signal.

The reference value generation circuit 16 generates a reference value set at the time of design and outputs it to the comparison circuit 15.

Next, with reference to FIG. 1 and FIG. 2 which is a waveform diagram showing an example of the operation thereof, the operation of the voice determination circuit of the present embodiment will be briefly described.

FIG. 2 (a) is a waveform diagram showing an example of a voice signal input to the voiced decision circuit of the present invention.
(F) is a waveform diagram in which the value of the digital signal inside the sound determination circuit of the present invention is expressed on the same time axis as that of FIG. 2 (a).

When a voice signal as shown in FIG. 2A is input to the voice input terminal 17, the output of the AD conversion circuit 11 has a constant digital code proportional to the input signal as shown in FIG. 2B. Appears at every sampling interval. The output signal of the AD conversion circuit 11 is output by the absolute value conversion circuit 12 as shown in FIG.
As shown in (c), the lower limit value conversion circuit 13 converts the absolute value.
It is input to the low-pass filter 14 via.

FIG. 3 is a circuit diagram showing an example of a detailed configuration of the smoothing means 10 in the sound determination circuit of FIG.

Since the smoothing means 10 in the present embodiment is configured to limit the lower limit value of the input signal of the filter means, a lower limit value limiting circuit 13 including a lower limit value generator 31, a comparator 32 and a selector 33 is provided. , A first multiplier 34, a second multiplier 35, an adder 36, and a delay device 37.
(Infinite Impulse Responses
e) It can be separated into a filter part.

In the lower limit value limiting circuit 13, when the comparator 32 determines that the output of the absolute value converting circuit 12 is smaller than the output of the lower limit value generator 31, the selector 3
3 selects the output of the lower limit value generator 31, and otherwise selects the output of the absolute value conversion circuit 12. Thus, from the selector 33, as the output signal of the lower limit value limiting circuit 13,
As shown in FIG. 2D, a signal whose lower limit is limited is output.

The output signal of the lower limit value limiting circuit 13 is then smoothed by an IIR filter section consisting of a first multiplier 34, a second multiplier 35, an adder 36, and a delay unit 37, as shown in FIG. The signal becomes as shown in (e) and is output as the output signal of the smoothing means 10.

Finally, the output signal of this smoothing means 10 is
The output of the reference value generation circuit 16 is compared by the comparison circuit 15. In the comparison circuit 15, the output signal of the smoothing means 10 is used as the determination output signal shown in FIG.
When it is larger than the output of 6, a logical value "1" is output, and in other cases, a logical value "0" is output.

As described above, in the sound presence judgment circuit of this embodiment, when the strength of the voice signal input to the AD conversion circuit 11 exceeds the predetermined strength, the comparison circuit 15 outputs the logical value "1".
Is output, it is determined to be voiced. The point to be particularly noted here is that the output value of the smoothing means 10 is always larger than the output value of the lower limit value generator 31 due to the effect of the lower limit value limiting circuit 13, so that the comparison is performed after the voice starts to be input to the AD conversion circuit 11. That is, the delay until the circuit 15 outputs the logical value "1" is small.

Next, the effect of shortening the delay by providing the lower limit to the input signal of a general low-pass filter will be described using mathematical expressions.

The following expression (1) is an example of an expression representing the output value of the low-pass filter when a constant value is input with time t = 0 as a boundary. V _O is the input value at t ≧ 0, τ is the time constant, and V (t) is the output value at time t.

[0033]

In this case, the voice determination level is V _D = r _D
If V _O (provided that 0 ≦ r _D <1), the determination time t _d is the formula (2) it is derived from equation (1).

[0035]

On the other hand, the lower limit value is V _L = r _L · V _D
= R _L · r _D · V _O (where 0 ≦ r _L <1), the output value of the low-pass filter according to the present invention is expressed under the same condition as the expression (3).

[0037]

In this case, the voice determination level is set to V similarly.
_{When D} = r _D · V _O (where 0 ≦ r _D <1), the determination time t _d is derived from the formula (3) by the formula (4).

[0039]

FIG. 6 is a characteristic diagram showing the judgment time characteristics of the judgment output of the sound judging circuit of the present invention. FIG. 2 is a graph showing Equations (2) and (4), where the horizontal axis represents r _D as a voiced determination level and the vertical axis represents a voiced determination time t _d.
Is represented.

It can be seen from FIG. 6 that the present invention reduces the decision delay more than the conventional case. For example, when the lower limit level is set to 90% of the voice determination level (r _L = 0.9), the voice determination time of the present invention is shortened to one third or less of the voice determination time of the conventional technique. I understand that it will be done. Further, it is clear that if the difference between the lower limit level and the voiced judgment level is further shortened, that is, if r _L is made larger, the voiced judgment time is further shortened.

Incidentally, the portion realized by the digital system in the present invention can also realize the same function by the analog system.

However, if it is realized by an analog system, it is necessary to consider the manufacturing precision and uniformity of each element which is a constituent element. For example, when the relative error between the lower limit voltage value used in the lower limit value limiting circuit and the voltage value of the reference value generating circuit is 10%, the lower limit voltage must be set lower than the voltage of the reference value generating circuit by 10% or more. I won't. This is because there is a possibility that the sound is erroneously determined when there is no sound. However, in the present invention, since the lower limit value limiting circuit, the comparing circuit and the reference value generating circuit are constituted by digital circuits, they are not affected by manufacturing errors or non-uniformity.

Therefore, in the case of the voice determination circuit of this embodiment, even if the difference between the output of the reference value generation circuit and the lower limit value of the lower limit value limiting circuit is set to the minimum in the range that can be expressed as a digital value, No false positives will occur. Therefore, it is possible to extremely reduce the delay from the start of the voice input to the AD conversion circuit 11 until the comparison circuit 15 outputs the logical value "1".

It should be noted that FIG. 3 merely shows one embodiment of the smoothing means 10 of the sound determination circuit according to the present invention, and the present invention can be easily realized by other configurations.

For example, FIG. 4 is a circuit diagram showing a second embodiment of the smoothing means in the sound determination circuit of the present invention.

Referring to FIG. 4, the lower limit value limiting circuit 13 is constituted by the lower limit value generator 41, the comparator 42, and the selector 43, as in FIG. However, the first multiplier 4
4 is different from the circuit of FIG. 3 in that the lower limit value limiting circuit 13 is placed on the output side of the adder 46 of the IIR filter unit constituted by the second multiplier 45, the adder 46, and the delay device 47. ing.

The resulting characteristic of the circuit of FIG. 4 is that the lower limit value as the output of the smoothing means is equal to the output value of the lower limit value generator 41, and the lower limit value of the output changes even if the coefficient of the IIR filter is changed. There is nothing. Therefore,
When the circuit of FIG. 4 is used, the amount of circuit modification accompanying a filter design change can be reduced.

FIG. 5 is a circuit diagram showing a third embodiment of the smoothing means in the sound determination circuit of the present invention.

Referring to FIG. 5, the lower limit value limiting circuit 13 is composed of the lower limit value generator 51, the comparator 52, and the selector 53, which is similar to FIGS. 3 and 4, but the first multiplier 54 is used. In the IIR filter section composed of the second multiplier 55, the adder 56, and the delay device 57, the lower limit value limiting circuit 13 is placed on the signal feedback path from the output terminal in FIGS. Different from the circuit.

The characteristic of the circuit of FIG. 5 generated by this is that there is no lower limit value limiting means on the signal processing path from the input terminal as the smoothing means to the output terminal, so that a circuit with less delay due to digital processing can be realized. .

The absolute value conversion circuit 1 described in this embodiment is used.
2, the smoothing means 10 including the lower limit value limiting circuit 13, the comparison circuit 15, and the reference value generating circuit 16 are functional means that can be realized by software sequential control of a system using a microcomputer, in addition to means by hardware.

[0053]

As described above, the voiced sound judging circuit according to the present invention comprises the lower limit value limiting means for limiting the lower limit value of the processed signal in the signal processing path or the signal feedback path of the smoothing means composed of the digital filter means. Since it is provided, there is an effect that the delay time from the start of voice input until it is determined that the voice is present can be significantly shortened.

Further, since the lower limit value limiting means is provided, there is no influence of signals below the set lower limit level. That is, since any signal below the lower limit level is raised to the lower limit level, the processing after the smoothing means is not affected by noise. Therefore, erroneous determination due to noise is reduced even though the voice determination time is shortened.

Further, the voiced decision circuit according to the present invention is
Since the D conversion circuit and the subsequent steps are realized by digital processing, it is possible to easily set the lower limit level and the sound determination level to extremely close values. In other words, there is an effect that there is no need to design with a margin in the difference between the lower limit level and the voiced judgment level in order to prevent malfunctions due to manufacturing errors of elements constituting the circuit, usage conditions and environmental changes.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of a sound determination circuit of the present invention.

FIG. 2 is a waveform diagram showing an example of the operation of the sound determination circuit of FIG.

FIG. 3 is a circuit diagram showing a detailed configuration example of a smoothing unit in the sound determination circuit of FIG.

FIG. 4 is the second smoothing means in the voice determination circuit of the present invention.
3 is a circuit diagram showing an embodiment of FIG.

FIG. 5 is a third example of the smoothing means in the sound determination circuit of the present invention.
3 is a circuit diagram showing an embodiment of FIG.

FIG. 6 is a characteristic diagram showing the determination time characteristic of the determination output of the sound determination circuit of the present invention.

FIG. 7 is a block diagram showing an example of a conventional sound determination circuit.

8 is a waveform chart showing an example of an operation in the voice determination circuit in FIG.

[Explanation of symbols]

 10 smoothing means 11 AD conversion circuit 12 absolute value conversion circuit 13 lower limit value limiting circuit 14 low-pass filter 15 comparison circuit 16 reference value generation circuit 31, 41, 51 lower limit value generator 32, 42, 52 comparator 33, 43, 53 selection Unit 34,44,54 First multiplier 35,45,55 Second multiplier 36,46,56 Adder 37,47,57 Delay device 71 Rectifier circuit 72 Smoothing circuit 73 Comparison circuit 74 Reference value circuit

Claims (2)

[Claims] 1. A rectifying means for rectifying an input audio signal, a smoothing means for removing and smoothing an AC component of an output signal of the rectifying means, and an output signal of the smoothing means with respect to a preset reference value. And a comparison means for comparing the result as a determination output signal with a voice output determination circuit, which comprises AD conversion means for converting the voice signal into a digital signal at regular time intervals, and the rectification means performs the AD conversion. The output signal of the means is converted into an absolute value, and the smoothing means has an adding means in a signal processing path from its input terminal to an output terminal, and the smoothing means has the output terminal to the other input of the adding means. A digital filter means having a delay means in a signal feedback path to reach, and a lower limit value limiting means which is inserted in the signal processing path or the signal feedback path and limits a lower limit value of a signal to be processed. Sound presence judgment circuit, wherein the door. 2. The lower limit value limiting means sets a lower limit value smaller than the reference value, and a comparator for comparing the output of the lower limit value generator with the input of the lower limit value limiting means. And a selection circuit for selecting one of the output of the lower limit value generator and the input of the lower limit value limiting means as the output of the lower limit value limiting means according to the output of the comparator. Sound judgment circuit.