JP2000148196A - Voice input control method and device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent generation of a large noise in regeneration or the sudden change of volume to auditorially deteriorate the sound quality when a silent voice signal or a voice signal containing a fixed noise or unexpected noise is inputted. SOLUTION: This method comprises steps S102, S104, S105 for detecting whether an input voice signal is silent or a fixed noise, or whether the input contains a large noise or not, step S103 for muting the coded input voice date when the input signal is silent or fixed noise, steps S111, S112 for comparing an average value of the input voice data with a set level and changing the inputlevel by one step from the comparison result, and step S113 for regulating the timing of changing the input level to prevent the output of an unnecessary noise in regeneration or the sudden change of volume.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech processing system, and more particularly, to a method and apparatus for controlling a speech input signal containing silence, constant noise, or sudden noise that is suddenly large noise.

[0002]

2. Description of the Related Art For example, in "Audio Editing Method and Apparatus" described in Japanese Patent Application Laid-Open No. H08-115098, when music or voice is taken into a computer, an analog signal is digitized by an AD converter or the like. A method of monitoring an output value for a certain period of time, calculating an average value thereof, comparing the average value with a predetermined optimum value, and automatically setting a signal level input to an AD converter or the like is adopted.

The audio editing method described in the above publication will be described in detail with reference to FIG.

First, in step S901, a specified level value and a monitoring time are set. The specified level value is a value for determining whether or not audio data obtained by digitizing an audio input signal from a sound source by an AD converter or the like has an appropriate level. The monitoring time is a time for monitoring the converted digital value.

Next, in step S902, audio data is fetched, and the audio data is added. In step S903, the number of input data is counted. Then, the monitoring time is determined in step S904. As a result, if the monitoring time has not expired, the process returns to step S902,
The input voice data is added again. On the other hand, if the monitoring time has expired, the average value of the audio data is calculated in step S905.

Subsequently, in step S906, the average value is compared with the specified level value set in step S901, and if the average value is larger than the specified level value, step S907 is performed.
To set the input level low, while if the average value is smaller than the specified value, the input level is increased in step S908.

[0007] By performing such processing, music, voice, and the like at an appropriate level can be taken into the computer, so that a good voice signal can be reproduced using the voice data subjected to this processing. .

[0008]

The above-described audio editing method has two major problems. The first problem is that the quality of the reproduced audio signal is degraded depending on the state of the input audio signal. The second problem is that when the level of the input audio signal changes suddenly, the sound suddenly increases when the audio signal is reproduced,
That is, the reproduced sound becomes undesirably unacceptable due to a decrease in volume.

Therefore, two problems will be described in detail below.

First, the first problem will be described. The quality of the reproduced sound is degraded when the input audio signal is silent or constant noise, or when the input includes breakthrough large noise.

First, the case where the input audio signal is silent will be described with reference to FIGS.

FIG. 10 (a) shows a signal waveform when silence continues from time t1 to time t2, and the vertical axis shows the amplitude of the input audio signal. As can be seen from the figure, in this case, the input audio signal amplitude is lower than the input determination specified value.

FIG. 10B shows input audio data obtained by digitally encoding the input audio signal shown in FIG. 10A, and corresponds to the audio data input in step S901 in FIG. When this silent voice data is input, it is determined in step S906 in FIG. 9 that the average value is smaller than the specified value, and the level is increased in step S908.

FIG. 10C shows how the level of the input voice data mainly consisting of n3 to n4 noise following the data string number n2 in FIG. 10B is increased. When the input audio data is reproduced, as shown as a reproduced audio signal in FIG.
From No. 3, audio data mainly including noise is greatly amplified, and is output as auditory noise.

Next, the case where the input sound is constant noise will be described with reference to FIGS.

FIG. 11A shows a signal waveform when constant noise continues from time t1 to time t2.
As in the case of the silence shown in FIG. 3A, the amplitude of the input audio signal is always lower than the specified input determination value.

FIG. 11 (b) shows input voice data obtained by digitally encoding the input voice signal shown in FIG. 11 (a). 9, it is determined that the average value is smaller than the specified value, and the level is increased in step S908.

FIG. 11 (c) shows how the level of the input voice data mainly consisting of noises n3 to n4 following the data string number n2 in FIG. 11 (b) is increased. At the time of reproduction, as shown as a reproduced audio signal in FIG.
From No. 3, voice data mainly including noise is greatly amplified and output.

Finally, a case where a large sudden noise is mixed in the input voice signal will be described with reference to FIGS. 12 (a) and 12 (b).

When noise much larger than the specified value is input at time t1, it is determined in step S906 in FIG. 9 that the average of the input values is larger than the specified value, and step S9 is performed.
At 07, the level is reduced so as to reduce the input level. As a result, the amplitude of the input voice data suddenly decreases as shown in FIG. 12B in the period of the data string numbers n2 to n3 corresponding to the time t2 to the time t3 after the time t1. Therefore, when this signal is reproduced, the sound is suddenly reduced for a certain period of time, which is extremely unacceptable for hearing.

Next, a second problem will be described with reference to FIG.
This will be described with reference to (a) and (b).

In FIG. 13A, when a sound signal considerably larger than the input determination specified value for determining the input level is continuously input from time t1 to t2, the input level is lowered in step S907 in FIG. However, the level of the input audio signal drops sharply in order to make the average value of the input audio signal close to the specified value. Therefore, the amplitude of the input audio signal suddenly decreases from time t4 immediately after the level change of the input audio signal is performed.

Accordingly, as shown in FIG. 13 (b), the volume of the reproduced audio signal obtained by reproducing the audio signal suddenly decreases from time t4. Further, when a signal having a sound considerably smaller than a specified value is input, which is not shown in the figure, the volume suddenly increases at a certain time during reproduction. as a result,
Hearing of the reproduced sound is extremely poor.

Therefore, an object of the present invention is to control the input level in accordance with the type and size of the input audio signal, optimize the digitally encoded audio data, and change the level of the input audio signal. It is an object of the present invention to provide a voice input control method and apparatus which does not impair hearing by appropriately adjusting timing.

[0025]

Therefore, a voice input control method according to the present invention is a voice input control method for storing and processing voice data obtained by digitally encoding an input voice signal from a sound source in block units. A first step of determining whether or not audio data is silent in units of the block; and, if determining that the audio data is silent, determining an amplitude level of the input audio signal corresponding to a block next to the block. A second step of not changing and muting the audio data of the block, a third step of determining whether or not the audio data has constant noise in units of the block, and the audio data having constant noise. If it is determined that there is no change in the amplitude level of the input audio signal corresponding to the next block of the block, and the A fourth step of muting voice data; a fifth step of determining whether or not the voice data includes suddenly large sudden noise; and a fifth step of determining whether or not the voice data includes suddenly large sudden noise. A sixth step of not changing the amplitude level of the input audio signal corresponding to the next block after the block, and a seventh step of calculating an average value from the amplitude information of the audio data. And comparing the average value and the set value, and when the average value is larger than the set value, lowers the input level of the input audio signal, and when the average value is smaller than the set value, An eighth step of raising the input level of the input audio signal.

Further, the audio input control device according to the present invention comprises an input level adjusting circuit for changing an input level of an input audio signal from a sound source by a control signal, and digitally encoding an output signal of the input level adjusting circuit in block units. A / D converter for converting the audio data into speech data, and determining whether or not the voice data is silent, whether or not the noise is constant, or whether or not the voice data includes suddenly large sudden noise. An input audio signal determination circuit that outputs a signal, a mute circuit that mutes the audio data of the block corresponding to the determination result when the determination result is silent or constant noise, and an average from the amplitude information of the audio data. A level calculating circuit for calculating a value, comparing the average value with a set value, and when the average value is larger than the set value, An input level determination circuit that outputs the control signal that increases the input level of the input audio signal to the input level adjustment circuit when the input level of the input signal is lowered and the average value is smaller than the set value. Provided.

[0027]

Next, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a flowchart showing a voice input control method according to the present invention. First, step S101 in FIG.
Use to set each parameter. Here, each parameter determines whether or not the input voice signal is silent or constant noise, whether or not the input voice signal includes sudden noise, and whether or not the input voice signal is at an appropriate level. This is the set value for

Next, in step S102, it is determined whether or not the input audio signal is silent. If the input audio signal is silence, a mute process is performed to set the input audio data corresponding to the silence to "0" in the input audio data digitally encoded in block units in step S103. Further, it is set so that the level of the input audio signal is not changed irrespective of the result of the data processing performed after step S104. If it is determined in step S102 that the input audio signal is not silent, the process of the next step S104 is performed.

In step S104, it is determined whether or not the input voice signal has a constant noise. Here, the constant noise means noise in which the amplitude of the noise is substantially constant. If the input voice signal is constant noise, steps S102 and S103
The same processing is performed. If it is determined in step S104 that the input voice signal is not constant noise, the next step S1
05 is performed.

In step S105, it is determined whether or not sudden noise is included in the input voice signal. If sudden noise is included in the input audio signal, control is performed so that the level of the input audio signal is not changed in step S105. If no sudden noise is included in the input voice signal in step S105, the process of the next step S106 is performed.

In step S106, the input voice data obtained by digitizing the input voice signal is added, and in step S107, the number of input voice data added is counted.
Then, in step S108, the set count set in step S101 is compared with the count N calculated in step S107, and if the count N has not reached the set count, the process returns to step S106 to continue the addition process. .

On the other hand, if the count number N has reached the set count number, the average of the input voice data is calculated in step S109.

Subsequently, in step S110, the set level set in step S101 to determine whether or not the input voice data is at an appropriate level, and in step S10
Compare with the average value calculated in 9. If the average value is smaller than the set level, the process proceeds to step S1 in step S111.
In step S06, all audio data used for the addition
A level-up process for adding the adjustment addition value set in 101 is performed.

On the other hand, when the average value is larger than the set value, a level down process is performed in step S112 in which the adjustment subtraction value set in step S101 is subtracted from all the input voice data used in the addition process in step S106. .

Next, in step S113, when level up or level down is performed, timing for executing level up or level down is determined.

Then, in step S114, step S1
When the processing of step S113 is performed, the level of the input voice signal is increased, and when the processing of step S112 is performed, the level of the input voice signal is reduced, so that the input is performed at the timing determined in step S113. Change the level of the audio signal. Then, in step S115,
The count value of the number of input voice data counted in step S107 is cleared.

Next, the audio data processed in step S116 is transferred to a storage unit (not shown) for storing the audio data.

Finally, in step S117, step S
It is determined whether a series of processing from 102 to step S116 has been completed. If the processing is in progress, step S1
The processing from 02 is repeated, but if a series of processing is completed, all data processing ends here.

Next, a description will be given of a method for determining whether or not the input speech signal contains silence, constant noise, or sudden noise.

First, when the input audio signal is silent or has constant noise, the temporal change in amplitude of the input audio signal is extremely small. Therefore, if the value of the input voice data is extremely small and the data change is small, it can be determined that there is no sound. Also,
Whether the value of the input audio data is smaller than the first threshold,
Alternatively, if the value is larger than the second threshold and the change in data is small, it can be determined that the noise is constant. That is, for a normal input audio signal, the value of the input audio data is in the range between the first threshold value and the second threshold value, and includes some frequency components.

Further, if sudden noise is included in the input speech signal, it can be determined that sudden noise is included because the change between data is extremely large.

Next, the silence determination processing in step S102 will be described in detail with reference to FIG.

First, in step S201, a counter for counting the number of times of processing of the input voice data processed in the silence determination processing S102 is cleared. Then, step S2
In 02, it is determined whether or not the input voice data is greater than a silence determination value for determining whether or not the input voice data set in step S101 is silent.

If the input voice data is smaller than the silence determination value, there is a possibility that there is silence.
The process of determining whether or not there is no sound is further continued in the process of 203,
If the input voice data is larger than the silence determination value,
Step S104 is performed.

Next, from step S203 to step S2
At 05, input voice difference data which is a difference between continuous input voice data is calculated. Then, step S206
Then, the input voice difference data is compared with a silence change value that is the first determination value for the input voice difference data set in step S101. If the input voice difference data is smaller than the silence change value, in step S207, "1" is set to the silence count number of the input voice data.

Next, in step S208, it is determined whether or not the number of processes from step S202 to step S207 has reached the set count number Nbl representing the number of audio data constituting one block of audio data. For example, the process returns to step S202 to continue the above processing.

In step S208, step S2
When the number of times of processing from 02 to step S207 has reached the set count number Nbl, in step S209, the silence count number is compared with the silence count determination value set in step S101, and the silence count number is set to the silence count determination value. If it is lower than the above, the process of step S104 is performed, and if the silence count number exceeds the silence count determination value, the mute process of step S103 is performed.

In this way, if the input voice data is smaller than the silence determination value and the number of input voice data satisfying the condition that the input voice difference data is smaller than the silence change value is larger than the silence count determination value, it is determined that there is no sound. Then, a mute process is performed in step S103. The muted input audio data is all “0”. At this time, the amplitude level of the input audio data of the next block is not changed, and the state before silence is maintained, so that the input audio signal can be easily adjusted even when valid audio is input again.

If the input voice data is larger than the silence judgment value in step S202, it is judged that the sound is not silence, and it is judged whether or not the noise is constant in step S104 in FIG.

Next, the constant noise determination processing in step S104 will be described in detail with reference to FIG.

First, in step S301, a counter for counting the number of times of processing of the input voice data processed in the constant noise determination processing S104 is cleared. Subsequently, in step S302, the input voice data falls within a range of a fixed noise determination value 1 for determining whether or not the input voice data set in step S101 is constant noise and a constant noise determination value 2 larger than this. It is determined whether or not it has entered.

If the input voice data is out of the range between the fixed noise determination value 1 and the fixed noise determination value 2, that is, if the input voice data is small or very large, there is a possibility that the noise is constant noise. In the process of step S303, the process of determining whether the noise is constant noise is further continued. If the input voice data is within the range of the fixed noise determination value 1 and the fixed noise determination value 2, the process of step S105 in FIG. I do.

Next, from step S303 to step S3
At 05, input voice difference data which is a difference between continuous input voice data is calculated. Then, in step S306,
The input voice difference data is compared with a constant noise change value that is a second determination value for the input voice difference data set in step S101. If the input voice difference data is smaller than the constant noise change value, in step S307, "1" is set to the constant noise count number of the input voice data.

The silence noise change value as the first judgment value of the input speech difference data and the constant noise change value as the second judgment value for the input speech difference data are different even if they are the same value. May be set as follows. This selection is appropriately selected according to the characteristics of the noise.

Next, in step S308, a set count number N representing the number of data constituting one block of the audio data is set.
It is determined whether or not the number of processes from step S302 to step S307 has reached bl, and if not, the process returns to step S302 to continue the above process.

Also, in step S308, step S3
When the number of times of processing from 02 to step S307 has reached the set count number Nbl, in step S309, the fixed noise count number is compared with the fixed noise count determination value set in step S101, and the fixed noise count number becomes constant noise. If the count is less than the count determination value, step S
If the processing of step 105 is performed and the fixed noise count number exceeds the fixed noise count determination value, step S103
Mute processing.

As described above, the number of input voice data which is outside the range determined by the fixed noise determination value 1 and the fixed noise determination value 2 and whose input voice difference data satisfies the condition smaller than the constant noise change value is: If it is larger than the constant noise count determination value, it is determined that the noise is constant, and a mute process is performed in step S103. The muted input audio data is all “0”. At this time, the amplitude level of the input audio data of the next block is not changed, and the state before silence is maintained, so that the input audio signal is input even when valid audio is input again as in the case where silence is input. Can be easily adjusted.

If the input voice data is within the range between the fixed noise judgment value 1 and the fixed noise judgment value 2 in step S302, it is judged that the noise is not constant noise, and step S1 in FIG.
It is determined whether or not sudden noise 05 is included.

Finally, the process of determining whether or not sudden noise is included in the input voice signal in step S105 will be described in detail with reference to FIG.

First, in step S401, the value of the counter for counting the sudden noise count is cleared, and in step S402, the counter for counting the number of processes in step S105 for determining whether or not the sudden noise is included is cleared. .

Next, steps S403 to S4
At 05, input voice difference data which is a difference between continuous input voice data is calculated. Then, in step S406,
The input voice difference data is compared with the sudden noise change value that is the third determination value for the input voice difference data set in step S101. If the input voice difference data is larger than the sudden noise change value, in step S407, "1" is set to the sudden noise count number.

Next, in step S409, a set count number N representing the number of data constituting one block of the audio data is set.
It is determined whether or not the number of processes from step S403 to step S408 has reached bl. If not, the process returns to step S403 to continue the above process.

In step S409, step S4
When the number of times of processing from 03 to S408 reaches the set count, in step S410, the sudden noise count is compared with the sudden noise count determination value set in step S101, and the sudden noise count is calculated. If it is less than the determination value, step S106
When the sudden noise count number exceeds the sudden noise count determination value, the process of step S115 is performed.

As described above, if there is sudden noise, no change is made to all data of this voice block. Then, in step S116, the audio data is transferred to the memory.

Next, the processing of the sudden noise described above will be described with reference to FIG.

FIG. 5A shows an input audio signal at time t.
The sudden noise occurs at b. However, at time t2
As shown in FIG. 5B, the level of the reproduced audio signal suddenly decreases because the level of the reproduced signal from time t3 to time t4 subsequent to the time t3 is not changed due to the sudden noise at time tb. That is, the problem that the reproduced sound signal, which has conventionally been an auditory problem, suddenly drops is improved.

Although not particularly described above, FIG.
As shown in (c), a smooth reproduced audio signal can be generated by complementing the sudden noise at time tb using the input audio data before and after.

Next, an embodiment of the voice input control device of the present invention will be described with reference to the drawings.

FIG. 6 is a block diagram showing an embodiment of the voice input control device according to the present invention, wherein an AD converter 601 for converting an input voice signal into a digital signal and an AD converter 60 are provided.
An input level adjusting circuit 602 for adjusting the level of the input audio signal to the input signal 1; an input audio signal determination circuit for determining whether the input audio signal is silent or constant noise, or determining whether or not the input audio signal includes sudden noise; 603, a mute circuit 604 for muting the input audio data digitally encoded by the AD converter 601 based on the judgment result of the input audio signal judgment circuit 603, a level calculation circuit 605 for calculating an average value of the input audio data, and a level calculation circuit An input level determination circuit 606 that compares the average value calculated in 605 with the determination value and determines whether to adjust the input level of the input audio signal of the AD converter 601. A timing adjustment circuit 607 for determining whether to perform the timing, an input audio signal determination circuit 603, a level calculation circuit 605, and a timer. A timing adjustment circuit 607, the CPU608 giving the set values corresponding to the parameters described in step S101 in FIG. 1, and a storage unit 609 for storing the processed voice data.

Next, the operation of the voice input control device shown in FIG. 6 will be described.

The input audio signal input to the AD converter 601 is digitally converted into input audio data in which one block is composed of a plurality of audio data. The input audio data is input to an input audio signal determination circuit 603 and a mute circuit 60.
4 and the level calculation circuit 605. The input voice signal determination circuit 603 determines whether the input voice data input is silent or constant noise, or whether the input voice data includes sudden noise.

When the input voice data is silent or constant noise, the input voice signal determination circuit 603
3 and outputs an input level change prohibition signal S2 to the input level determination circuit 606.

When the mute signal S 1 is output from the input audio signal determination circuit 603, the mute circuit 604 sets the input audio data to “0” and transfers this data to the storage unit 609. If the input audio data includes sudden noise, the input audio signal determination circuit 603 outputs the mute signal S1
And outputs an input level change prohibition signal S2 to the input level determination circuit 606. At this time, the mute circuit 604 outputs the input audio data to the storage unit 609 without performing the mute process.

If the input voice data is not any of the above, that is, none of the input voice data including silence or constant noise or sudden noise, the input voice data output from the AD converter 601 is output to the mute circuit. The processing is not received in 604 and stored in the storage unit 609.

The level calculation circuit 605 calculates an average value of the input audio data. This average may be an addition average of sequentially adding input audio data, a square average representing the power of the audio signal, or a moving average. The obtained average value is compared with a set value for judging the appropriateness of the input level in the input level judgment circuit 606. If the average value is larger than the set value, the input level is decreased by one step. , A level adjustment signal S3 for setting the input level to be raised by one step is output to the input level adjustment circuit 602 via the timing adjustment circuit 607.
Here, the timing adjustment circuit 607 adjusts the output timing of the level adjustment signal S3, that is, the timing at which the input level is changed.

With the above configuration, the same processing as the voice input control method shown in FIG. 1 can be performed.

As shown in FIG. 7A, when the level of the input audio signal is large, the input level is changed one to several steps at a time. The signal amplitude can be gradually reduced. Therefore, the reproduced sound is good in hearing. In addition, since the time t4 at which the level change is started can be arbitrarily set by the timing adjustment circuit 607, a more perceptually preferable reproduced sound can be obtained.

Next, an embodiment of the voice input control device shown in FIG. 6 will be described with reference to FIG. Note that components common to those in FIG. 6 are denoted by common reference characters / numbers.

The audio input control device shown in FIG. 8 includes a variable amplifier 6021 forming an input level adjustment circuit 602,
A D converter 601, a data register with reset 6041 constituting the mute circuit 604, a storage unit 609,
Counter 6071 constituting timing adjustment circuit 607
, A register 6072, and a CPU 608.

Further, the input audio signal determination circuit 603
A latch circuit 6031 for latching input audio data output from the D converter 601, and a subtractor 6 for subtracting output data from the latch circuit 6031 and input audio data
032, comparators 6033 and 6034, an UP counter 6035, and registers 6036 and 6037.

Further, the level calculation circuit comprises an adder 6051
, A latch circuit 6052, an averaging circuit 6053, a register 6054, and an UP counter 6055.

The input level judging circuit 606 comprises a comparator 6061, a register 6062, an UP / DOWN counter 6063, and a decoder 6064.

Next, the operation of the voice input control device of FIG. 8 will be described.

First, the input audio signal is the level adjustment signal S3
AD through a variable amplifier 6021 whose gain is controlled by
The data is input to the converter 601 and one block is digitally converted into input audio data composed of a plurality of audio data.

The input audio data is input to a latch circuit 6031 constituting an input section of the input audio signal determination circuit 603, and when the mute signal S1 is input to the reset terminal, the output becomes "0" and the data register with reset 60 performs mute processing.
41, input to an adder 6051 constituting an input unit of the level calculation circuit 605.

First, the input audio data D2 and the input audio data D2 latched by the latch
The input voice data D1 before the data is subtracted by the subtractor 6032, and the input voice difference data that is the result of the subtraction is input to the comparator 6033.

The comparator 6033 sequentially compares the inputted input voice difference data with the silent change value X1, the constant noise change value X2, and the sudden noise change value X3 stored in the register 6036.

The silence change value X1 is a set value for judging whether or not there is silence, the constant noise change value X2 is a set value for judging whether or not there is constant noise, and the sudden noise change value X3 is a sudden noise Is a setting value for determining whether or not the data includes the data, and has already been described with reference to FIGS.

First, the CPU 608 sets the register 6036
The silence change value X1 is set in the register 6037, the silence counter determination value is set in the register 6037, and the UP counter 6035 is cleared.

The comparator 6033 determines the magnitude of the silence change value X1 and the input speech difference data output from the subtractor 6032, and if the input speech difference data is smaller than the silence change value, sets “1”. If the data is larger than the silent change value, “0” is output to the UP counter 603.

The UP counter 603 counts up each time "1" is output from the comparator 6033, and
034 is output as the number of silence counters. This behavior is
This is repeated by the count number Nbl.

The comparator 6034 compares the number of silence counters with the silence counter judgment value set in the register 6037. If the silence counter number is larger than the silence counter judgment value, that is, if the input voice data is silence. If it is determined, “1” is output to the data register with reset 6041 and the comparator 6061 as a mute signal S1.

The data register with reset 6041 is
When the mute signal S1 becomes "1", a mute operation is performed. That is, the audio data stored in the data register with reset 6041 is set to “0”,
9 to output this “0” data.

The operation described above substantially corresponds to the silence determination processing described with reference to FIG.

Next, the CPU 608 sets the constant noise change value X2 in the register 6036, the constant noise counter judgment value in the register 6037, and sets the UP counter 6035.
And performs an operation corresponding to the constant noise determination process shown in FIG. Since the operation is the same as the silence determination processing described above, the description is omitted. However, when the constant noise counter number is larger than the constant noise counter determination value, the comparator 6034
That is, when it is determined that the input voice data has constant noise, “1” is output to the data register with reset 6041 and the comparator 6061. Receiving the mute signal S1, the data register with reset 6041 performs a mute operation also in this case.

Next, the CPU 608 sets the sudden noise change value X3 in the register 6036, the sudden noise counter judgment value in the register 6037, and sets the UP counter 6035.
Is cleared, and an operation corresponding to the sudden noise determination process shown in FIG. 4 is performed. The comparator 6034 outputs “1” only to the data register with reset 6041 when the number of sudden noise counters is larger than the sudden noise counter determination value, that is, when it is determined that the input voice data includes sudden noise.

The input audio data input to the level calculation circuit 605 is added by the adder 6051 and the latch circuit 6052 by the count number Nbl, and the average value is obtained by the averaging circuit 6053 based on the addition result. Note that when counting up to the value set in the register 6054, the UP counter 6055 outputs a control signal to the latch circuit 6052 so as to transfer the addition result of the latch circuit 6052 to the averaging circuit 6053.

The average value, which is the output signal of the level calculation circuit 605, is calculated by the comparator 6 constituting the input level determination circuit 606.
061. Here, the input level setting value stored in the register 6062 is calculated by the comparator 6061,
Comparison with the average value is performed. If the average value is smaller than the input level set value, UP / DOW is set to increase the input level.
The count value of the N counter 6063 is increased by one to three, and when the average value is larger than the input level set value, the count value of the UP / DOWN counter 6063 is decreased by one to three to reduce the input level. UP / DOWN counter 606
3 is input to the decoder 6064,
The decode value which is the output of 064 determines the amount of amplification and attenuation of the variable amplifier 6021.

Here, the UP / DOWN counter 6063
Is one step to several steps, and the amount of amplification and attenuation of the variable amplifier 6021 is also one step to several steps. The number of steps is selected by a program stored in the CPU 608.

Also, the comparator 60 of the input audio signal determination circuit
If it is determined in step 34 that the input audio signal is silent or constant noise or that the input audio signal includes sudden noise, the UP / DOW is determined regardless of the comparison result of the comparator 6061.
The output of the N counter 6063 does not change.

The output of the decoder 6064 is
After the counter 6071 counts up to the value set to 72, the variable amplifier 6
021 is applied to a control terminal (not shown). In response, the variable amplifier 6021 amplifies or attenuates the input audio signal. That is, the counter 6071 and the register 6
072 determines the timing of supplying the control signal from the decoder 6064 to the variable amplifier 6021.

With the above configuration, the same operation as that of the voice input control device shown in FIG. 6 can be realized.

In the above description, the circuit configuration and operation of the input voice signal determination circuit 603 have been described as an example of serial processing such as silence determination → constant noise determination → sudden noise determination. A determination circuit, a constant noise determination circuit for performing constant noise determination, and a sudden noise determination circuit for performing sudden noise determination may be provided, and these may be operated in parallel. At this time, the output of the silence determination circuit and the output of the constant noise determination circuit may be input to an OR gate, and the data register with reset 6041 may be reset by the output of the OR gate.

Although this method has a larger circuit scale than the serial method described above, it is excellent in high-speed performance, and is suitable for a case where real-time performance is strongly required.

The output of the comparator 6034 has a configuration in which a signal line for outputting the mute signal S1 and a signal line for outputting the determination result of the sudden noise are connected to the input of the comparator 6061. However, a selector circuit is provided at the output of the comparator 6034, and the output of the selector circuit is switched between the case of silent or constant noise and the case of sudden noise, and the signal line from the selector circuit to the comparator 6061 is reduced to one. It is possible to

In the level calculation circuit 605, the input audio data is sequentially added and an average is taken.
A squaring circuit for squaring the input audio data may be provided between the output of the converter 601 and the input of the adder 6051 to take the root mean square.

Further, the averaging circuit 6053 may be deleted, and the addition result of the latch circuit 6052 may be directly input to the comparator 6061. In this case, the set value set in the register 6062 may be multiplied by the count number Nbl of the set value when the averaging circuit is also provided. This eliminates the necessity of the division circuit forming the averaging circuit 6053, so that the circuit scale can be reduced.

[0109]

As described above, the voice input control method and apparatus according to the present invention detects a silent or constant noise or a sudden noise included in an input voice signal, and detects AD when a silent or constant noise is input. The input audio data encoded by the converter is muted, and the muted audio data is transferred to a predetermined storage unit.

Further, when no sound, constant noise, or sudden noise is input, the input level is not changed. Thus, unnecessary noise can be eliminated during reproduction, and the level of the reproduced audio signal can be optimized.

If the input voice signal does not include silence, constant noise, or sudden noise, the average value of the coded input voice data is calculated, and as a result, the level of the input voice signal needs to be adjusted. In this case, the level to be changed is one stage to several stages, and the timing for changing the input level is adjusted.

As a result, the level of the input audio signal does not change abruptly, so that the change in the volume of the reproduced sound is smooth, and a reproduced sound that is aurally favorable can be obtained.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a processing procedure of a voice input control method according to an embodiment.

FIG. 2 is a flowchart showing a detailed processing procedure of silence determination processing S102 in FIG. 1;

FIG. 3 is a flowchart illustrating a detailed processing procedure of a constant noise determination processing S104 in FIG. 1;

FIG. 4 is a flowchart showing a detailed processing procedure of the sudden noise determination processing S105 of FIG. 1;

FIG. 5 is a signal waveform diagram for explaining a process in a case where sudden noise is included in an input audio signal in the audio input control method according to the present embodiment.

FIG. 6 is a block diagram illustrating a voice input control device according to the present embodiment.

FIG. 7 is a signal waveform diagram for describing an operation when an excessive input audio signal is input in the audio input control device of the present embodiment.

FIG. 8 is a circuit diagram showing a main part of the voice input control device of FIG. 6;

FIG. 9 is a flowchart showing a processing procedure of a conventional voice input control method.

FIG. 10 is a signal waveform diagram for describing an operation when silence is input in a conventional voice input control method.

FIG. 11 is a signal waveform diagram for describing an operation when constant noise is input in the conventional voice input control method.

FIG. 12 is a signal waveform diagram for explaining an operation when an input audio signal including sudden noise is input in the conventional audio input control method.

FIG. 13 is a signal waveform diagram for describing a reproduction process when an excessive input audio signal is input in the conventional audio input control method.

[Explanation of symbols]

601 AD converter 602 Input level adjustment circuit 603 Input audio signal determination circuit 604 Mute circuit 605 Level calculation circuit 606 Input level determination circuit 607 Timing adjustment circuit 608 CPU 609 Storage unit 6021 Variable amplifier 6031, 6052 Latch circuit 6032 Subtractor 6033, 6034 , 6061 Comparator 6035, 6055 UP counter 6036, 6037, 6054, 6062, 6072
Register 6041 Data register with reset 6051 Adder 6053 Averaging circuit 6063 UP / DOWN counter 6064 Decoder 6071 Counter

Claims (12)

[Claims] 1. An audio input control method for accumulating and processing audio data obtained by digitally encoding an input audio signal from a sound source in units of blocks, wherein a method of determining whether or not the audio data is silent in units of blocks. Step 1; if it is determined that the audio data is silent, the amplitude level of the input audio signal corresponding to the block next to the block is not changed, and the audio data of the block is muted. A second step, a third step of determining whether or not the audio data has constant noise in units of the block; and a determination that the audio data has constant noise corresponds to a block next to the block. A fourth step of not changing the amplitude level of the input audio signal to be performed and muting the audio data of the block; A fifth step of determining whether or not the audio data includes suddenly large sudden noise; and determining that the audio data includes suddenly large sudden noise. A sixth step of not changing the amplitude level of the input audio signal corresponding to the block, a seventh step of obtaining an average value from the amplitude information of the audio data, and comparing the average value with a set value. When the average value is larger than the set value, the input level of the input audio signal is lowered, and when the average value is smaller than the set value,
An eighth step of increasing the input level of the input audio signal. 2. The method according to claim 1, wherein the first step is a ninth step of calculating input voice difference data which is a difference between the voice data and the next input voice data. A tenth step of comparing the input voice difference data and calculating a count number of the voice data belonging to the one block that satisfies a condition that is smaller than the difference set value; The voice input control method according to claim 1, further comprising an eleventh step of determining that there is no sound when the volume is large. 3. The third step is a twelfth step of calculating input voice difference data that is a difference between the voice data and the next input voice data. A thirteenth step of calculating a count number of the audio data belonging to the one block that satisfies a condition that the input audio difference data is smaller than the difference set value; The voice input control method according to claim 1, further comprising a fourteenth step of determining that the noise is constant when the noise is large. 4. The fifth step is a fifteenth step of calculating input voice difference data which is a difference between the voice data and the next input voice data. And calculating a count number of the audio data belonging to the one block that satisfies a condition that the input voice difference data is larger than the difference setting value; and the count number is larger than a count determination value. The voice input control method according to claim 1, further comprising: a seventeenth step of determining a sudden noise when the noise is large. 5. The voice input control method according to claim 1, further comprising an eighteenth step of adjusting a timing of changing the input level when changing the input level. 6. The voice input control method according to claim 1, wherein the average value is calculated by an addition average, a square average, or a moving average. 7. An input level adjusting circuit for changing an input level of an input audio signal from a sound source by a control signal, and an AD converter for converting an output signal of the input level adjusting circuit into audio data digitally encoded in block units. An input audio signal determination circuit that determines whether or not the audio data is silent, whether or not it is constant noise, or whether or not the audio data includes suddenly large sudden noise, and outputs a determination result. When the determination result is silent or constant noise, a mute circuit that mutes the audio data of the block corresponding to the determination result, and a level calculation circuit that calculates an average value from amplitude information of the audio data, Comparing the average value with a set value, and if the average value is larger than the set value, lower the input level of the input audio signal; No. is less than the set value, the input speech signal of the input level judging circuit for outputting the control signal to the input level adjusting circuit to raise the input level, an audio input control apparatus provided with. 8. When the determination result is any of silent, constant noise, or sudden noise, the input level determination circuit determines the input level regardless of a result of comparison between the average value and the set value. The voice input control device according to claim 7, wherein the control signal is output so as not to change the control signal. 9. The audio input control device according to claim 7, further comprising a timing adjustment circuit for inputting the control signal and adjusting timing of the input level change when changing the input level. 10. The input audio signal determination circuit includes: a latch circuit that latches the audio data; a subtractor that generates input audio difference data that is a difference between output data of the latch circuit and the audio data; A first comparator that compares the input voice difference data with the change determination value and outputs a comparison result; an UP counter that counts up the comparison result and outputs a counted number; And comparing the count value with the count determination value, and when the count number is larger than the count determination value, outputs a signal that becomes active.
The audio input control device according to claim 7, further comprising: a comparator. 11. The level determination circuit, comprising: a third comparator for comparing the average value with the set value; and when the average value is smaller than a previous set value, based on an output of the third comparator. , Increase the count value, and if the average value is greater than the set value, decrease the count value.
A P / DOWN counter and an output signal of the UP / DOWN counter,
The audio input control device according to claim 7, further comprising: a decoder that outputs the control signal to the input level adjustment circuit. 12. The audio input control device according to claim 9, wherein the timing adjustment circuit has a counter that outputs the control signal from the input level determination circuit to the input level adjustment circuit after counting a set value.