JP2014052553A - Sound volume correction device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sound volume correction device capable of reducing the amount of volume correction processing.SOLUTION: A buffer part 11 has a storage region capable of storing sound input signals for a predetermined period of time. When the storage region has no space, the buffer part 11 overwrites new sound input signals in a storage region where the oldest sound input signals are stored. Every time when all sound input signals stored in the buffer part 11 are replaced with new ones, a gain calculation section 13 calculates the gain based on a representative value of the sound input signals stored in the buffer part 11. A gain limitation section 14 calculates the amount of change between the gain previously calculated by the gain calculation section 13 and the gain calculated this time. When the absolute value of the amount of change is larger than a predetermined first threshold value, a limitation value which is obtained by changing the previously calculated gain by an amount equivalent to a predetermined changing upper limit so that the previously calculated gain gets closer to the gain calculated this time is determined as the gain his time. A gain multiplication section 15 multiplies the gain which is set by the gain calculation section 13 and a gain limitation section 14 by a sound input signal read out from the buffer part 11 and outputs the same.

Description

  The present invention relates to a volume correction device.

  2. Description of the Related Art Conventionally, there has been a voice transmission system that includes a voice transmission terminal device that transmits a voice signal by a packet and a voice reception terminal device that outputs a voice based on the received voice signal, and performs a voice call in real time (for example, Patent Documents). 1).

  In this audio transmission system, after the audio receiving terminal device decodes one audio frame of audio data included in the packet received from the audio transmitting terminal device, the anti-noise processing is performed on the decoded digital audio signal, The generation of abnormal noise due to transmission errors was suppressed. That is, on the voice transmitting terminal device side, the first difference calculation means calculates the absolute value of the difference between the quantized value of the sample point of the digital voice signal and the quantized value of the immediately preceding sample point. The maximum absolute value is obtained for the frame. Then, the noise countermeasure processing means of the voice receiving terminal device calculates the absolute value of the difference between the quantized value of the sample point of the digital voice signal and the quantized value of the previous sample point, and this absolute value is When the above maximum value is exceeded, it is determined that the noise is abnormal, and processing for suppressing the abnormal noise is performed.

JP 2008-292961 A

  In the voice transmission system that performs a voice call in real time as described above, it is necessary to adaptively adjust the gain as the signal level of the voice input signal changes. Therefore, conventionally, every time an audio input signal is sampled, the amplitude of the entire audio is estimated from the sampled instantaneous value, and the gain is calculated based on the estimated value. Since these processes are performed for each sampling, the amount of signal processing increases, and it is necessary to use a high-performance microcomputer for the arithmetic processing unit.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a volume correction apparatus that reduces the volume of volume correction processing.

  The volume correction apparatus of the present invention includes a buffer unit, a gain calculation unit, a gain limiting unit, and a gain multiplication unit. The buffer unit has a storage area capable of storing a voice input signal for a predetermined time, and when the storage area is full, a new voice input signal is overwritten and saved in the storage area for storing the oldest voice input signal. . The gain calculating unit calculates the gain based on the representative value of the audio input signal stored in the buffer unit every time the audio input signal stored in the buffer unit is replaced. The gain limiting unit obtains a change between the gain calculated by the gain calculation unit last time and the gain calculated this time, and when the absolute value of the change is equal to or greater than a predetermined first threshold, the gain calculated last time is calculated. The limit value obtained by changing by a predetermined fluctuation upper limit so as to approach the gain obtained is set as the current gain. When the gain is set by the gain calculation unit and the gain limiting unit, the gain multiplication unit reads out the audio input signal from the buffer unit in the oldest order, and multiplies the read audio input signal by the gain and outputs the result. .

  In this volume correction apparatus, it is also preferable that the gain calculation unit obtains a maximum absolute value of the audio input signal stored in the buffer unit and sets the maximum value as the representative value.

  In this volume correction apparatus, it is also preferable that the gain calculation unit obtains an average value of absolute values of the audio input signal stored in the buffer unit, and sets the average value as the representative value.

  In this volume correction apparatus, it is also preferable that the gain calculation unit sets a gain so that a value obtained by multiplying the representative value by a gain is equal to a preset target value.

  In this volume correction device, when the gain limiting unit sets the limit value as the current gain and the product of the gain and the representative value exceeds a predetermined second threshold, the gain limiting unit It is also preferable that the gain calculated by the gain calculation unit is the current gain so that a value obtained by multiplying the gain by the gain becomes a predetermined target value.

  In this volume correction device, when the absolute value of the change is equal to or greater than the first threshold value and the gain limiter uses the limit value as the current gain, the buffer unit reduces the size of the storage area. It is also preferable.

  In this volume correction device, if the representative value is equal to or lower than a determination level for determining whether the sound is in a silent state or a sound state, the gain multiplication unit calculates the gain previously calculated by the gain calculation unit from the buffer unit. It is also preferable to multiply the read audio input signal and output it.

  The volume correction apparatus preferably further includes a sound determination unit that determines whether a sound input signal input to the buffer unit is sound or non-speech. When it is determined that all the audio input signals input to the buffer unit by the audio determination unit are non-audio, the gain multiplication unit reads the gain previously calculated by the gain calculation unit from the buffer unit. Multiply the audio input signal and output.

  In this volume correction apparatus, it is also preferable that the gain calculation unit sets the gain to the upper limit value when the calculated gain exceeds a predetermined upper limit value.

  According to the present invention, every time the audio input signal stored in the buffer unit is replaced, the gain calculation unit calculates the gain based on the representative value of the audio input signal stored in the buffer unit. Therefore, compared with the conventional example in which the gain is calculated for each sampling, the processing amount of the volume correction can be reduced, and the gain can be calculated only by the delay corresponding to the storage capacity of the buffer unit, so that the delay of the output sound can be reduced. Further, a change between the previously calculated gain and the currently calculated gain is obtained by the gain calculating unit, and when the absolute value of the change is equal to or greater than the first threshold, the gain limiting unit calculates the previously calculated gain The limit value obtained by changing by a predetermined fluctuation upper limit so as to approach the calculated gain is used as the current gain. As a result, since the volume level of the input voice has suddenly changed, even if the gain calculation result fluctuates more than the first threshold value, the gain fluctuation can be suppressed to the predetermined fluctuation upper limit, and thus the volume of the output voice suddenly changes. Can be suppressed.

It is a block diagram of this embodiment. It is a wave form diagram explaining operation | movement same as the above, (a) is a gain, (b) is a figure which shows the absolute value of an audio | voice input signal. It is a wave form diagram explaining another operation | movement same as the above, (a) is a gain, (b) is a figure which shows the absolute value of an audio | voice input signal. It is explanatory drawing of the method of calculating | requiring a representative value from the audio | voice input signal preserve | saved at the buffer part same as the above. It is explanatory drawing of another method which calculates | requires a representative value from the audio | voice input signal preserve | saved at the buffer part same as the above. It is a wave form diagram explaining another operation | movement same as the above, (a) is a gain, (b) is a figure which shows the absolute value of an audio | voice input signal. It is a wave form diagram explaining another operation | movement same as the above, (a) is a gain, (b) is a figure which shows the absolute value of an audio | voice input signal. It is a wave form diagram explaining another operation | movement same as the above, (a) is a gain, (b) is a figure which shows the absolute value of an audio | voice input signal.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram of an audio output device 1 using a volume correction device 10 according to the present invention.

  The audio output device 1 includes a microphone 2, an A / D conversion unit 3, a volume correction device 10, an encoding / packetization processing unit 4, and a wireless transmission unit 5.

  The microphone 2 converts the input sound into an electrical signal and outputs it to the A / D conversion unit 3.

  The A / D converter 3 performs A / D conversion on the electrical signal input from the microphone 2. Then, the audio input signal converted into a digital signal by the A / D conversion unit 3 is output to the volume correction device 10.

  The volume correction apparatus 10 includes a buffer unit 11, a gain setting unit 12 (consisting of a gain calculation unit 13 and a gain limiting unit 14), and a gain multiplication unit 15 as main components, and is input from the A / D conversion unit 3. The volume level of the input audio signal is corrected. The gain setting unit 12 and the gain multiplication unit 15 are realized by causing a microcomputer to execute a built-in program, for example.

  The buffer unit 11 includes a storage area capable of storing a voice input signal for a predetermined time input from the A / D conversion unit 3. When there is no more free space in the storage area, the buffer unit 11 overwrites and saves the new audio input signal in the storage area for storing the oldest audio input signal.

  The gain setting unit 12 includes a gain calculating unit 13 and a gain limiting unit 14, and determines the gain every time when all the audio input signals stored in the buffer unit 11 are replaced (that is, every predetermined time).

  The gain calculation unit 13 calculates the gain based on the representative value of the audio input signal stored in the buffer unit 11 every time the audio input signal stored in the buffer unit 11 is replaced. Specifically, the gain calculation unit 13 obtains the maximum value of the absolute value of the audio input signal based on the audio input signal for a predetermined time stored in the buffer unit 11, and uses this maximum value as a representative value. Yes. FIG. 4 shows an example of the waveform of the audio input signal (absolute value). The periods T1, T2, T3, and T4 in FIG. 4 correspond to the time lengths of the audio input signals that can be stored in the buffer unit 11, respectively. All the audio input signals stored in the are exchanged. The gain calculation unit 13 obtains the maximum values P1, P2, P3, and P4 of the audio input signals stored in the buffer unit 11 at the end of the periods T1, T2, T3, and T4, and the maximum values P1, P2, P3 , P4 is a representative value of the audio input signal in each period. Then, the gain calculation unit 13 calculates the gain so that a value obtained by multiplying the maximum value (representative value) of the audio input signal in each period by the gain becomes a predetermined target value.

  When the gain calculation unit 13 calculates the gain, the gain limiting unit 14 obtains a change between the gain previously calculated by the gain calculation unit 13 and the gain calculated this time. When the absolute value of this change is equal to or greater than the predetermined first threshold, the gain limiting unit 14 obtains a limit value obtained by changing the previously calculated gain by a predetermined fluctuation upper limit ΔG so as to approach the gain calculated this time. Is output to the gain multiplier 15 as the current gain. On the other hand, if the absolute value of the change described above is less than the first threshold value, the gain limiting unit 14 does not perform the process of limiting the gain, and the gain calculating unit 13 outputs the gain calculated this time to the gain multiplying unit 15 as it is. To do. The first threshold value is set to a value that is at least larger than the variation upper limit ΔG.

  FIG. 2B shows an example of the waveform of the voice input signal, and FIG. 2A shows the gain determined based on the voice input signal. The gain calculation unit 13 calculates the gain every time the audio input signals stored in the buffer unit 11 are replaced, and the gains calculated at the end of the periods T1, T2, T3, and T4 are G1, G2, and G3, respectively. , G4.

  Here, when the gain is calculated by the gain calculating unit 13 at the end of the period T2, the gain limiting unit 14 obtains a change between the previously calculated gain G1 and the currently calculated gain G2, and calculates the change. The absolute value (| G1-G2 |) is compared with the first threshold value. Since the absolute value of the change between the gain G1 and the gain G2 is less than the first threshold, the gain limiting unit 14 sets the gain G2 calculated this time as the gain in the period T2, and outputs the gain to the gain multiplication unit 15. .

  When the gain is calculated by the gain calculating unit 13 at the end of the period T3, the gain limiting unit 14 obtains a change between the previously calculated gain G2 (gain of the period T2) and the currently calculated gain G3. The absolute value of this change is compared with the first threshold value. Since the absolute value of the change between the gain G2 and the gain G3 is less than the first threshold, the gain limiter 14 sets the gain G3 calculated this time as the gain in the period T3 and outputs the gain to the gain multiplier 15. .

  On the other hand, when the gain is calculated by the gain calculation unit 13 at the end of the period T4, the gain limiting unit 14 calculates the change between the previously calculated gain G3 (gain of the period T3) and the gain G4 calculated this time. The absolute value of this change is compared with the level of the first threshold. Here, since the maximum value of the audio input signal is significantly lower in the period T4 than in the period T3, the absolute value of the change between the gain G3 and the gain G4 is equal to or greater than the first threshold value. Therefore, the gain limiter 14 sets a limit value G4b (= G3 + ΔG) obtained by changing the previously calculated gain G3 by a predetermined variation upper limit ΔG in a direction approaching the currently calculated gain G4 as the current gain. . Here, the variation upper limit ΔG is preferably set to about 3 dB in view of human auditory characteristics, and it becomes difficult to notice the change in volume that occurs with the change in gain, so that the user does not feel uncomfortable. The volume change can be suppressed.

  When the gain is set by the gain setting unit 12, the gain multiplication unit 15 reads the audio input signal from the buffer unit 11 in the oldest order, and multiplies the read audio input signal by the gain input from the gain setting unit 12. And output to the encoding / packetization processing unit 4.

  The encoding / packetization processing unit 4 performs voice encoding processing and packetization processing on the signal whose volume has been corrected by the volume correction device 10, and then outputs the signal to the wireless transmission unit 5.

  The wireless transmission unit 5 wirelessly transmits the audio signal subjected to the audio encoding process and the packetization process by the encoding / packetization processing unit 4.

  As described above, in the present embodiment, the voice input from the microphone 2 is A / D converted by the A / D converter 3 and the volume level is corrected by the volume correction device 10, and then the voice encoding process, packet The wireless transmission is performed from the wireless transmission unit 5.

  In the volume correction device 10, the gain calculation unit 13 calculates the gain based on the representative value of the audio input signal stored in the buffer unit 11 every time the audio input signal stored in the buffer unit 11 is replaced. ing. Therefore, compared to the conventional example in which the gain is calculated for each sampling, the processing amount of the volume correction can be reduced, and the gain can be calculated only by the delay according to the storage capacity of the buffer unit 11, so that the delay of the output sound can be reduced. . Further, a change between the previously calculated gain and the currently calculated gain is obtained, and when the absolute value of the change is equal to or greater than the first threshold, the gain limiting unit 14 sets the previously calculated gain to the currently calculated gain. The limit value obtained by changing the previously calculated gain by a predetermined fluctuation upper limit is set as the current gain so as to approach. As a result, since the volume level of the input voice has suddenly changed, even if the gain calculation result fluctuates more than the first threshold value, the gain fluctuation can be suppressed to the predetermined fluctuation upper limit, and thus the volume of the output voice suddenly changes. Can be suppressed.

  By the way, if the time length of the audio signal stored in the buffer unit 11 is too short, the volume correction device 10 reacts excessively to the change in the signal level of the input audio, and the volume correction processing is frequently performed. Therefore, the processing amount of the sound volume correction device 10 increases. Further, if the time length of the audio signal stored in the buffer unit 11 is too long, the buffer unit 11 having a large storage capacity is required, and there is a problem that a delay corresponding to the time that can be stored in the buffer unit 11 occurs. .

  Moreover, it is known that the average time length of one syllable or one character in a normal utterance is about 130 (mSec) (reference: “Basics of Speech Information Processing”, Shuzo Saito, Kazuo Nakata, Published by Ohmsha (1981). In this embodiment, the time of the audio signal stored in the buffer unit 11 is set to about several hundreds (mSec) to several thousand (mSec) so that the volume can be adjusted using data of at least several syllables. It is possible to suppress frequent volume correction processing while shortening the delay due to the ring.

  The gain calculating unit 13 preferably sets the output gain to the upper limit value when the calculated gain exceeds a predetermined upper limit value. In this way, by limiting the gain output from the gain calculation unit 13 within the upper limit value, it is possible to prevent the voice input signal from being amplified with an excessive gain, thereby reducing the floor noise included in the output voice. it can. Further, when used in a call system such as an interphone, it is possible to suppress howling by suppressing the output level.

  Here, when the absolute value of the change between the previously calculated gain and the currently calculated gain is equal to or greater than the first threshold, the gain limiting unit 14 uses the limit value obtained as described above as the current gain. In addition, it is also preferable that the buffer unit 11 reduces the size of the storage area.

  FIG. 3B shows an example of the waveform of the voice input signal, and FIG. 3A shows the gain determined based on the voice input signal. The gain calculation unit 13 calculates the gain every time the audio input signal stored in the buffer unit 11 is replaced. The gain calculation unit 13 also stores the audio input signal stored in the buffer unit 11 at the end of the periods T11, T12, and T13. The gains calculated in the above are G11, G12, and G13.

  At the end of the period T12, since the absolute value (| G11−G12 |) of the change between the previously calculated gain G11 and the currently calculated gain G12 is less than the first threshold, the gain limiting unit 14 The calculated gain G12 is defined as the current gain.

  On the other hand, in the period T13, the volume level of the audio input signal is significantly lower than in the period T12. Therefore, the absolute value of the change between the gain G12 calculated in the previous period (in the period T12) and the gain G13 calculated this time is It is greater than or equal to the first threshold. Therefore, the gain limiting unit 14 sets the limit value G13b obtained by changing the previously calculated gain G12 by the predetermined fluctuation upper limit ΔG so as to approach the currently calculated gain G13 as the current gain. The gain limiting unit 14 controls the buffer unit 11 to reduce the size of the storage area used for storing the voice input signal. The gain limiter 14 reduces the size of the voice input signal that can be stored by reducing the size of the storage area. The time length is shortened. That is, the time length D2 of the audio input signal that can be stored in the buffer unit 11 after the period T14 is shorter than the time length D1 from the period T11 to T13. Even at the end of the periods T14 and T15, since the absolute value of the change between the gain calculated by the gain calculating unit 13 and the previous gain is equal to or greater than the first threshold, the gain limiting unit 14 A limit value obtained by adding the variation upper limit ΔG to the gain is set as the current gain. At the end of the period T16, since the absolute value of the change between the gain calculated this time by the gain calculation unit 13 and the previous gain is less than the first threshold value, the gain calculation unit 13 uses the value calculated this time as the gain G16. And substantially matches the gain G13 calculated at the end of the period T13.

  Thus, since the time length of the audio input signal stored in the buffer unit 11 is shortened after the period T14, the gain is updated in a short time. Therefore, even when the amount of change in gain is limited within the above-described variation upper limit ΔG, the gain is calculated by the gain calculation unit 13 before the gain is limited in a shorter time before the gain change is limited by the gain limiting unit 14. The calculated gain G13) can be reached. After the buffer unit 11 reduces the size of the storage area, if the absolute value of the change between the gain calculated by the gain calculation unit 13 and the previous gain is less than the first threshold, the buffer unit 11 The size of the storage area used for storing the input signal may be returned to the original size.

  In this embodiment, after all the audio input signals stored in the buffer unit 11 are replaced, the gain calculation unit 13 obtains the maximum value of the absolute values of the audio input signals stored in the buffer unit 11. The maximum value is the representative value. The gain calculation unit 13 sets the gain so that a value obtained by multiplying the representative value (maximum value) of the audio input signal by the gain becomes a predetermined target value. Thereby, the maximum value of the signal output from the sound volume correction apparatus 10 can be made constant.

  If the average level of the signal output from the volume correction device 10 is to be constant, the average value of the audio input signal stored in the buffer unit 11 is used as a representative value, and a gain is added to this representative value (average value). The gain may be set so that a value obtained by multiplication becomes a predetermined target value. FIG. 5 shows an example of the waveform of the audio input signal (absolute value). The time lengths of the periods T1, T2, T3, and T4 correspond to the time length of the audio input signal stored in the buffer unit 11. Each time the audio input signals stored in the buffer unit 11 are all switched (that is, at the end of the periods T1, T2, T3, and T4), the gain calculating unit 13 performs the audio input signal in each of the periods T1, T2, T3, and T4. Average values A1, A2, A3, and A4 are obtained. Then, the gain calculation unit 13 uses the average value obtained in each period as a representative value of the audio input signal in each period, and calculates the gain so that a value obtained by multiplying the representative value by the gain becomes a predetermined target value. As a result, the average value of the signals output from the sound volume correction device 10 can be set as a predetermined target value, and the average value of the sound volume level can be made constant.

  Incidentally, as shown in FIGS. 6A and 6B, when the volume level of the audio input signal suddenly increases in the period T4, the gain G4 calculated by the gain calculating unit 13 is larger than the previously calculated gain G3. Decrease significantly. Here, when the absolute value of the change between the gain G3 and the gain G4 exceeds the first threshold value, the gain limiting unit 14 sets the previously calculated gain G3 to a variation upper limit in a direction approaching the currently calculated gain G4. A value G4b changed by the amount ΔG is set as the current gain. In this case, since the gain in the period T4 is not sufficiently reduced, the product of the representative value of the audio input signal and the gain G4b in the period T4 may exceed a predetermined second threshold value. When the limit value is the current gain and the product of the gain and the representative value exceeds a predetermined second threshold value, the gain limiting unit 14 causes the value obtained by multiplying the representative value by the gain to be the predetermined target value. The gain calculated by the gain calculation unit 13 is defined as the current gain. As a result, when the signal level of the sound input to the microphone 2 suddenly increases, it is possible to prevent an excessively loud signal from being output by changing (decreasing) the gain exceeding the first threshold. it can. The second threshold value may be set to a value larger than the target value and corresponding to the upper limit level of the output volume.

  In the present embodiment, if the representative value of the audio input signal stored in the buffer unit 11 is equal to or lower than the determination level for determining whether the sound is in the silent state or the voiced state, the gain calculating unit 13 does not calculate the gain. The gain multiplying unit 15 preferably uses the gain previously calculated by the gain calculating unit 13.

  FIG. 7B shows an example of the waveform of the audio input signal, and FIG. 7A shows the gain determined based on the audio input signal. In the illustrated example, since the representative value (for example, the maximum value) of the audio input signal stored in the buffer unit 11 exceeds the predetermined determination level L1 in the periods T1, T2, and T3, the gain calculation unit 13 The gain is calculated based on the representative value of the voice input signal at. On the other hand, since the representative value (maximum value) of the audio input signal stored in the buffer unit 11 is equal to or lower than the determination level L1 in the period T4, the gain calculation unit 13 does not calculate the gain. Then, the gain multiplication unit 15 multiplies the voice input signal read from the buffer unit 11 by the gain G3 using the gain G3 previously calculated by the gain calculation unit 13 in the period T4, and thereby encodes and packetizes the processing unit 4. Output to.

  When the representative value of the audio input signal is equal to or lower than the determination level L1, that is, when the gain calculation unit 13 calculates the gain in a silent state in which no meaningful signal is input, the gain is set to an excessive value. As described above, if the representative value is equal to or less than the determination level L1, the gain calculation unit 13 does not calculate the gain, but multiplies the voice input signal by the previously calculated gain, so the floor noise is amplified by an excessive gain. The possibility of being reduced can be reduced. The determination level L1 is preferably set to a value larger than the volume level of the floor noise and smaller than the lower limit of the representative value of the meaningful signal (that is, the audio signal). It can be reliably distinguished from floor noise.

  In the present embodiment, it is also preferable to include an audio determination unit (not shown) that determines whether the audio input signal input to the buffer unit 11 is audio or non-audio. In this case, when all the audio input signals input to the buffer unit 11 are determined as non-speech by the audio determination unit, the gain calculation unit 13 does not calculate the gain, and the gain multiplication unit 15 does not calculate the gain. Is multiplied by the voice input signal using the previously calculated gain.

  For example, in the case of the audio input signal shown in FIG. 8B, at least a part of the audio input signal stored in the buffer unit 11 is determined to be audio during the periods T1, T2, and T3. Therefore, the gain calculation unit 13 calculates the gain in the periods T1, T2, and T3, and the gains in the periods T1, T2, and T3 are G1, G2, and G3, respectively (see FIG. 8A). On the other hand, in the period T4, all of the audio input signals stored in the buffer unit 11 are determined as non-speech by the audio determination unit, and in this case, the gain calculation unit 13 does not calculate the gain. Then, the gain multiplication unit 15 multiplies the voice input signal read from the buffer unit 11 by the gain G3 using the gain G3 previously calculated by the gain calculation unit 13 in the period T4, and thereby encodes and packetizes the processing unit 4. Output to. Here, when the gain is obtained from the non-speech signal in the period T4, the gain G21 is significantly larger than the gain obtained from the speech signal (for example, the gain G3 in the period T3), and this gain G21 is used. Multiplying the audio input signal will amplify the noise with high gain. On the other hand, in the present embodiment, since the calculation of the gain is not performed in the period T4 in which all the audio input signals are determined to be non-audio, and the audio input signal is amplified using the gain G3 obtained last time, Noise can be suppressed from being amplified with high gain.

  As described above, when all the audio input signals input to the buffer unit 11 are determined to be non-speech by the audio determination unit, the gain multiplication unit 15 calculates the gain previously calculated by the gain calculation unit 13 as the buffer unit 11. Is multiplied by the audio input signal read from. Therefore, it is possible to suppress noise other than voice from being amplified with a high gain.

  The sound determination unit determines whether the sound is sound or non-speech by the following method. That is, the voice determination unit obtains a long-time average value of a voice input signal input from the A / D conversion unit 3 for a relatively long time and a short-time average value of a voice input signal for a relatively short time. Here, the level of the short-time average value is determined by the voice component included in the voice input signal, and the level of the long-time average value is determined by the noise component (sound component other than voice) included in the voice input signal. It is thought that it is done. Thus, the sound determination unit determines that the sound is a sound if the ratio of the short-time average value to the long-time average value is equal to or greater than a predetermined reference value, and determines that the sound is non-speech if it is less than the reference value. It is possible to reliably distinguish the voice. Note that the method for determining voice or non-speech is not limited to the above-described method. For example, it may be determined whether the voice is input or not from the input of a switch operated by the speaker at the time of voice input.

DESCRIPTION OF SYMBOLS 10 Volume correction apparatus 11 Buffer part 12 Gain setting part 13 Gain calculation part 14 Gain limit part 15 Gain multiplication part

Claims (9)

A storage area capable of storing a voice input signal for a predetermined time, and when there is no space in the storage area, a buffer section for overwriting and saving a new voice input signal in a storage area for storing the oldest voice input signal;
A gain calculating unit that calculates a gain based on a representative value of the audio input signal stored in the buffer unit every time the audio input signal stored in the buffer unit is replaced;
The gain calculation unit obtains a change between the previously calculated gain and the currently calculated gain, and when the absolute value of the change is equal to or greater than a predetermined first threshold, the previously calculated gain approaches the currently calculated gain. A gain limiter that uses the limit value obtained by changing the amount of change as much as a predetermined fluctuation upper limit as the current gain,
A gain multiplier that reads out the audio input signals from the buffer unit in the oldest order when the gain is set by the gain calculator and the gain limiter, multiplies the read audio input signals by the gain, and outputs the result The volume correction apparatus characterized by the above-mentioned.   The volume correction apparatus according to claim 1, wherein the gain calculation unit obtains a maximum absolute value of the audio input signal stored in the buffer unit and sets the maximum value as the representative value.   The volume correction apparatus according to claim 1, wherein the gain calculation unit calculates an average value of absolute values of audio input signals stored in the buffer unit, and uses the average value as the representative value.   The volume according to any one of claims 1 to 3, wherein the gain calculation unit sets the gain so that a value obtained by multiplying the representative value by a gain is equal to a preset target value. Correction device.   When the gain limiting unit sets the limiting value as the current gain and the product of the gain and the representative value exceeds a predetermined second threshold, the gain limiting unit multiplies the representative value by the gain. 5. The volume correction apparatus according to claim 1, wherein the gain calculated by the gain calculation unit is set to be a current gain so that the value becomes a predetermined target value. 6.   The buffer unit reduces the size of the storage area when the absolute value of the change is equal to or greater than the first threshold and the gain limiting unit sets the limiting value as the current gain. Item 6. The sound volume correction device according to any one of Items 1 to 5. If the representative value is below the determination level for determining whether the sound is silent or sound,
7. The gain multiplication unit according to claim 1, wherein the gain multiplication unit multiplies the audio input signal read from the buffer unit by the gain previously calculated by the gain calculation unit and outputs the result. Volume correction device. A voice determination unit that determines whether a voice input signal input to the buffer unit is voice or non-voice;
When it is determined that all audio input signals input to the buffer unit by the audio determination unit are non-audio,
7. The gain multiplication unit according to claim 1, wherein the gain multiplication unit multiplies the audio input signal read from the buffer unit by the gain previously calculated by the gain calculation unit and outputs the result. Volume correction device.   The volume correction apparatus according to claim 1, wherein the gain calculation unit sets a gain to the upper limit value when the calculated gain exceeds a predetermined upper limit value.

Images