JP2009122204A - Sound volume control unit, method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique to control sound volume without changing the predetermined feature amounts of the sound such as frequencies as much as possible. <P>SOLUTION: The inputted sound is divided into frames of a certain time length. The outer shape value which is the feature amount representing the sound volume in a frame is calculated for each frame. By defining the sound section consisting of the preset number B<SB>2</SB>of the frames or more between the silent frames each continuing for the preset number B<SB>1</SB>or more as a first sound section, and removing some of the outer shape values of the frames constituting the first sound section starting from the larger one, and the maximum remaining one is decided to be the outer shape value of the first sound section. The information for adjusting the inputted sound volume (the first volume control information, hereinafter) is decided so that the obtained outer shape value of the first sound section is within the preset range, and the sound volume is adjusted by using a first volume control instruction means to output and the outputted first volume control information. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a volume adjusting device, method, and program for automatically adjusting the volume of sound input from a sound input device such as a telephone or a microphone for voice recognition and the like.

A conventional automatic volume control apparatus will be described with reference to FIG. The conventional automatic volume control device is intended to adjust a sound such as music and voice to a volume that is easy for a person to hear and hear.
The dashed-dotted line in FIG. 7 represents the input peak (allowable maximum value) of the device to which the sound output from the automatic volume control device (that is, the sound adjusted automatically) is input.
(1) If the volume of the input sound exceeds the peak of the input, the gain is reduced rapidly until the volume does not exceed the peak of the input.
(2) The average power of the input sound is calculated for a short time (10 to 30 seconds), and the gain is gradually increased and decreased so that the average power approaches the preset target value of the average power. At this time, the short-term power (power per frame) of the sound is calculated to distinguish between the sounded section (sounded section) and the silent section, and the average power is calculated using only the sound of the sounded section. Thus, the volume can be adjusted appropriately.

Conventional automatic volume control devices combine the methods (1) and (2) above to perform automatic volume control, and control so that the input loud and small sounds are always at a constant volume that is easy for humans to hear. (For example, refer to Patent Document 1).
JP 58-14410 A

Conventional automatic volume control devices constantly adjust the volume little by little by increasing and decreasing the gain. As described above, when the input sound is constantly adjusted, there is a problem that a predetermined feature amount of the sound whose waveform is distorted or the like is easily lost.
A specific example of the volume adjusting method (2) will be described with reference to FIG. For example, as shown in FIG. 8, the average value P1 (average power P1) of the power L1 of the sound with small amplitude dispersion is brought close to the target value P ^* of the average power. The power of a sound with small amplitude dispersion after being adjusted to be close is indicated by a symbol L2, and the average value thereof is indicated by P2 (average power P2). At this time, when a sound L3 having a large amplitude dispersion (the vertical axis of L3 represents the amplitude) is input, a portion exceeding the input peak intermittently occurs and the portion exceeds the input peak. There is a problem that a predetermined feature amount of sound changes.

  The present invention provides a volume adjusting device, method, and program for adjusting a volume of a target sound to be equal to or lower than a predetermined volume (peak of input sound) while maintaining a predetermined feature amount of the target sound. Objective.

The volume control device according to the present invention includes a frame dividing unit that divides an input sound into frames of a certain length of time, and a frame outer shape that obtains an outer shape value that is a feature amount representing the magnitude of sound included in the frame for each frame. configuration value determining means, sandwiched silence consecutive frames number B ₁ than the predetermined, the sound interval which is composed of several B ₂ or more frames which is predetermined as a first sound zone, the first sound segment The first sound interval that obtains the maximum value of the remaining outer shape value as the outer shape value of the first sound interval by excluding the plurality of outer shape values from the larger outer shape value from the outer shape values of the plurality of frames Information for adjusting the volume of the input sound (hereinafter referred to as first volume adjustment information) so that the outer shape value determining means and the determined outer shape value of the first sound section fall within a predetermined range. .) Comprising a first sound volume adjustment instruction means for force, and a first volume control means for adjusting the volume of the input sound using a first volume adjustment information the output.

  While maintaining a predetermined feature amount of sound, the volume of the sound can be reduced to a predetermined volume (peak of input sound) or less.

With reference to FIG. 1, an automatic volume control apparatus 1 according to an embodiment of the present invention will be described.
Sound is input from the input unit 11. The input unit 11 is a microphone, for example. As the input unit 11, a transmission / reception adapter installed between a telephone and a transmitter / receiver or a headset may be used in order to extract one or both of the transmitted voice and the received voice of the telephone. The sound input from the input unit 11 is converted into an electrical signal and output to the volume adjustment unit 12.
The volume adjustment unit 12 adjusts the volume of the input sound based on volume adjustment information described later, and outputs the adjusted sound. A part of the output sound is input to the AD converter. The volume adjustment unit 12 may be analog or digital. Details of the processing of the volume adjusting unit 12 will be described later.

The AD conversion unit 13 digitizes the sound analog signal by quantizing the analog signal at a predetermined sampling frequency, and sends the digitized signal to the frame division unit 14. Note that an AD conversion unit 13 may be provided in front of the volume adjustment unit 12 as indicated by a dotted line in FIG. In this case, the volume adjustment unit 12 is digital. Hereinafter, the digitized sound signal is referred to as a sound signal.
The frame dividing unit 14 divides the input sound into frames having a certain time length. For example, the length of one frame is set to 100 ms (when the sampling frequency is 16 kHz, the number of samples constituting the frame is 1600). In this way, by making the time length of the frame sufficiently longer than, for example, the male speech waveform and the basic period of the power supply noise, the volume can be adjusted stably regardless of the voice level and the power supply noise. The framed sound signal is sent to the buffer 15.

The buffer 15 temporarily stores a number A ₁ of frames equal to or greater than a predetermined number 1.
The DC bias calculator 16 reads the framed sound signal stored in the buffer 15 and calculates the average value of the amplitude of the sound signal by observing it for a long time. The average value, that is, the value of the DC component is sent to the subtracting unit 17.
The subtractor 17 subtracts the value of the DC component calculated by the DC bias calculator 16 from the sound signal read from the buffer 15 to generate an unbiased sound signal. The generated sound signal is sent to the end-to-end determination unit 18, the external shape determination unit 19, and the second volume adjustment instruction unit 26. Hereinafter, the term “sound signal” refers to an unbiased sound signal.

The end-to-end determination unit 18 determines the start time of sound generation and the end time of sound generation by observing the average value of the absolute values of sound signals for each frame. The sound section at the beginning and end of pronunciation is defined as pronunciation. The start and end of pronunciation are the beginning and end of a call when the sound is a voice such as a phone call. In this case, the pronunciation corresponds to a so-called call section.
Specifically, the average value calculation unit 181 of the end-to-end determination unit 18 calculates the average value of the absolute values of the amplitudes of the input sound signals for each frame. Then, throughout the determination unit 18, to determine whether the average value of the calculated absolute value of the amplitude is greater than the threshold value A ₂ with a predetermined sequence, if it is determined to be larger sound is started and Determination is made and a signal to that effect is sent to each part of the automatic volume control device 1 including the end-time volume control unit 33. If the average value of the calculated absolute value of the amplitude is determined to be larger than the threshold value A ₂ of predetermined phonetic since predated predetermined time length from the time when the it is determined (e.g., 0.5 seconds) You may determine with having started.

Further, the end-to-end determination unit 18 has a state in which the average value of the calculated absolute value of the amplitude is smaller than a predetermined threshold A ₃ (the threshold A ₃ is smaller than the threshold A ₂ ). If lasted a certain time length defined in or, when followed by pre-determined number a ₄ frames, it is determined that the sound is completed, the end volume adjusting unit 33 a signal to that effect It is sent to each part of the automatic volume control apparatus 1 including it.
The outline value determination unit 19 that has received a signal indicating that the sound generation has been started obtains an outline value that is a feature amount representing the loudness of the frame for each frame. For example, the outer shape value is the maximum absolute value of the amplitude of the sound signal. In other words, the outer shape value is the maximum value of the values of a plurality of samples constituting the frame. The obtained outer shape value for each frame is sent to the sound / silence frame determination unit 20 and the first sound volume adjustment instruction unit 25. 3A and 3B show specific examples of external value extraction. FIG. 3A shows a waveform of an unbiased sound signal. FIG. 3B illustrates the maximum value (outer shape value) of the absolute value of the amplitude obtained for each frame from the waveform of the sound signal shown in A. FIG.

The description will be given again with reference to FIG. Voice activity frame determination unit 20 compares the threshold value A ₅ with a predetermined outer shape value, the larger the better contour value determines the frame as voiced frame, otherwise the frame Judged as a silent frame. The threshold A _5, without the predetermined value, for example, may be dynamically changed the threshold A ₅ as the value of the constant multiple of the minimum value of the outline values of the silent frame of the past 10 seconds (e.g., 3 times) . Information on whether the frame is a sound frame or a sound frame is sent to the sound / silence section determination unit 21.
The voiced / silent section determination unit 21 sets the silent frame for a predetermined number A ₆ (for example, A ₆ is set so that the time length is 0.5 seconds). The sound section composed of the continuous frames is determined as a silent section, and the sound section composed of other frames is determined as a sound section. Information about the voiced section and the silent section is sent to the first sound section extraction unit 22 of the first volume adjustment instruction section 25.

Hereinafter, the first volume adjustment instruction unit 25 will be described with reference to FIG. The first sound segment extraction unit 22 of the first volume adjustment instruction unit 25 determines whether or not the determined sound segment is longer than a predetermined time length A ₇ (for example, 2 seconds) or the above determination. When the number of frames A ₈ (for example, 20 frames) constituting the sound section is larger than the predetermined number A ₈ , the sound section is set as the first sound section. When the input sound is a voice such as a telephone, the first sound section corresponds to a so-called speech section. The utterance section is a section of a sound that a person utters with one breath. In this way, by extracting the first sound section, it is possible to cut out the "Hello" or "little question is you, but" such as the sense to close the length of a person's sound section. FIG. 3B shows a specific example of extraction of the first sound section. For example, as shown in FIG. 3B, a lump of sounded sections of 2 seconds or longer is extracted as a first sound section using a silent section of 0.5 seconds or longer.

The first sound segment extraction unit 22 sends, for example, information regarding the frames constituting the first sound segment and the outer shape values of those frames to the first sound segment outer value extraction unit 23. As the outer shape value of the frame constituting the first sound section, information on the outer shape value of the frame received by the first sound section extracting unit 22 from the outer shape value determining unit 19 is used.
The exclusion unit 231 of the first sound section outer shape value extraction unit 23 excludes a plurality of outer shape values from the larger outer shape value from the outer shape values of a plurality of frames constituting the first sound section. The number of external values to be excluded is preferably increased as the number of frames constituting the first sound section is larger. For example, multiply the number of frames that make up the first sound section by a preset ratio A ₉ (for example, 10 to 30%, this time 20%), and exclude the number of external values that are rounded down, rounded, or rounded up. To do. It may be to exclude a predetermined outer shape of the number A _10. The outline value remaining without being excluded is sent to the maximum value determination unit 232.

The maximum value determination unit 232 obtains the maximum value of the external shape value that remains without being excluded, and stores the maximum value as the external value of the first sound section. The external value of the first sound section is sent to the first gain determination unit 24.
The first gain determination unit 24 determines information (hereinafter, referred to as first volume adjustment information) for adjusting the input sound so that the outer shape value of the first sound section falls within a predetermined range. Then, it is sent to the volume adjustment unit 12. For example, an input peak is input to the first gain determination unit 24. The first gain determination unit 24 determines the gain so that the outer shape value of the first sound section is within a range obtained by multiplying the input peak by a predetermined ratio A ₁₁ (for example, 10% to 25%). . In this case, the gain is the first volume adjustment information.

When the first sound volume adjustment information is determined, the first sound volume adjustment instruction unit 25 does not perform the above process for a frame corresponding to the delay time of the buffer 15.
A specific example will be described with reference to FIG. 3C. The excluding unit 231 excludes a predetermined number (seven in this example) of outer shape values having large outer shape values from the outer shape values of the frames constituting the first sound section. It is an outer shape value from which the outer shape value shown in white in FIG. 3C is excluded. The maximum value determination unit 232 selects the largest outline value among the outline values remaining without being excluded as the outline value of the first sound section. The outer shape value remaining without being excluded is the outer shape value indicated by black and rays in FIG. 3C, and the outer shape value of the first sound section, which is the maximum value, is the outer shape value indicated by rays.

Assuming that a predetermined range in which the outer shape value of the first sound section is to be entered is 3000 to 8000, in this example, the outer shape value of the first sound section is not in that range. The first gain determination unit 24 calculates the difference between the outer shape value of the first sound section and the range thereof, and determines the gain so that the outer shape value of the first sound section falls within the range. If the external value of the first sound section is within that range, no processing is performed.
Another specific example will be described. It is assumed that the outer shape value of the first sound section is 5% of the input peak, and the predetermined range in which the outer shape value of the first sound section is to enter is 10% to 25% of the input peak. In this case, the first gain determination unit 24 determines the gain so that the outer shape value of the first sound section is 10% of the input peak. Thus, the external value of the first sound section after volume adjustment becomes equal to the value closer to the external value of the first sound section before volume adjustment, among the upper limit value or lower limit value of the predetermined range. Thus, by determining the gain, the volume adjustment amount can be minimized, and the change in the predetermined feature amount of the sound can be minimized.

In addition, in this way, when a predetermined range in which the outer shape value of the first sound section is to be entered is provided and the outer shape value of the first sound section is included in this range, the above gain calculation is not performed. By doing so, the number of times of changing the gain can be reduced. As a result, the number of times the waveform of the sound is distorted can be reduced, so that the change in the predetermined feature amount of the sound can be reduced.
In this method, “Yes”, “Ah”, “Eh”, etc. are not short sound intervals where the volume is unstable, but “Thank you for calling.” “I have something I want to hear” A sound section having a certain length and a stable volume is used as a reference for volume adjustment. In addition, by excluding a plurality of contour values having a large contour value from the contour values of a plurality of frames constituting the first sound interval, the maximum value of the remaining contour values not excluded is used as the contour value of the first sound interval. The gain is adjusted using the external value of the first sound section.

This makes it less susceptible to sudden noise such as coughing and sneezing, and the volume after adjusting the volume does not exceed the input peak even if the amplitude of the target sound varies.
In the above example, out of the outer shape values of the frames constituting the first sound section, the outer shape value of 20% from the larger one is excluded, and a predetermined range in which the outer shape value of the first sound section should be entered is the input peak. 10% to 20%. This is because, as a result of the experiment, the peak of the input was less than about 4 times the external value of the first sound section, excluding sudden noise.

The description will be given again with reference to FIG. The first volume adjustment unit 121 of the volume adjustment unit 12 uses the first volume adjustment information (for example, gain) determined by the first volume adjustment instruction unit 25 to adjust and output the volume of the input sound. The first volume adjustment unit 121 performs volume adjustment based on the already sent first volume adjustment information until new first volume adjustment information is sent from the first volume adjustment instruction unit 25.
Thus, in the present invention, the volume is adjusted based on the same first volume adjustment information for a long time compared with the volume adjustment device of the prior art. This makes it difficult to lose the predetermined feature amount of the sound as compared with the case where the gain for frequently adjusting the sound volume is changed as in the prior art.

You may have the 2nd sound volume adjustment instruction | indication part 26 and the 2nd sound volume adjustment part 122 which adjust a sound volume on the basis of the sound area (2nd sound area) shorter than a 1st sound area described below.
The second volume adjustment instruction unit 26 will be described with reference to FIG. The sound signal output from the subtraction unit 17 is input to the excessive input sample number determination unit 27 of the second volume adjustment instruction unit 26. The excessive input sample number determination unit 27 determines the number of samples larger than a predetermined value A ₁₂ (for example, 90% of the upper limit of values that can be expressed by sample values) (hereinafter referred to as excessive input sample number). .) Is determined for each frame. The determined number of excessive input samples for each frame is sent to the excessive input frame determination unit 28 and the storage unit 29.

The excessive input frame determination unit 28 determines for each frame whether or not the excessive input sample number is larger than a predetermined number A ₁₃ (30% of the number of samples in one frame). Hereinafter, a larger frame than the number A ₁₃ excessive number of input samples has been determined in advance and excessive input frame. Information on the excessive input frame (for example, a flag indicating that it is an excessive input frame) is sent to the storage unit 29.
The second sound section excessive input sample number determination unit 30 selects a sound section composed of frames of a number A ₁₄ (for example, 10 for a time length of 1 second) smaller than the number of frames constituting the first sound section. As a two-tone section, the total number of excessive input samples for the frames constituting the second sound section is calculated, and the total number is sent to the second gain determining unit 32. Specifically, when the second sound interval is the past 10 frames, the number of excessive input samples of the past 10 frames is read from the storage unit 29 and added to obtain the total number of excessive input samples. Ask for.

The second sound interval excessive input frame number determination unit 31 determines the number of excessive input frames in the frames constituting the second sound interval, and sends the number to the second gain determination unit 32. Specifically, when the second sound interval is the past 10 frames, information on the excessive input frames of the past 10 frames is read from the storage unit 29, and the number of excessive input frames is determined.
The second gain determination unit 32 is configured such that the total number of excessive input samples is larger than a predetermined number A ₁₅ (for example, 20% of the total number of samples constituting the second sound interval), and When the number is larger than a predetermined value A ₁₆ (for example, 3 when the second sound interval is 10 frames), information for lowering the volume of the input sound by a predetermined volume (hereinafter referred to as “the volume”) , Second volume adjustment information) is sent to the volume adjustment unit 12. The second volume adjustment information may be a specific gain value (for example, 0.7, 3 dB as a volume), or the like, or information indicating that the volume is simply decreased without a specific numerical value. May be.

The second volume adjustment unit 122 of the volume adjustment unit 12 decreases the volume of the input sound based on the second volume adjustment information. When the gain is lowered, the second volume adjustment instruction unit 26 sets a short-time volume adjustment flag for the frame, and thereafter does not perform processing on the frame corresponding to the time corresponding to the delay of the buffer 15.
As a result, it is possible to reduce the volume by avoiding noise having a relatively short duration from among sudden noises and reducing the gain.
After the start / stop determination unit 18 detects the start of sound generation, the volume is adjusted according to the instructions of the first volume adjustment instruction unit 25 and the second volume adjustment instruction unit 26 as described above. When the end-to-end determination unit 18 detects the end of the sound generation, information indicating that the sound generation has ended is sent to the end-time sound volume adjustment unit 33.

When receiving the information that the sound generation has ended, the end-time volume adjustment unit 33 reads the gain at the end of the sound generation set in the sound volume adjustment unit 12 and stores it in the storage unit 331 of the end-time volume adjustment unit 33. . Then, at the end of the volume adjusting unit 33 reads the respective gains at the end of the past pronounce number A ₁₇ predetermined by the last sound from the storage unit 331, obtains the average value thereof, the average value Set in the volume adjustment unit 12.
When the current gain value cannot be obtained from the volume adjustment unit 12, the end-time volume adjustment unit 33 sets the gain in the volume adjustment unit 12 as follows. The case where the current gain value cannot be obtained from the volume adjustment unit 12 means that, for example, the volume adjustment unit 12 has only a relative gain specifying means for increasing the 3 dB volume and decreasing the 3 dB volume. This is a case where the adjustment range is exceeded or there is no means for notifying that the adjustment has failed.

1. When the gain is not changed to adjust the volume according to the instruction of the first volume adjustment instruction unit 25, the end-time volume adjustment unit 33 does nothing.
2. When lowering the gain to decrease the volume according to an instruction of the first sound volume adjustment instruction unit 25, volume control gain at the end volume adjustment portion 33 having a reduced value by a value A ₁₈ set in advance from the current gain Set in part 12.
3. When the gain is increased in order to increase the volume in accordance with an instruction from the first volume adjustment instruction unit 25, the end-time volume adjustment unit 33 performs the following processing.

3-1. When the gain is lowered to lower the volume according to the instruction from the second volume adjustment instruction unit 26, the end-time volume adjustment unit 33 does nothing.
3-2. In cases other than “3-1.”, The end-time volume adjustment unit 33 sets a gain obtained by increasing the value by a preset value A ₁₉ from the current gain in the volume adjustment unit 12.
By adjusting the volume at the end of pronunciation in this way, the volume at the beginning of the next pronunciation can be brought close to an appropriate value, and the sound collection environmental conditions such as speaker, microphone position, and volume can be changed. The volume can be adjusted appropriately following the above.

As illustrated in FIG. 9, the sound input from the input unit 11 is input to the AD conversion unit 13 and the volume adjustment unit 12, and the volume adjustment information is similar to the above from the sound input to the AD conversion unit 13. Is determined, and the volume adjusting unit 12 may adjust the volume based on the determined volume adjustment information.
For example, when an operator changes seats every day at a call center, the same sound collection condition continues for a certain period of time, but the sound collection condition changes every day. In such an environment, the volume can be adjusted to meet the sound collection conditions such as the voice volume and microphone position of each operator in a short time of several calls. The volume can be adjusted appropriately.

The second volume adjustment instruction unit 26 and the second volume adjustment unit 122 may be omitted. Further, the end-time volume adjustment unit 33 may not be provided.
FIG. 5 shows a system for recording a conversation between an operator and a user using the automatic volume control device 1 in a call center.
An operator wears a headset 35 connected to the telephone 34 and has a conversation with the user. A handset branch adapter 36 having a volume control unit 12 is connected between the headset 35 and the telephone 34, and the audio is input to the PC 37 using audio input or USB. The voices of the operator and the user captured in the PC are transmitted through the echo canceling unit, and the operator voice that is in the user voice side is suppressed as a side sound. As shown in FIG. 6, when the echo canceling unit 38 is attached to the handset separation adapter, the processing of the echo canceling unit is bypassed.

  When the end-of-call determination unit 18 detects the beginning of a call based on the respective voices sent from the echo cancel unit 38, the transmission-side automatic volume adjustment device 1a determines the volume of the operator voice as described above. Adjust in the same way. The reception-side automatic volume adjustment device 1b adjusts the volume of the user voice in the same manner as the automatic volume adjustment device 1 described above. The transmission-side automatic volume adjustment device 1a and the reception-side automatic volume adjustment device 1b do not have the volume adjustment unit 12 and the end-to-end determination unit 18, but the end-to-end determination of the volume adjustment unit 12 of the handset branch adapter 36 and the PC 37. The unit 18 functions as the volume adjustment unit 12 and the end-to-end determination unit 18 of the transmission-side automatic volume adjustment device 1a and the reception-side automatic volume adjustment device 1b. The other points are the same as those of the automatic volume control device 1.

The operator voice can be adjusted to an appropriate volume with a few calls because the sound collection conditions are substantially the same while the operator is the same. However, the user voice has a different telephone, transmission line, etc. for each call. For this reason, the reception-side automatic volume adjustment device 1b does not give an instruction for tone adjustment by the end-time volume adjustment unit 33.
When the end-to-end determination unit 18 detects the end of the call, the sound whose volume is adjusted is stored in the disk 40 of the PC 37 through the recording unit 39.
When the above configuration is realized by a computer, the processing contents of the functions that each device should have are described by a program. The processing functions are realized on the computer by executing the program on the computer.

The program describing the processing contents can be recorded on a computer-readable recording medium. The computer-readable recording medium may be any medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, or a semiconductor memory. Specifically, for example, the magnetic recording device may be a hard disk device or a flexible Discs, magnetic tapes, etc. as optical discs, DVD (Digital Versatile Disc), DVD-RAM (Random Access Memory), CD-ROM (Compact Disc Read Only Memory), CD
-R (Recordable) / RW (ReWritable), etc., MO (Magneto-Optical disc), etc. as a magneto-optical recording medium, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. as a semiconductor memory it can.

The program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Furthermore, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.
As an execution form different from the above-described embodiment, the computer may read the program directly from the portable recording medium and execute processing according to the program. Each time is transferred, the processing according to the received program may be executed sequentially. Also, the program is not transferred from the server computer to the computer, and the above-described processing is executed by a so-called ASP (Application Service Provider) type service that realizes the processing function only by the execution instruction and result acquisition. It is good. Note that the program in this embodiment includes information that is used for processing by an electronic computer and that conforms to the program (data that is not a direct command to the computer but has a property that defines the processing of the computer).

In this embodiment, the present apparatus is configured by executing a predetermined program on a computer. However, at least a part of these processing contents may be realized by hardware.
In addition, the various processes described above are not only executed in time series according to the description, but may be executed in parallel or individually according to the processing capability of the apparatus that executes the processes or as necessary. Needless to say, other modifications are possible without departing from the spirit of the present invention.

The figure which illustrates the function structure of the automatic volume control apparatus 1 which is one Example of this invention. The figure which illustrates the function structure of the 1st volume adjustment instruction | indication part. A is a figure which illustrates the waveform of a sound signal. B is a diagram illustrating a first sound section (speech section). C is a diagram illustrating an outer shape value of a first sound section. The figure which illustrates the functional structure of the 2nd volume adjustment instruction | indication part. The figure which illustrates the system which records the conversation of an operator and a user. The figure which illustrates the system which records the conversation of an operator and a user. The figure which illustrates the conventional automatic volume adjustment. The figure for demonstrating the problem of the conventional automatic volume adjustment. The figure which illustrates the function structure of the automatic volume control apparatus 1 '.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Automatic volume adjustment apparatus 1a Transmission side automatic volume adjustment apparatus 1b Reception side automatic volume adjustment apparatus 11 Input part 12 Volume adjustment part 13 Conversion part 14 Frame division part 15 Buffer 16 DC bias calculation part 17 Subtraction part 18 End-of-life determination part 19 Outline value Determining unit 20 Sounded / silent frame determining unit 21 Sounded / silent segment determining unit 22 First sound segment extracting unit 23 First sound segment outer shape value extracting unit 24 First gain determining unit 25 First volume adjustment instruction unit 26 Second volume adjustment Instructing unit 27 excessive input sample number determining unit 28 excessive input frame determining unit 29 storage unit 30 second sound segment excessive input sample number determining unit 31 second sound segment excessive input frame number determining unit 32 second gain determining unit 33 volume at end Adjustment unit 34 Telephone 35 Headset 36 Handset branch adapter 38 Echo canceling unit 39 Recording unit 40 Disk 121 First Volume Adjustment Unit 122 Second Volume Adjustment Unit 181 Average Value Calculation Unit 231 Exclusion Unit 232 Maximum Value Determination Unit 331 Storage Unit

Claims (7)

Frame dividing means for dividing the input sound into frames of a certain time length;
Frame outer shape value determining means for obtaining an outer shape value, which is a feature amount representing a loudness included in a frame, for each frame;
A plurality of frames constituting the first sound interval are defined as a first sound interval between sound frames composed of frames equal to or more than a predetermined number B ₂ and sandwiched between a predetermined number B ₁ or more of continuous silence frames. A first sound section outer shape value determining unit that excludes a plurality of outer shape values from the outer shape value from the larger outer shape value and obtains a maximum value of the remaining outer shape values as the outer shape value of the first sound section without being excluded; ,
Information for adjusting the volume of the input sound (hereinafter, referred to as first volume adjustment information) is determined so that the obtained outer shape value of the first sound section falls within a predetermined range. First volume adjustment instruction means for outputting;
First volume adjustment means for adjusting the volume of the input sound using the output first volume adjustment information;
A volume control device comprising: The volume control device according to claim 1,
The outer shape value of the frame is the maximum absolute value of the sample values included in the frame.
A volume control device characterized by that. In the volume control apparatus according to claim 1 or 2,
If greater than the threshold value B ₃ of external value of a frame is predetermined to determine the frame and voiced frame, a silence frame and determines voice activity frame determination means that frame otherwise,
A voiced silence section determining means for determining a sound section composed of continuous silence frames of a predetermined number B ₁ or more as a silent section and determining other sound sections as a voiced section;
A first sound segment extraction means that sets a sound segment longer than a predetermined time length among the determined sound segments as the first sound segment;
A volume control device comprising: In the volume adjusting device according to any one of claims 1 to 3,
The number of the absolute value of a sample is greater than the threshold value B ₄ a predetermined sample value (hereinafter referred to. An excessive number of input samples) and excessive input sample number determining means for determining for each frame,
Excessive input frame determination means for determining whether or not the number of excessive input samples is larger than a predetermined number B ₅ (hereinafter, the number of excessive input samples is larger than a predetermined number B ₆ ); The frame is an excessive input frame),
A sound interval composed of a number of frames smaller than the number of frames constituting the first sound interval is defined as a second sound interval, and the determined number of excessive input samples for the frames constituting the second sound interval is determined. When the total number is larger than the predetermined number B ₇ and the number of excessive input frames in the frames constituting the second sound section is larger than the predetermined number B ₈ , the input Second volume adjustment instruction means for outputting information (hereinafter referred to as second volume adjustment information) for lowering the volume of the generated sound by a predetermined volume;
Second volume adjustment means for lowering the volume of the input sound using the output second volume adjustment information;
A volume control device comprising: The volume control device according to any one of claims 1 to 4,
Means for calculating an average value of the absolute value of the amplitude of the input sound for each frame;
When a frame having an average value larger than a predetermined threshold B ₉ is detected, it is determined that sound generation has started, and a predetermined number B _{11 of} frames having an average value smaller than the predetermined threshold B ₁₀ is determined. Means for determining the end of sound generation only when it continues continuously,
When it is determined that the sound generation has ended, the first sound volume adjustment information and / or the second sound volume adjustment information at the time of sound generation end is stored in the storage means, and a predetermined number B ₁₂ of past is determined from the latest sound generation. The first volume adjustment information and / or the second volume adjustment information at the end of pronunciation is read from the storage means, the average value thereof is obtained, and the setting is set in the first volume adjustment means and / or the second volume adjustment means Hour volume adjustment means,
A volume control device comprising: A frame dividing step for dividing the input sound into frames of a certain length of time;
A frame outer shape value determining step for obtaining an outer shape value, which is a feature amount representing a loudness included in the frame, for each frame;
A plurality of frames constituting the first sound interval are defined as a first sound interval between sound frames composed of frames equal to or more than a predetermined number B ₂ and sandwiched between a predetermined number B ₁ or more of continuous silence frames. A first sound interval outer shape determination step for excluding a plurality of outer shape values from the outer shape value from the larger outer shape value and obtaining the maximum value of the remaining outer shape values as the outer shape value of the first sound interval without being excluded; ,
Information for adjusting the volume of the input sound (hereinafter, referred to as first volume adjustment information) is determined so that the obtained outer shape value of the first sound section falls within a predetermined range. A first volume adjustment instruction step to output;
A first volume adjustment step for adjusting the volume of the input sound using the output first volume adjustment information;
A volume adjustment method comprising:   A sound volume adjusting program for causing a computer to function as each means of the sound volume adjusting device according to claim 1.

Images