KR101410523B1 - Apparatus and method for controlling a buffer for processing voice data - Google Patents

Abstract

The present invention relates to a buffer control apparatus and method for processing voice data, comprising: measuring an initial read speed of a buffer; Calculating a difference between the initial reading speed and a predetermined output speed of the buffer; Determining a ratio of an amount of data input to or output from the buffer according to the calculated speed difference; Calculating a gap between the read pointer and the write pointer by measuring a position of a read pointer and a write pointer of the buffer for inputting or outputting data according to a ratio of the determined data amount; And adjusting a ratio of the amount of data according to whether the interval between the read pointer and the write pointer is changed.

Voice data, buffer, data compensation, interpolation, decimation, silence interval

Description

[0001] APPARATUS AND METHOD FOR CONTROLLING A BUFFER FOR PROCESSING VOICE DATA [0002]

The present invention relates to buffer control. And more particularly, to a buffer control apparatus and method for processing voice data.

In a voice digital system, voice data decoded in a vocoder is temporarily stored in a buffer and output to a voice reproduction device (for example, a sound card of a computer). That is, the vocoder outputs the decoded audio data to the buffer writer at fixed intervals (usually 20 msec), and the buffer writer records the audio data in the buffer. The buffer reader then reads the voice data recorded in the buffer and outputs it to the voice reproduction apparatus.

1 is a block diagram of a general buffer for storing voice data.

In Fig. 1, the state in which the voice data 105 is stored in the buffer 100 is shown. The buffer 100 includes a write pointer (WP) 101 and a read pointer (RP) 103. The WP 101 indicates the next address of the buffer address at which the current buffer writer (not shown) records the voice data, and the RP 103 stores the address of the buffer address currently read by the reader It points to the address in the next sequence. Hereinafter, an interval between the RP 103 and the WP 101 is defined as an 'RW value 107' below.

The rate at which the WP 101 moves the buffer depends on the rate at which the buffer writer writes the voice data (hereinafter, referred to as 'writing speed'), and the writing speed is dependent on the speed at which the vocoder outputs the voice data. Eventually, the write speed depends on the output speed of the vocoder. However, the output speed of the vocoder is fixed as described above. This is because the output speed of the vocoder is processed in software.

In addition, the speed at which the RP 103 moves the buffer depends on the speed at which the buffer reader reads the voice data (hereinafter, referred to as 'read speed'). However, the reading speed is set according to a predetermined sampling time frequency set in the sound card. In order for the audio data to be reproduced smoothly, the reading speed must be equal to the writing speed (i.e., the RW value 107 must be constant during audio data reproduction). If the writing speed is faster than the reading speed, And in the opposite case, a buffer underflow occurs. Therefore, it is ideal that the read speed should be set dependent on the write speed. For this purpose, the sampling frequency is generally set in accordance with the output speed of the vocoder. However, it is common for a sound card to have a set sampling frequency and error. This is due to the hardware cause, and the sampling frequency, which is normally set, and the error within the range of about 5% are treated as normal products. Therefore, there arises a problem that the difference between the read speed and the write speed occurs in the buffer.

A conventional technique for solving such a problem is disclosed in Korean Patent Publication No. 2007-0081766. To briefly summarize the contents of the above-mentioned patent, if the RW value is measured in a buffer where audio data is recorded and the RW value is out of the standard, the oscillation frequency of the VCXO is controlled in hardware and the RW value is adjusted according to the oscillation frequency Method. 2 is a view for explaining an audio data processing apparatus according to the prior art. 3 is a conceptual diagram for explaining a speech data processing method according to the prior art. Figures 1 and 3 of Japanese Patent Application No. 2007-0081766 are attached to Figures 2 and 3 for a description of the technology according to the prior art. A detailed description thereof can be found in Japanese Patent Application Laid-Open No. 2007-0081766.

However, the conventional apparatus and method cause 'tick, tick' sound whenever the oscillation frequency of the VCXO is controlled by hardware. This is a factor that can be sufficiently perceived by the user, which causes a deterioration in the quality of the audio data reproduction. In a case where audio data from another application is stored in a buffer by multitasking or the like while voice data is stored in a buffer by one application and is reproduced, the position of the RP jumps to the address where the other voice data is stored (jump) occurs. In this case, a reversal phenomenon occurs in which RP is located ahead of WP. This reverse phenomenon adversely affects quality in reproduction of audio data. That is, there may be a problem in voice reproduction due to a software problem.

In order to solve the above problems, the present invention provides a buffer control apparatus and method for processing voice data for preventing buffer overflow or buffer underflow.

The present invention also provides a buffer control apparatus and method for processing audio data for smoothly reproducing audio data by solving a reverse phenomenon.

In addition, the present invention provides a buffer control apparatus and method for processing voice data that can improve reproduction quality because a user does not recognize a process of resolving voice distortion.

The apparatus provided in the present invention,

A buffer control apparatus for processing voice data, comprising:

(RP) and a write pointer (WP) of the buffer in a real-time processing step, in real time, in real time, And outputs the measurement result,

A silence detection unit for detecting silence period information of the voice data and outputting the silence period information,

(RW) value using the RP and the WP during the real time processing step, and outputs the value of the reader / writer (RW) ,

Performing initial data compensation to adjust the ratio of the amount of voice data to be input to and output from the buffer using the speed difference value in the buffer initialization step, and changing the RW value in the real time processing step, And real-time data compensation for adjusting the ratio of the amount of voice data stored in the buffer.

In addition, the present invention provides a method,

A buffer control method for processing voice data,

Measuring an initial read rate of the buffer during a buffer initialization step;

Obtaining a speed difference value with a preset output speed using the initial reading speed;

An initial data compensation step of adjusting a ratio of data to be input to and output from the buffer in correspondence with the speed difference value;

Measuring a position of a read pointer and a write pointer of the buffer receiving audio data in real time;

Detecting a silent section of the voice data;

Calculating a reader-writer (RW) value using the positions of the pointers measured in real time;

And adjusting a ratio of the amount of voice data stored in the buffer when the RW value is changed and the current voice data is a silence period.

Effects of the embodiment of the present invention will be described as follows.

A buffer overflow or a buffer underflow that may occur due to a hardware error generated in the sound card connected to a buffer for processing voice data is prevented to minimize distortion in voice data reproduction.

In addition, the present invention solves the inversion phenomenon that can occur due to multitasking of a buffer for processing voice data, thereby facilitating voice data reproduction.

Further, the present invention can improve the reproduction quality in the audio data reproduction by processing the distortion of the reproduction of the audio data generated in the buffer for processing the audio data into a silent section in which the user can not recognize.

The operation principle of the preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear. The following terms are defined in consideration of the functions of the present invention, and may be changed according to the intentions or customs of the user, the operator, and the like. Therefore, the definition should be based on the contents throughout this specification.

4 is a flowchart illustrating a buffer control method for processing voice data according to an embodiment of the present invention.

In step 401, the state of the buffer is measured in the step of initializing the buffer to drive the application. That is, the initial reading speed of the buffer is measured, and the speed difference with the sampling rate preset in the sound card is obtained. Here, it is assumed that the set sampling rate is set equal to the write speed. Therefore, if you do not know the writing speed, you can measure the initial writing speed and the initial reading speed of the buffer to obtain the difference between the two. Since step 401 is a stage before voice data is input to the buffer, it can be performed by generating test data for measuring the output speed and inputting it into the buffer. If the voice data continues to be input / output to the buffer in the state where the speed difference has occurred, buffer overflow or underflow may occur. Therefore, the speed difference must be compensated to prevent this. This is done in step 403.

The method of compensating the speed difference is as follows. And adjusts the ratio of data input / output to / from the buffer. That is, since the write speed and the read speed are values already determined in the buffer, it may be difficult to change them. Therefore, instead of changing the writing speed and the reading speed, the ratio of the amount of data input / output to the buffer is adjusted. In other words, since the number of times that the buffer reader and the buffer writer can read or write per unit time can not be adjusted, the amount of data input to and output from the buffer by adjusting the amount of data that the buffer reader and the buffer writer can read or write at one time Is maintained constant.

Adjusting the amount of data is defined as 'data compensation'. In addition, since data compensation in the buffer initialization step is performed, the data compensation in step 403 is referred to as 'initial data compensation'. There are two types of initial data compensation. The first is interpolation and the second is decimation. The interpolation means changing the data input / output ratio so that the amount of data stored in the buffer increases. The decimation means changing the input / output ratio so that the amount of data stored in the buffer decreases. A buffer overflow or a buffer underflow that may occur when voice data is input using the speed difference, and adjusts a decimation or an interpolation ratio to prevent the buffer overflow or the buffer underflow.

Then, in step 405, the voice data is input to the buffer. In step 407, the RW value is measured while the voice data is input / output to the buffer. As described above, the RW value means the interval between the read pointer and the write pointer. While the audio data is being input to and output from the buffer, the RW value must be maintained constant so that the audio data can be reproduced while maintaining the quality.

In step 401, the buffer read speed is measured by the test data. In step 407, however, the RW value of the buffer that changes when the voice data is input / output to the buffer is measured. This is because the initial data compensation can be sufficiently performed only by measuring the output speed in step 401.

In step 409, a change in the RW value is monitored in real time to determine whether the RW value has changed. If it has not been changed, the process returns to step 407, and if it has been changed, the process proceeds to step 411. In step 411, it is determined whether the currently received data is a silent period. The meaning of the 'silence interval' will be described later. If it is not the silence period, the process returns to step 411, and if it is the silence period, the process proceeds to step 413. In step 413, data compensation is performed according to the changed RW value. The data compensation in step 413 is called 'real-time data compensation' in order to distinguish it from the initial data compensation in step 403.

The silent interval in step 411 refers to a section of the speech data in which the speech signal is not actually included. The voice data can be distinguished as a section containing human voice information (hereinafter, referred to as 'voice section') and a section having no voice information. A section without this voice information is called a silence section.

The reason for determining whether or not the silence period is included is as follows. If the RW value is changed, data compensation is performed in step 413 as described above. However, if the data compensation is performed in the voice section, the reproduction quality of the voice data may be adversely affected to such an extent that the user can recognize the compensation. Therefore, the data compensation can be performed in the silence period, and the RW value can be kept constant without affecting the reproduction quality of the audio data. The real-time data compensation in step 413 will be described in detail with reference to FIG.

5 is a flowchart for explaining a real-time data compensation method and an initial data compensation method in a buffer control method according to an embodiment of the present invention.

First, the real-time data compensation method in step 411 will be described.

In step 501, it is checked whether the RW value exceeds the RW threshold value. The RW threshold value refers to a maximum value and a minimum value of a value that the RW value can take. The RW threshold value should be predetermined. If the RW threshold is exceeded, the process proceeds to step 503; otherwise, the process proceeds to step 505. It is known that when the RW value exceeds the RW threshold, it usually occurs due to multitasking in a computer system. That is, when a part of the buffer stores different sound data by an instruction to execute another application while reading one sound data, the RP can indicate the address of the buffer where the other sound data is stored. In this case, the positions of RP and WP can be reversed, which is the case when the RW threshold is exceeded. The range of the RW threshold value may vary depending on the setting. For example, if a given RW value is 200 bytes and the threshold ratio is set to 50%, 300 to 500 bytes will be the possible range of RW values. If the RW value exceeds the RW threshold, in step 503, the voice data of the silent interval is output so that the buffer reader can read the voice data of the silent interval instead of the voice data of the current buffer. In this case, the RP in which the reversal occurs can be returned to its original position, and the RW value can be maintained constant again.

In step 505, it is determined whether the RW value determined to have changed in step 409 of FIG. 4 is increased or decreased. The process proceeds to step 507, and if it is decreased, the process proceeds to step 509. [ In step 507, the ratio of decimation is adjusted because the RW value is increased. In step 509, the ratio of interpolation is adjusted because the RW value is decreased. The meaning of the decimation and interpolation is as described in the initial data compensation described in FIG. The initial data compensation in step 403 will be described below. However, the difference from the real-time data compensation in step 411 will be mainly described.

In step 505, the initial data compensation is performed when the determination criterion is changed. That is, the difference is determined not by the increase or decrease of the RW value but by the speed difference measured in the step 401. Also, it is very difficult to assume that the RW value exceeds the RW threshold in the buffer initialization step. This is because the initialization of the buffer implies that the application is to be executed, so that the probability that another application will be executed is extremely rare. Therefore, it is not necessary to judge step 501. That is, data compensation can be performed with interpolation or decimation predicted from the measured speed difference.

6 is a block diagram illustrating a buffer control apparatus for processing voice data according to an embodiment of the present invention.

The buffer 601 receives voice data. The buffer 601 includes a writer 601a and a reader 601b. Writer 601a receives the voice data and writes it into the buffer 501. The reader 601b reads and outputs the voice data stored in the buffer 601. [ The measuring unit 603 measures the reading speed by the test data in the buffer initializing step by the reader 601b and transmits it to the controller 605. [ While the audio data is being processed in real time in the buffer 601 (hereinafter referred to as a 'real-time processing step'), the reader 601a and the writer 601b measure the addresses of the RP and the WP and transmit the measured values to the controller 605 . The addresses of the RP and WP may be read by reading directly from the buffer. The silence detection unit 607 receives the voice data, checks whether the currently received data is a silent period, and transmits the information to the control unit 605. [

The control unit 605 may generate test data and transmit it to the reader 601a in the buffer initialization step. Further, a speed difference between the sampling rate of the sound card and the sampling rate of the sound card is calculated using the reading speed information received from the measuring unit 603, and the difference is transmitted to the data compensating unit 609. In the real time step, the controller 605 obtains the RW value using the information of the RP and the WP received from the measuring unit 603, and transmits the RW value to the data compensator 609. The information received from the silence detection unit 607 is transmitted to the data compensation unit 609.

The data compensator 609 receives the speed difference information from the controller 605 in the initialization step and determines the initial data compensation according to the speed difference. That is, whether or not the interpolation or decimation described in FIG. 5 is performed and the ratio is determined and transmitted to the buffer 601. In the real time step, if the RW value is received from the control unit 605 and the RW value exceeds the RW threshold value, the control unit 605 outputs the silence data received from the control unit 605.

1 is a block diagram of a general buffer for storing voice data,

2 is a view for explaining a speech data processing apparatus according to the prior art,

3 is a conceptual diagram illustrating a speech data processing method according to the prior art,

4 is a flowchart illustrating a buffer control method for processing voice data according to an embodiment of the present invention;

FIG. 5 is a flowchart illustrating a real-time data compensation method and an initial data compensation method in a buffer control method according to an embodiment of the present invention;

6 is a configuration diagram for explaining a buffer control apparatus for processing voice data according to an embodiment of the present invention;

Claims (10)

In the buffer control method, Measuring an initial read speed of the buffer; Calculating a difference between the initial reading speed and a predetermined output speed of the buffer; Determining a ratio of an amount of data input to or output from the buffer according to the calculated speed difference; Calculating a gap between the read pointer and the write pointer by measuring a position of a read pointer and a write pointer of the buffer for inputting or outputting data according to a ratio of the determined data amount; And adjusting the ratio of the amount of data according to whether the interval between the read pointer and the write pointer is changed or not. The method according to claim 1, Wherein the step of determining the ratio of the data amount comprises the step of determining a ratio of the amount of data so that the amount of data stored in the buffer is increased or decreased. The method according to claim 1, Wherein the step of adjusting the ratio of the amount of data comprises the step of outputting voice data that does not include a human voice if the interval between the read pointer and the write pointer exceeds a predetermined threshold value. The method according to claim 1, The step of adjusting the ratio of the amount of data may include adjusting the ratio of the amount of data so that the amount of data stored in the buffer is reduced when the interval between the read pointer and the write pointer does not exceed a predetermined threshold but increases, ≪ / RTI > The method according to claim 1, The step of adjusting the ratio of the amount of data may include adjusting the ratio of the amount of data to increase the amount of data stored in the buffer when the interval between the read pointer and the write pointer does not exceed a predetermined threshold but decreases, ≪ / RTI > A buffer control device comprising: A measuring unit for measuring an initial reading speed of the buffer and measuring a position of a read pointer and a write pointer of the buffer for inputting or outputting data; A control unit for calculating a difference between the initial reading speed and a predetermined output speed of the buffer and calculating an interval between the read pointer and the write pointer; Determines a ratio of an amount of data input or output to the buffer according to the difference between the initial reading speed and the predetermined output speed of the buffer, and determines a ratio of the amount of data to be written to the buffer according to whether the interval between the read pointer and the write pointer is changed And a data compensation unit for adjusting the ratio. The method according to claim 6, Wherein the data compensator determines a ratio of the amount of data to be stored in the buffer so as to increase or decrease the amount of data stored in the buffer. The method according to claim 6, Wherein the data compensator outputs voice data that does not include a human voice when the interval between the read pointer and the write pointer exceeds a predetermined threshold value. The method according to claim 6, Wherein the data compensator adjusts the ratio of the amount of data so that the amount of data stored in the buffer is reduced when an interval between the read pointer and the write pointer does not exceed a predetermined threshold but increases. The method according to claim 6, Wherein the data compensator adjusts the ratio of the data amount so that the amount of data stored in the buffer increases when the interval between the read pointer and the write pointer does not exceed a predetermined threshold value but decreases.

Images