KR100325135B1 - A Voice Signal Error Compensation Algorithm - Google Patents

Abstract

In the present invention, when a loss or error occurs in a voice signal communicated in a digital mobile communication system, the voice signal is silently processed. In particular, the voice data signal is continuously lost or lost a predetermined number of times on a wireless transmission line. In the audio signal error compensation method of the digital mobile communication system, a first process of demodulating and digitally sampling a received high frequency signal and then performing a cyclic redundancy check to determine whether there is an error. And a second process of determining whether an error flag is set when there is no error in the first process, and a third process of converting and outputting a voice when the error flag is not set in the second process. In the process 2, when the error flag is set, the voice in which the error occurred due to triangulation A fourth process of predicting and sequentially converting the signals, and in the first process, if there is an error, comparing the number of occurrences of the error with a predetermined value, and then outputting a level attenuated again. The fifth process is included to compensate for the silence processing for a long time in the urban area where the error of the voice signal is repeated, thereby eliminating the temporary interruption of the phenomenon, and also improving the reliability of the system. There is an industrially available effect to improve.

Description

Voice Signal Error Compensation Algorithm}

According to the present invention, when a loss or error occurs in a voice signal communicated in a digital mobile communication system, the voice signal is muted or compensated for output. In particular, a wireless transmission path The present invention relates to compensating for a voice signal within a predetermined range when a loss or an error occurs continuously due to a sliding collision occurring repeatedly in a phase.

As mobile communication technology has been converted from analog to digital, the quality of audio signals has been greatly improved. However, due to the additional service of data communication, an advantage of digital mobile communication systems, In addition, a large amount of data communication is generated, and an influence on a voice communication signal transmitted by a sliding collision caused by various signal transmission paths affects a lot of errors or losses of digital voice signal data. Was done.

In voice communication using the digital mobile communication system as described above, check whether a data loss or a data error has occurred in the digital voice signal, and correct or error an error or lost voice data signal. By using the Cyclic Redundancy Checking (CRC) method, which is a method of detecting the signal, in addition to the digital voice signal, it is possible to more efficiently cope with an error occurrence of the voice data signal.

That is, by adding a CRC code (Cord) to the digital voice signal to be transmitted, and checking the CRC error at the receiving side, converting only the digital voice signal without error into an analog voice signal, and removing the error signal. In addition, clean voice quality enables smooth voice communication.

Hereinafter, a voice signal error compensation method according to the related art will be described with reference to the accompanying drawings.

1 is a general functional block diagram of a voice signal error compensation terminal according to an embodiment, and FIG. 2 is a flowchart of a voice signal error compensation method according to an embodiment of the prior art.

Referring to the drawings, the conventional method for compensating for the error of a voice signal according to a conventional terminal is described by removing a certain high frequency signal portion from a radio frequency (RF) signal received from an antenna. A low pass filter (LPF) 12 that passes only frequencies; An amplifier (AMP) 14 for amplifying a low level signal applied from the low pass filter 12; A high frequency unit (10) comprising a detector (16) for detecting or demodulating a digital voice signal from the signal passing through the amplifier (14),

ADPCM (Adaptive Differential) converts the analog audio signal from the high frequency unit 10 into a digital audio signal, and converts it into a 4-bit digital signal by sampling. Pulse Coding Modulation) section 20,

After temporarily storing the digital voice signal converted and sampled by the ADPCM unit 20 in a memory buffer 35, the cyclic redundancy check (CRC) of the signal temporarily stored in the buffer 35 is performed. A CPU 30 for simultaneously outputting a digital voice signal and a CRC test result stored in the buffer 35;

A mute control unit 40 for muting or passing the digital voice signal simultaneously applied from the buffer 35 by the CRC test result signal applied from the CPU 30,

A voice converter 50 for converting the digital voice signal applied from the silence controller 40 into an analog voice signal;

The speaker 60 is configured to convert the electric voice signal applied from the voice conversion unit 50 into an audible frequency so that it can be heard by the human ear.

In addition, the prior art method for compensating for a voice signal error by using the apparatus according to the above configuration comprises: a reception determination step (S1) of determining whether a digital mobile communication signal is received through an antenna;

When the digital mobile communication signal is received by the reception determination step (S1), high frequency demodulation is performed through the high frequency unit 10, and the digital signal is converted into a 4-bit digital signal sampled through the ADPCM unit 20. Processing step (S2),

A check step (S3) of performing a cyclic redundancy check (CRC) of the signal applied from the digital processing step (S2) in the CPU 30;

An error judging step S4 for judging whether an error is found by the checking step S3;

If an error is found in the error judging step S4, after the mute processing is performed by the mute control unit 40, the mute control step S6 returns to the reception judging step S1;

If it is determined that there is no error in the error determination step (S4), and passes through the silence control unit 40, the voice conversion unit 50 is converted to an analog voice signal, and then the reception determination step (S1) Is converted into a speech conversion step (S5).

Hereinafter, a detailed operation of the conventional speech signal error compensation device and method by the above configuration will be described with reference to the accompanying drawings.

If it is determined that the voice communication signal is applied as a high frequency radio signal from the base site of the digital mobile communication system (S1), the high frequency component is applied to the low pass filter 12 of the high frequency unit 10. Removes and removes the high frequency component to the amplifier 14 to amplify the signal to an appropriate level, and applies the signal amplified to the appropriate level to the detector 16 to convert or demodulate the digital audio signal. do. The ADPCM unit 20, which receives an analog digital voice signal from the high frequency unit 10, samples the quantized signal and quantizes the received signal into a 4-bit digital voice signal. To be molded (S2).

The CPU 30, which has received the shaped 4-bit signal from the ADPCM unit 20, temporarily stores the shaped 4-bit signal in the buffer memory 35, and at the same time, the shaped 4-bit signal. By performing a cyclic redundancy check (CRC) (S3), it is determined whether an error occurs in the transmission process (S4).

As a result of performing a CRC check, the CPU 30, together with a control signal that determines whether there is an error or no error, is applied from the ADPCM unit 20 and stores a digital voice signal stored in the memory buffer 35 as a frame ( When the silence control unit 40 is applied to the silence control unit 40 and the silence control unit 40 receives the control signal of the error determination result received from the CPU 30, that is, there is no error. The digital voice signal stored in the memory buffer 35 as applied through the CPU 30 is applied to the voice converter 50 as it is, so that the voice of the analog voice is converted into analog voice. By converting into a signal and applying it to the speaker 60, it is converted into an audible signal.

However, when it is determined by the CPU 30 that there is an error in the error determination control signal applied to the mute control unit 40, the mute control unit 40 receives the 4-bit digital received from the memory buffer 35. By not allowing the voice signal to pass through, the voice converter 50 does not receive a digital signal to be converted into an analog voice signal.

Therefore, the digital voice signal of the frame in which the error is detected is not converted to analog, and the user of the terminal does not hear the error signal, but feels that the voice signal is cut off in a very short moment.

In addition, a signal applied to an antenna through a reflection path reflected by a high-rise building or the like in a digital mobile communication signal overlaps with a signal applied to the antenna through a direct path, thereby causing a wavelength difference of λ / 4. In the case of the sliding collision, the CRC check in the CPU 30 causes an error, and the error causes a phenomenon that a voice signal is continuously heard and not heard for a long time. Therefore, there is a problem that exceeds the emotional limit of the terminal user, there is a problem of reducing the reliability of the digital mobile communication system.

According to the present invention, the sound signal input in the previous step is attenuated instead of the sound signal that is error-prone and silenced, and then outputted, and once again, before the output of the good sound signal input in the next step, It is an object of the present invention to provide a method for outputting a signal obtained by triangulating a speech signal and a good speech signal of the previous stage.

The present invention devised to achieve the above object, in the voice signal error compensation method of the digital mobile communication system, demodulating and digitally sampling the received high-frequency signal, the cyclic redundancy check to determine whether there is an error A first step, a second step of determining whether an error flag is set when there is no error in the first step, and a third step of converting and outputting a voice when the error flag is not set in the second step; In the second process, when the error flag is set, a fourth process of predicting a normal state of the speech signal having an error by triangulation and sequentially converting the speech; In this case, a fifth process of comparing the number of occurrences of the error with a predetermined value and then attenuating and outputting the output voice conversion signal is performed. The voice signal error compensation method is characterized by the above.

1 is a general functional block diagram of a voice signal error compensation terminal according to an embodiment;

2 is a flowchart of a voice signal error compensation method according to an embodiment of the prior art;

3 is a flowchart of a voice signal error compensation method according to an embodiment of the present invention.

Explanation of symbols on the main parts of the drawings

10: high frequency unit 12: low pass filter 14: amplifier

16: detector 20: ADPCM 30: CPU

35: memory buffer 40: silence control unit 50: voice conversion unit

Hereinafter, a voice signal error compensation method according to the present invention will be described with reference to the accompanying drawings.

3 is a flowchart of a voice signal error compensation method according to an embodiment of the present invention.

1 to 3 attached to the above, the voice signal error compensation method of the present invention is partially overlapped with the method according to the prior art, and whether a digital mobile communication data signal is wirelessly received through an antenna. A reception determination step (S1) for judging;

When the digital mobile communication signal is received by the reception determination step (S1), high frequency demodulation is performed through the high frequency unit 10, and the digital signal is converted into a 4-bit digital signal sampled through the ADPCM unit 20. Processing step (S2),

A check step (S3) of performing a cyclic redundancy check (CRC) of the signal applied from the digital processing step (S2) in the CPU 30;

A first process comprising an error judging step S4 for determining whether an error is found by the checking step S3;

If there is no error in the first process, a second process of determining by the CPU 30 whether an error flag is set;

In the second process, when the error flag is not set, the CPU 30 sets a first flag setting step of setting the error flag internally to '0';

A first storing step of storing an error-free digital voice signal in F1, which is a specific place of the memory buffer 35;

In the first storing step, the voice signal stored in the specific place F1 of the memory buffer 35 is applied to the mute control unit 40, and at the same time, the mute control unit 40 is controlled to transmit the signal to the voice conversion unit 50. A third process consisting of a first speech conversion step of converting the analog speech signal into an analog speech signal, applying it to the speaker 60, and then feeding back to the reception determination step S1;

In the second process, when the error flag is set to '1', the CPU 30 sets a second flag setting step (S20) of setting the error flag to '0';

A second storage step (S30) of reading the digital voice signal stored in the memory buffer 35 and storing the current error-free voice signal in the memory buffer 35;

A triangulation step (S40) of triangulating a digital voice signal read from the memory buffer 35 and the current digital voice signal stored in the memory buffer 35 by a specified formula in step S30;

By the triangulation step (S40), the prediction step (S50) for predicting the steady state of the error-produced speech signal,

The speech signal predicted by the prediction step S50 and the current speech signal stored in the memory buffer 35 are sequentially output to the mute control unit 40, and at the same time, the CPU 30 applies a control signal to the The second voice conversion step (S60) to be applied to the voice conversion unit 50, to be converted into an analog voice signal, to be applied to the speaker 60, and fed back to the reception determination step (S1) of the first process. The fourth process consists of,

In the first process, if there is an error in the digital voice signal by CRC checking, a third flag setting step (S70) of setting an error flag of the CPU 30 to '1';

A counter step S80 of increasing the value of the error counter EX by adding '1' to the error counter 'EX' of the CPU 30,

A determination step (S90) of comparing and determining whether the value of the error counter EX is greater than a preset counter value 'X' in the step S50;

In the determining step S90, when the EX value of the error counter is smaller than the preset X value, the digital voice signal recorded in the memory buffer 35 is read, the level is lowered by a predetermined level, and the silence control unit ( 40 is applied to the voice conversion unit 50 by a control signal applied from the CPU 30, and is converted into a low level analog voice signal and outputted to the speaker 60. A third voice conversion step (S100) fed back to the reception determination step (S1) of step 1,

In the determination step (S90), if the value of the error counter EX is greater than the preset 'X' value, the CPU 30 applies a control signal to the mute control unit 40, thereby blocking the signal from passing through Silence step (S110) and mute processing,

After the initializing of the value of the error counter EX to '0', it consists of a fifth process consisting of an initialization step (S120) for feeding back to the reception determination step (S1) of the first process.

Hereinafter, a voice signal error compensation method according to the present invention having the above configuration will be described in detail with reference to the accompanying drawings.

Since the first process operates in the same manner as in the prior art, a detailed description is not added.

If it is determined in the error determination step (S4) that there is no error in the digital voice signal, it is determined whether the '1' of the indication that there is an error is set in the error flag (S10), and if it is determined that there is no error, the error flag is determined. Is set to '0' (S21), the error-free digital voice signal is stored in the memory buffer 35 (S31), and then applied to the mute control unit 40 as well. At this time, the CPU 30 controls the mute control unit 40 so that the error-free digital voice signal is applied to the voice conversion unit 50 and converted into an analog voice signal (S61), so that the analog voice signal is a speaker. Is applied to 60, and is converted into an audible signal.

In addition, in the second process, when the error flag is set to '1' and it is determined that there is an error in the past digital voice signal (S10), the error flag is set to '0' ( S20), a voice signal that has been recorded in the buffer 35 as a memo without a previous error is read out, and a current error-free voice signal is recorded in the memory buffer 35 (S30).

The CPU 30 performs triangulation according to the following formula.

Sk (n) = [((Gk-Fk) / (N-1)) * (n + 1) + Fk] (Equation 1)

0 ≤ k ≤ N-1, 0 ≤ n ≤ N-1

Sk; Triangulation value = expected normal signal coefficient of the errored speech signal

Gk; Kth conversion coefficient of the current normal speech signal,

Fk; Kth transform coefficient of the past normal speech signal

Equation (1) can be predicted from the normal speech signal of the error-produced speech signal by performing linear interpolation by dividing the difference between the k-th conversion coefficient Fk and Gk by being extracted from the current and past normal speech signals.

N denotes an input frame order of the digital voice signal in the time domain.

In more detail and specifically, the equation 1 is described. For example, a signal in which an error occurs is referred to as frame 2, and the signal is generated before the frame 2 in which the error occurs. The normal audio signal is called Frame 1, and it is assumed that the normal signal is Frame 3, which is generated after Frame 2 in which the error occurs. The normal signal frame 1 or frame 3 is stored in the memory buffer 35. If a CRC error occurs in the audio signal of the frame 2, the normal audio signal frame 1 recorded in the memory buffer 35 is read, and after the frame 2 in which the error occurs, By applying the received frame 3 (Frame 3) to the equation 1 together with the frame 1 (Frame1) in order, triangulation is performed, and the audio signal of the frame 2 (CRC 2) in which the CRC error is generated by the triangulation It is predicted (S50).

The CPU 30 applies the predicted speech signal and the current error-free speech signal to the mute controller 40 in order, and converts the analog speech signal by the speech converter 50.

In addition, if it is determined that there is an error by the error determination step (S4), the error flag is set to '1' to indicate that there is an error (S70), and '1' is added to the error counter 'EX'. It counts up (S80) and determines whether the value of the error counter EX is large or small compared to a preset value 'X' (S90).

In the above process, the error compensation according to the present invention is to set a certain limit because the signal predicted by Equation (1) becomes inaccurate when the compensation by the continuous error is continued.

That is, when an error occurs less than a preset value of X in the determination step (S90), after reading the normal normal voice signal stored in the memory buffer, lowering the level to a certain level, the mute control unit ( 40) and the voice converter 50 to output a low level analog voice signal.

Therefore, the silent signal is generated so that the call cannot be continued and there is no feeling of being interrupted.

Nevertheless, if the value of the error counter EX is greater than the value of X previously set, the silence control unit 40 is controlled to prevent the signal from being applied to the voice conversion unit 50 (S110). After initializing the value to '0' (S120) and returning to the reception determination step of the first process.

Summarizing briefly the above-mentioned operation, the CRC test of the digital voice signal, if there is an error, calls back the error-free voice signal input in the previous step, and lowers the level to output the error. When the normal signal is applied in the next step, by predicting and generating the normal signal of the errored speech signal by triangulating the normal signal currently input and the currently input normal signal, the current normal signal is output. By converting the analog outputs one after the other, the error can be compensated more reliably.

According to the present invention having the above-described configuration, by silently processing a signal in which an error has occurred, outputting a signal which has been normally input in the past just before, at a low level, and outputting a prediction signal by a normal signal input in the next step again. In addition, it is possible to compensate for the silent processing for a long time in the urban area where the error of the voice signal is continuously repeated by the sliding collision phenomenon, which has the effect of eliminating the temporary disconnection of the user, and also improves the reliability of the system. There is an effect that can be used industrially.

Claims (4)

(Correction) In the audio signal error compensation method of a digital mobile communication system, Demodulating and digitally sampling the received high frequency signal, and performing a cyclic redundancy check to determine whether there is an error; A second step of determining whether an error flag is set when there is no error in the first step; In the second process, if the error flag is not set, a third process of voice conversion and output; In the second process, when the error flag is set, a fourth process of predicting the normal state of the speech signal having an error by triangulation and sequentially converting the speech; In the first process, if there is an error, a voice signal error comprising a fifth process of comparing the number of occurrences of the error with a predetermined value and then attenuating and outputting the output voice conversion signal again Compensation method. According to claim 1, The third process may include setting a first flag to set an error flag to 0; A first storage step of storing an error-free voice signal in a memory buffer; And a first voice conversion step of converting a voice signal stored in a memory buffer into an analog voice signal in the first storage step and then feeding back to the first process. (Correction) The method according to claim 1, The fourth process includes: setting a second flag to set an error flag to 0; A second storage step of reading a voice signal stored in a memory buffer and storing a current error-free voice signal; A triangular conversion step of triangular conversion using the read-out voice signal and the stored voice signal; A prediction step of predicting a steady state of an audio signal in which an error is generated by the triangulation step; And a second speech conversion step of sequentially outputting the speech signal predicted by the predicting step and the speech signal stored in the memory buffer and then feeding back to the first process. According to claim 1, The fifth process may include setting a third flag to set an error flag to 1; A counter step that increments the error counter, A judgment step of judging whether the value countered in the step is greater than a preset value; A third voice conversion step of reading a voice signal recorded in a memory buffer, lowering a level, and performing voice conversion if the value is smaller than a set value, and then feeding back to the first step; In the determining step, if greater than the set value, the silent step of silence processing so that the voice signal is not converted and output; And an initialization step of returning to the first process after initializing the error counter.