JP5599487B2 - Audio processing apparatus for ADPCM audio transmission system and audio processing method therefor - Google Patents

Description

  The present invention relates to an audio processing device of an ADPCM audio transmission system that transmits audio data differentially quantized by the ADPCM method and an audio processing method thereof.

  Cordless telephones and PHS (Personal Handyphone System) are known as examples of a voice processing device that converts voice into differentially quantized voice data by the ADPCM method and transmits / receives the voice data in units of transmission frames.

  For example, in (Patent Document 1), the received signal strength of ADPCM audio data in a transmission frame is detected, and only the difference between ADPCM audio data having a received signal strength of a certain value or less among ADPCM audio data determined to have an error. An error processing method of an ADPCM audio transmission system applied to PHS which is decoded after a value is reduced or no difference is described is described.

JP-A-8-288914

  However, in the error processing method of the ADPCM audio transmission system described in (Patent Document 1), correction processing is performed only when the received signal strength is a certain value or less. Accordingly, when a reception error occurs due to radio wave interference even when the received signal strength is sufficient, clickable noise (hereinafter referred to as a popping sound) that may be violently generated may occur. . Further, if only the difference value is replaced with a small value including “0” indicating no difference, a pop sound may be generated depending on the audio data.

  Therefore, the present invention is able to cope with a reception error due to radio wave interference or the like, thereby preventing the generation of a pop sound and minimizing a decrease in voice quality when a reception error occurs. It is an object of the present invention to provide a voice processing apparatus and a voice processing method thereof.

  The present invention is an audio processing apparatus of an ADPCM audio transmission system that transmits audio data differentially quantized by the ADPCM method, wherein the transmission frame is composed of a plurality of subframes, and each of the plurality of subframes has an error. And an audio processing unit that executes correction processing when an error is detected by the error detection unit, and the audio processing unit determines that there is an error by the error detection unit. Whether or not the subframe has reached the predetermined range in the transmission frame. If the subframe in which the error has occurred does not reach the predetermined range, the correction is performed at the first time interval. If the subframe in which an error has occurred extends over the predetermined range, correction is performed at a second time interval, and the second time interval is Shorter than the first time interval.

  The present invention corrects the sound data by the sound processing unit in accordance with the presence or absence of individual errors for a plurality of subframes. Therefore, even if the error occurrence state is different due to radio wave interference or the like, the sound data is appropriately corrected. Correction can be performed. Therefore, the present invention can cope with a reception error due to radio wave interference or the like, thereby preventing the generation of a popping sound and minimizing a decrease in voice quality when a reception error occurs.

(A) Block diagram showing transmission unit of parent device, (B) Block diagram showing reception unit of child device Diagram for explaining transmission frame format Flow chart showing audio processing of the receiving unit of the slave unit Flow chart showing reception error processing Flow chart showing subframe processing Diagram for explaining error value and correction interval value Flow chart showing error counter processing Illustration for explaining the error counter Flow chart showing received voice data processing Flow chart showing processing without receiving error

A first aspect of the present invention is an audio processing apparatus of an ADPCM audio transmission system that transmits audio data differentially quantized by the ADPCM method, wherein the transmission frame is composed of a plurality of subframes, and the received transmission frame An error detection unit that individually detects the presence or absence of an error for the plurality of subframes; and an audio processing unit that executes correction processing when an error is detected by the error detection unit, the audio processing unit Determines the frequency of occurrence of a subframe error by the error detection unit, performs correction at a first time interval when the frequency of occurrence of the subframe error is less than a threshold , and determines the frequency of occurrence of the subframe error. If it is equal to or greater than the threshold value , correction is performed at the second time interval, and the second time interval is shorter than the first time interval.

In the second aspect of the present invention, when the frequency of occurrence of the subframe error determined by the error detection unit is less than a threshold , the audio processing unit corrects the error according to the error weighting degree indicated by the first value. If the occurrence frequency of the subframe error is equal to or greater than a threshold value, correction is performed according to the error weighting degree indicated by the second value, and the error weighting degree indicated by the second value is It is more severe than the error weight indicated by the value of.

In the second aspect , since the audio data is corrected by the audio processing unit with a correction value corresponding to the error weight, the audio data is appropriately corrected even if the error occurrence state is different due to radio wave interference or the like. be able to.

According to a third aspect of the present invention, in the second aspect , the sound processing unit subtracts or adds a value corresponding to the error weighting degree from the sound data according to a code accompanying the sound data of ADPCM. Then, the processing in the direction to suppress the change of the value after ADPCM decoding is performed.

In the third aspect , when an error occurs, if the sign of ADPCM audio data (difference value) indicates positive (addition of difference value), a value corresponding to the error weight is subtracted from the audio data. If the sign of the audio data (difference value) indicates a negative value (subtraction of the difference value), a value corresponding to the error weight is added to the audio data. This suppresses the change in the value after ADPCM decoding. Thus, the volume after decoding is reduced according to the error weight. By this processing, it is possible to prevent an excessive correction compared to the correction that replaces the value indicating mute.

According to a fourth aspect of the present invention, in the second aspect or the third aspect , the error detection unit detects an error in a synchronization word, control data, or audio data that constitutes a transmission frame, and the audio processing unit As the error weighting, the error of the voice data is made heavier than the synchronization word and control data.

In the fourth aspect , in the error that occurs in each of the synchronization word, control data, and audio data, the weight of the error that occurs in the audio data is heavy, thereby preventing excessive deterioration in sound quality. doing.

A fifth aspect of the present invention, in to free the second aspect one of the fourth aspect, the audio processing unit, the sub-frame error by performing the integration based on the value corresponding to the weight of the error The occurrence frequency is calculated, and the audio data is corrected with a correction value based on the occurrence frequency of the subframe error .

In the fifth aspect , even if the error weight is light, if the error occurrence frequency is high, the effect of the error can be suppressed by increasing the correction degree. By performing integration based on the value and correcting the audio data with the correction value based on the integration value, it is possible to cope with frequent occurrence of errors over a plurality of transmission frames.

According to a sixth aspect of the present invention, in the fifth aspect , the sound processing unit corrects the sound data with a correction value corresponding to the lightness when the error weight is light, and generates a subframe error. The correction by the correction value based on the frequency is delayed after a predetermined number of transmission frames including an error are received.

In the sixth aspect , when minor errors continuously occur, if correction is performed based on the integrated value, which is a correction value based on the error occurrence frequency, there is a risk of excessive correction. Therefore, when the error weight is mild, the audio data is corrected with the correction value according to the mildness, and the correction by the correction value based on the error occurrence frequency is received, and a predetermined number of transmission frames not including the error are received. By doing so, the correction by the integrated value is delayed. By doing so, it is possible to reduce the influence on sound quality while suppressing the generation of popping sounds.

According to a seventh aspect of the present invention, in the second aspect , the number of subframes in which an error has occurred in one transmission frame is counted, and an error value or a correction interval value is set according to the counted number of subframe errors. Set. As a result of this processing, the larger the number of subframe errors, the larger the error value or the smaller the correction interval value, so that the audio data can be corrected more appropriately.

An eighth aspect of the present invention is an audio processing method of an ADPCM audio transmission system for transmitting audio data differentially quantized by the ADPCM method, wherein the transmission frame is composed of a plurality of subframes, and the plurality of subframes are determining the frequency of occurrence of the sub-frame error by individually detecting the presence or absence of an error, if the frequency of occurrence of the sub-frame error is less than the threshold value, performs a correction in a first time interval, said sub-frame error When the occurrence frequency is equal to or higher than the threshold value , correction is performed at a second time interval, and the second time interval is shorter than the first time interval.

(Embodiment)
An ADPCM audio transmission system according to an embodiment of the present invention will be described with reference to the drawings by taking a cordless telephone as an example. 1A and 1B are diagrams showing a cordless telephone according to an embodiment of the present invention, in which FIG. 1A is a block diagram showing a transmission unit of a parent device, and FIG. 1B is a block diagram showing a reception unit of a child device. is there.

  A cordless telephone 1 shown in FIG. 1 that functions as an ADPCM audio transmission system includes a parent device 2 connected to a telephone line, and a child device 3 that transmits and receives the parent device 2 using radio signals. In FIG. 1, the base unit 2 shows only the transmitting unit 20 and the handset 3 shows only the receiving unit 30, but the base unit 2 demodulates and decodes the radio signal from the handset 3. The slave unit 3 is provided with a transmission unit that encodes and modulates a voice signal and transmits a radio signal to the master unit 2.

  The transmission unit 20 of the base unit 2 includes a telephone control unit 21, an A / D converter 22, a quantization compression unit 23, an ADPCM encoding unit 24, a transmission frame generation unit 25, and a wireless transmission unit 26. ing.

  The telephone control unit 21 sends and receives calls and voice signals to and from the telephone line. The A / D converter 22 converts an analog audio signal into digital audio data. The quantization compression unit 23 performs quantization processing and compression processing on the audio data based on the A-law or the μ-law. The ADPCM encoding unit 24 converts the voice data subjected to quantum compression into an ADPCM G. In step 726, a 4-bit adaptive prediction difference value is encoded. In the ADPCM method, a difference between actually measured values with respect to adaptively predicted values is output as voice data.

  The transmission frame generation unit 25 generates a transmission frame including a synchronization word, control data, audio data as a difference value, and CRC as check data as a transmission frame. The radio transmission unit 26 transmits a radio signal obtained by modulating the input transmission frame by the time division multiplexing method via the antenna 26a.

  The reception unit 30 of the slave unit 3 includes a radio reception unit 31, a reception frame separation unit 32, an error detection unit 33, an error determination unit 34, an audio processing unit 35, an ADPCM decoding unit 36, and an inverse quantization. An expansion unit 37, a D / A converter 38, and an amplification unit 39 are provided.

  The wireless reception unit 31 demodulates a wireless signal received via the antenna 31a and outputs it as a transmission frame. The reception frame separation unit 32 separates the transmission frame into a synchronization word, control data, voice data, and CRC. The error detection unit 33 detects whether or not an error has occurred in the separated synchronization word, control data, and audio data. The synchronization word is detected based on whether or not it has a predetermined bit pattern. The control data and the voice data detect the occurrence of an error by checking the corresponding CRC. The error determination unit 34 determines the degree of weight of the detected error or the occurrence frequency of the error.

  The voice processing unit 35 corrects the voice data that is the difference value (4 bits) before ADPCM decoding based on the correction value corresponding to the result determined by the error determination unit 34. When an error has occurred, the audio data is output after performing the processing described later, and when there is no error, it is output without correction.

  The ADPCM decoding unit 36 generates decoded audio data based on the input audio data (4-bit difference value) and a prediction value obtained by internal prediction. If an error has occurred, the audio data is corrected by the audio processing unit 35 while remaining 4 bits, and the ADPCM decoding unit 36 performs the decoding process based on the corrected audio data.

  The inverse quantization expansion unit 37 performs inverse quantization processing and expansion processing on the audio data based on the A-law or the μ-law. The D / A converter 38 converts audio data that is digital data into an analog audio signal. The amplifying unit 39 amplifies the analog audio signal and outputs it to the speaker 39a.

  In the present embodiment, the error detection unit 33, the error determination unit 34, the audio processing unit 35, and the ADPCM decoding unit 36 constitute an audio processing device.

  As described above, the cordless telephone 1 according to the embodiment of the present invention is configured. Here, the transmission frame generated by the transmission frame generation unit 25 of the base unit 2 and the transmission frame separated by the reception frame separation unit 32 will be described in detail with reference to FIG. FIG. 2 is a diagram for explaining the format of a transmission frame.

  The transmission frame includes audio data encoded by the ADPCM encoding unit 24. The transmission frame includes a synchronization word, control data, a CRC for this control data, each subframe from the first subframe to the fourth subframe obtained by dividing the bit string of audio data into four, and a CRC corresponding to each subframe. And have.

  Since the audio data transmitted from the wireless transmission unit 26 at 32 kbps is one transmission frame every 10 ms, one transmission frame includes 320-bit audio data. Therefore, one subframe is 80 bits, and one ADPCM word in 4 bits is 20 words.

  The transmission frame generation unit 25 generates such a transmission frame and outputs it to the wireless transmission unit 26. In addition, the reception frame separation unit 32 separates the transmission frames and outputs them to the error detection unit 33, and also outputs the audio data to the audio processing unit 35.

  Next, the procedure of the voice processing method of the receiving unit 30 of the handset 3 that has received this transmission frame will be described with reference to FIGS. FIG. 3 is a flowchart showing audio processing of the receiving unit 30 of the child device 3. FIG. 4 is a flowchart showing reception error processing. FIG. 5 is a flowchart showing subframe processing. FIG. 6 is a diagram for explaining the error value and the correction interval value. FIG. 7 is a flowchart showing the error counter process. FIG. 8 is a diagram for explaining the error counter. FIG. 9 is a flowchart showing received voice data processing. FIG. 10 is a flowchart showing the processing without receiving error.

  First, the wireless reception unit 31 that has received the transmission frame demodulates and outputs it to the reception frame separation unit 32. In S100 shown in FIG. 3, the reception frame separation unit 32 separates the synchronization word included in the transmission frame, the CRC corresponding to the control data and the control data, and the CRC corresponding to the voice data and the voice data as described above. Then, each is output to the error detection unit 33 and the audio data is output to the audio processing unit 35.

  In S200, the error detection unit 33 generates a CRC error from the first subframe to the fourth subframe, whether the received synchronization word is a correct bit arrangement, whether a CRC error has occurred in the control data, or not. And the result is output to the error determination unit 34 as reception error information.

  In S300, error determination unit 34 determines whether or not a reception error has occurred based on the reception error information. Here, it is assumed that a reception error has occurred as a determination result of the error determination unit 34.

  If a reception error has occurred in S400, reception error processing is performed. This reception error process will be described with reference to FIG.

  In the reception error process shown in FIG. 4, it is first determined in S401 whether or not an error has occurred in the synchronization word. If no error has occurred in the synchronization word, it is determined in S402 whether a CRC error has occurred in the control data. If no control data error has occurred, it is determined in S403 whether a CRC error has occurred in the audio data of the first to fourth subframes. If no audio data error has occurred, “0” is substituted as the error value in S404, and the reception error process is terminated.

  If an audio data error has occurred in S403, the subframe process is executed in S420a and the process is terminated. The subframe processing will be described later.

  If a control data error has occurred in S402, the process proceeds to S405 to determine whether or not an audio data error has occurred. If no audio data error has occurred, “0” indicating no error is substituted for the error value in S406. In this case, the correction interval value is set to “4” in S407. If an audio data error has occurred in S405, subframe processing is executed in S420b.

  In this example, if no error is detected or if it is a minor error (when there is no voice data error even if a synchronization word error or control data error has occurred), “0” is set as the error value. Assigned. The correction interval value is a time interval in which the presence or absence of an error is checked and correction is performed. If this value is large, error detection / correction processing is performed at a long time interval, and as this value decreases, the correction interval value is performed at a shorter interval. . If an error is detected but is insignificant as described above, instead of setting the error value to “0”, the correction interval value is set to a relatively large value “4”. Correction processing is performed.

  If a synchronization error has occurred in S401, the process proceeds to S408 to determine whether a control data error has occurred. If no control data error has occurred, it is determined in S409 whether or not an audio data error has occurred. If no audio data error has occurred, “0” is substituted into the error value in S410. In step S411, the correction interval value is set to “4”. Then, the reception error process ends. Details of the correction interval value will be described later. If an audio data error has occurred in S409, the subframe process is executed in S420c, and this process ends.

  Thus, even if a synchronization word error has occurred, if there is no control data error and no voice data error, “0” is assigned as the error value. Also in this case, the correction interval value is set to a relatively large value “4”, and thereafter error detection / correction processing is performed at a rough interval.

  If a control data error has occurred in S408, it is determined in S412 whether an audio data error has occurred. If no audio data error has occurred, “1” is substituted into the error value in S413. In the next step S414, the correction interval value is set to “4” and the reception error process is terminated. If an audio data error has occurred in S412, a subframe process is executed in S420d, and this process ends.

  As described above, in the reception error process, an error value corresponding to the error generated in the synchronization word or the error generated in the control data and the subsequent correction interval value are set. For errors that occur in audio data, error values and correction interval values are set in subframe processing described below.

  This subframe processing will be described with reference to FIG. In the subframe processing, first, the variable “number of subframe errors” is set to “0” as an initial value in S421. The variable i is set to “1”. This variable i is a counter for inspecting from the first subframe to the fourth subframe. In S422, subframe error information related to the subframe is acquired from the reception error information from the error detection unit 33.

  In S423, it is determined whether an error has occurred in the subframe indicated by the variable i. That is, since the variable i is “1” at the beginning, it is determined whether or not a CRC error has occurred in the first subframe. If no error has occurred, the process proceeds to S424. In S424, 1 is added to the variable i. Then, in S425, it is determined whether or not the variable i is larger than the number larger than the total number of subframes (“4” in the present embodiment). When the variable i is 4 or less, the process proceeds to S423.

  If a CRC error has occurred in the subframe in S423, “1” is added to the variable “number of subframe errors” in S426. Then, in S427, it is determined whether or not the variable “number of subframe errors” is equal to or smaller than a head mute threshold “2” described later in detail. If the variable “number of subframe errors” is larger than the head mute threshold “2”, the process proceeds to S424, and if it is less than the head mute threshold “2”, the head mute flag is set to be valid in S428. The head mute flag is for instructing the mute effect from the head subframe. When an error is detected and i is equal to or less than the threshold value “2” (first subframe, second subframe), the head mute flag is used. The mute flag is enabled and mute processing is performed from the first subframe.

  When all subframes have been inspected, it is determined in S429 whether or not the variable “number of subframe errors” is equal to or greater than the subframe error threshold. In this embodiment, the subframe error threshold is “3”. In other words, in S429, if there are three or more subframes in which a CRC error has occurred in one transmission frame, the subframe error threshold is exceeded.

  If the number of subframes in which an error has occurred is less than the threshold “3”, if it is determined in the subframe processing that a synchronization error and a control error have occurred, the subframe error value is set to “ “2” is set, otherwise “1” is set. In S431, the correction interval value is set to “4”.

  If the number of error-occurring subframes is greater than or equal to the threshold “3” in S429, “3” is set as the error value in S432. If it is determined in the subframe processing that a synchronization error and a control error have occurred, “1” is set as the correction interval value in S433, and “2” is set otherwise.

  In this way, in subframe processing, the presence of errors is individually checked for a plurality of subframes constituting a transmission frame, the number of error occurrences is counted, and an error value and a correction interval value corresponding to the counted number of subframe errors. Set. In addition, when the subframe where the error has occurred is positioned before the predetermined subframe, that is, when it is the first or second subframe, a setting is made to enable the head mute flag.

  Here, error values and correction interval values will be described with reference to FIG.

  No. Since 1 is a state in which no error occurs in the transmission frame, the error value is “0”, which means the lowest level, and the audio data is decoded without correction. Since no correction is performed, the correction interval value is not set.

  No. 2 and no. 3 is a case where either the synchronization error or the control data has an error, and the audio data has no error. At this time, “0” is set as the error value, and “4” is set as the correction interval value. When the transmission packet is transmitted from the transmission side by time division multiplexing, the reception side does not completely synchronize the transmission packet with the synchronization word, but synchronizes with the clock when it is normally received at the reception side. Voice data can be received. That is, even if there is an error in either the synchronization error or the control data, it is possible to reproduce the sound if there is no error in the sound data. In this case, by setting the correction interval value to “4”, the correction process is performed at a low frequency with a relatively large time interval.

  No. 4 is a state in which both the synchronization error and the control error are errors, but there is no error in the audio data. In this case, there is a possibility that the transmission frame cannot be normally received as a whole. Therefore, even if there is no error in the audio data, there is a possibility that an error due to CRC cannot be detected, so “1” is set as the error value. Also in this case, the correction frequency is set low by setting the correction interval value to “4”.

  No. 5 to No. 7 is a state in which both the synchronization error and the control data are not errors at the same time, and the total number of subframes in which the error has occurred is less than the threshold “3”. In this case, since the audio data error has occurred, it is necessary to correct the audio data. Since the number of error-occurring subframes is small, mild correction is sufficient. The error value is “1” and the correction interval value is “ 4 ”.

  No. 8 is a state in which both the synchronization error and the control data are errors, and the total number of subframes in which the error has occurred is less than the threshold “3”. In this case, the frequency of occurrence of errors in the subframe is low, but since both the synchronization error and the control data are errors, there is a possibility that an error also occurs in a normal subframe as described above. . Therefore, the error value is “2” indicating a medium error, and the correction interval value is “4”.

  No. 9 to No. 11 is a state in which both the synchronization error and the control data are not errors at the same time, and the total number of subframes in which the error occurs is equal to or greater than the threshold “3”. In this case, since the occurrence frequency of the subframe error is high, the error value is set to “3” indicating a severe error, and the correction interval value is set to “2”. By reducing the correction interval value, the degree of ADPCM word correction increases.

  No. 12 is a state in which both the synchronization error and the control data are errors, and the total number of subframes including the error is equal to or greater than the subframe error threshold. In this case, since the frequency of occurrence of errors in the subframe error is high, the error value is set to “3” indicating a severe error, and the correction interval value is set to “1”.

  As described above, in the present embodiment, it is determined that the error weight generated in the voice data is heavy among the errors generated in the synchronization word, the control data, and the voice data. By doing so, excessive sound quality degradation is suppressed.

  Returning to FIG. 3, when the reception error process is completed in S400, an error counter process is performed in S500. This error counter process will be described with reference to FIG.

  In the error counter process shown in FIG. 7, first, in S501, it is determined whether or not the error value set in the reception error process and the subframe process executed in the reception error process is “0”. If the error value is “0”, it indicates that none of the synchronization error, the control error, and the voice data error has occurred in the received transmission frame, and the process is terminated as it is.

  If it is determined in S501 that the error value is not “0”, it is determined in S502 whether or not the error value is equal to or less than a mild threshold value. That is, it is determined by comparing the error value with the mild threshold value whether or not the degree of the detected error is minor. In the present embodiment, it is determined that the error is minor with the minor threshold being “1”.

  If it is determined that the error value is less than or equal to the mild threshold value, the delay counter stored in the error determination unit 34 is acquired in S503. The delay counter is provided in order to obtain information on whether or not errors are continuously generated following the previous transmission frame when an error is detected. The audio processing by this delay counter will be described later.

  In S504, it is determined whether or not the delay counter is equal to or smaller than a predetermined threshold value. If the delay counter is equal to or smaller than the predetermined threshold, the error counter is set to “1” and the correction interval value is set to “4” in S505. This error counter is a counter that accumulates values corresponding to error values, and serves as a correction value for correcting audio data. In S505, “1” is fixedly assigned to the error counter. This is because the audio data is corrected with the smallest correction value during the period when the delay counter is less than or equal to the predetermined value, and when the delay counter becomes larger than the predetermined value, the correction delay ends and the integrated value by the error counter becomes the value of the audio data. It means that it becomes a correction value.

  Here, the predetermined threshold is determined as “6”, and when the delay counter is “6” or less, the error counter is set to be “1”. By doing so, a slight correction can be continuously applied to a transmission frame subsequent to a transmission frame in which an error has occurred until the delay counter falls below a predetermined threshold. This is likely to result in excessive correction if correction is performed based on the error counter when minor errors occur continuously. Therefore, when the error weight is mild, the audio data is corrected with a correction value corresponding to the lightness, and the correction by the integrated value is delayed by receiving the predetermined number of transmission frames. The influence on sound quality can be reduced while suppressing the generation of sound.

  In S506, it is determined whether or not the delay counter is “0” or whether or not the head mute flag is set. Whether or not the delay counter is “0” is determined by determining whether or not a delay of correction using the integrated value of the error counter as a correction value is performed subsequently from the previously received transmission frame. The determination as to whether or not the head mute flag is set is to determine whether or not the occurrence of an error in the audio data is in the first subframe or the second subframe.

  Therefore, when the delay counter is “0”, it indicates that the correction delay has not been performed subsequently from the previously received transmission frame, and therefore the correction to be performed when a minor error occurs is instructed. Therefore, the mute flag is set to “small” in S507. In addition, if the head mute flag is set, it indicates the occurrence of an error in the first subframe or the second subframe in which a pop sound is likely to occur, so in order to instruct the execution of the mute process, In step S507, the mute flag is set to “small”. In step S508, 1 is added to the delay counter, and the process ends.

  If the delay counter is not “0” in S506, it indicates that an error has occurred in a previously received transmission frame, and therefore correction using the error counter integrated value as a correction value is delayed. ing. Accordingly, the audio data correction is not newly set, and in order to continue the previous state, S507 is not executed (the mute flag is not changed), and S508 is executed and the process ends. If the head mute flag is not set, it indicates that the subframe in which the error has occurred is a subframe located after the subframe in which the pop sound is likely to be generated. Therefore, S508 is not executed. To exit.

  If the delay counter is larger than the predetermined threshold (“6”) in S504, “1” is added to the error counter in S509. Then, in S510, it is determined whether or not the error counter is greater than “7”.

  In the present embodiment, since the ADPCM word is 4-bit data, the maximum value of the error counter is 7. Therefore, if the error counter is larger than 7, “7” is substituted into the error counter in S511 and the process is terminated. If it is determined in S510 that the error counter is 7 or less, the processing is terminated as it is. Here, the upper limit value of the error counter is set.

  If it is determined in S502 that the error value is greater than the minor threshold, it is determined in S512 whether or not the error value is greater than or equal to the severe threshold. That is, it is determined from the error value whether the detected error is severe. In the present embodiment, it is determined that the error is severe as the severe threshold “3”.

  If the error value is less than the severe threshold value, it is determined in S513 whether the error counter is 0 or not. This is to determine whether or not an error has occurred in a previously received transmission frame. If the error counter is “0”, the mute flag is set to “small” in S514. In S515, “1” is added to the error counter. That is, although the degree of error is medium, there is no error in the previously received transmission frame, so “small” is set as the mute flag. Then, the process proceeds to S510.

  If it is determined in S512 that the error value is greater than or equal to the severe threshold value, “large” is set as the mute flag in S516. Further, in S517, “3” which is larger than the case of light or medium is added to the error counter. Then, the process proceeds to S510.

  In this way, in the error counter process, the error value is used as a value for measuring the degree of weight of the error that has occurred in the received transmission frame, and the error counter is added according to each of the mild, medium, and severe errors. Processing is in progress.

  Here, the error counter will be described with reference to FIG. The error counter is a counter that calculates an error occurrence frequency by accumulating values according to error values. For example, if the error is minor or moderate, “1” is added to the error counter in S509 or S515 of the error counter process (see FIG. 7). That is, when an error corresponding to an error value “1” or “2” occurs in one transmission frame, the error counter is incremented by “1” (indicated as a broken line S1 in FIG. 8). .) However, when a severe error corresponding to the error value “3” has occurred in the received transmission frame, the error counter increases by “3” at a time (in FIG. 8, the broken line S2). As shown).

  However, if the error is minor, the error counter is incremented by this error value when the delay counter becomes larger than “6”. While the delay counter is “6” or less, “1” is set in the error counter in S505, and when the delay counter becomes “7” or more, addition by the error value is performed.

  Returning to FIG. 3, when the error counter process is completed in S500, the voice processing unit 35 (see FIG. 1) performs the received voice data process (S600). This received audio data processing will be described with reference to FIG.

  In the received voice data processing shown in FIG. 9, the number of ADPCM words, which are voice data input from the received frame separation unit 32 in S601, is acquired. For example, if the audio data from the first subframe to the fourth subframe is 320 bits, the number of ADPCM words is 80 words.

  In S602, it is determined whether or not the error counter delivered from the error determination unit 34 is zero. When the error counter is “0”, it indicates that no error has occurred. Therefore, the audio data (ADPCM word) from the first subframe to the fourth subframe is not corrected, and ADPCM decoding is performed as it is. Output to the conversion unit 36, and the received audio data processing is terminated.

  If the error counter is not “0” in S602, it is determined in S603 whether or not the error counter is the maximum value “7”. If the error counter is not “7”, the initial value “0” is substituted into the variable j in S604.

  In S605, it is determined whether or not “Large” is set in the mute flag. This mute flag is a value set in S507, S514, and S517 of the error counter process (see FIG. 7).

  If “large” is not set in the mute flag, it is next determined in step S606 whether or not “small” is set. If “small” is set in the mute flag, the number of samples for the mute flag “small” set in the audio processing unit 35 is acquired in S607. In the present embodiment, the number of samples for the mute flag “small” is “8”.

  In S608, replacement is performed by writing “15” (“F” in hexadecimal) indicating mute to the ADPCM word corresponding to the number of samples (8 words) from the beginning of the audio data. That is, a mute process is performed in which all 32 bits from the first bit of the first subframe are set to “1”. By doing so, the volume of the received transmission frame is attenuated during the time of 8 words. This is because a pop sound is likely to occur near the beginning of the first subframe, so that the volume of the sound only for a short time when the error weighting degree is light or medium in the error counter process (see FIG. 7). By setting the to mute state, it is possible to reduce the influence on the sound quality while suppressing the generation of the popping sound.

  In the present embodiment, the mute flag is set to “small” when the error weight degree is light or medium, but the degree of correction by the mute process may be divided into light and medium. Good. When the error weight degree is increased from three stages, the number of samples to be subjected to mute processing may be increased according to the number of stages.

  In step S609, the number of samples for the mute flag “small” is substituted into the variable j. In the present embodiment, since the number of samples for the mute flag “small” is “8”, “8” is substituted into the variable j.

  If “small” is not set in the mute flag in S606, the process proceeds to the next without setting mute.

  If “large” is set in the mute flag in S605, the number of samples for the mute flag “large” set in the audio processing unit 35 is acquired in S610. In the present embodiment, the number of samples for the mute flag “large” is “16”.

  In S611, replacement is performed by writing “15” (“F” in hexadecimal) indicating mute in the ADPCM word corresponding to the number of samples from the beginning of the audio data. That is, a mute process is performed in which all 64 bits from the first bit of the first subframe are set to “1”. By doing so, the sound of the received transmission frame is attenuated during the time of 16 words, which is twice that when the mute flag is “low”. This is because, in the vicinity of the first bit of the first subframe, a popping sound is likely to occur, and since the error weight is severe, the degree of the popping sound is higher than that of light or medium. . Therefore, the generation of a popping sound can be more effectively suppressed by attenuating the volume over a long period of time.

  In step S612, the number of samples for the mute flag “large” is substituted into the variable j. In the present embodiment, since the number of samples for the mute flag “large” is “16”, “16” is substituted into the variable j.

In this way, if “small” or “large” is set in the mute flag, the volume is attenuated for the length of time corresponding to the set value, so that the error weighting degree is adjusted. Correction can be performed. Accordingly, it is possible to prevent the generation of a pop sound according to the error weighting degree , and to suppress a decrease in voice quality when a reception error occurs.

  In S613, it is determined whether or not the variable j is divisible by the correction interval value notified from the error determination unit 34. The variable j is “8” when the process proceeds from S609 to S613, and the variable j is “16” when the process proceeds from S612 to S613. For example, when only a synchronization error has occurred in the received transmission frame, the correction interval value is set to “4” in S412 of the reception error process (see FIG. 4), and therefore it is divisible by the variable j. Therefore, the process after S616 is performed.

  If the variable j is not divisible by the correction interval value, the process proceeds to S614, and “1” is added to the variable j. Then, in S615, it is determined whether or not the value indicated by the variable j is greater than the number of ADPCM words (80 words in the present embodiment). If the variable j is larger, it indicates that the processing for all ADPCM words has been completed, so the received voice data processing is terminated.

  If the variable j is equal to or less than the number of ADPCM words, the process proceeds to S613, and it is determined again whether the variable j is divisible by the correction interval value.

  If it is determined in S613 that it is divisible, the first bit (MSB) in the ADPCM word is extracted in S616. The ADPCM word at this time is the variable j-th from the voice data. That is, when the correction interval value is “4” and the process proceeds from S606 in which the mute process is not performed, first, the 0th ADPCM word (the first word of the first subframe) is targeted, and the next The fourth ADPCM word is the target.

  If the process proceeds from S609 in which the mute processing for the mute flag “small” has been performed, first, the eighth (9th word of the first subframe) ADPCM word is targeted, and then the twelfth ADPCM word is targeted. Become.

  For example, if the correction interval value is “2” and the process proceeds from S612 in which the mute process for the mute flag “large” is performed, the 16th ADPCM word is first processed, and then the 18th ADPCM word is processed. Is the target.

  In S617, it is determined whether or not the first bit of the extracted ADPCM word is “0”. Since the ADPCM word is represented by a signed binary number, if the first bit is “0”, it indicates that a positive number, that is, a difference value is added. If the first bit of the ADPCM word is “1”, it means that a negative number, that is, a difference value is subtracted.

  In the processing of S618 and S621, the error weight value is subtracted or added from the ADPCM word value according to the sign of “0” or “1” of the first bit. If the sign of the ADPCM word is “0” (addition of the difference value), the error counter value is subtracted from the ADPCM word value in S618, and processing in a direction to suppress the change in the value after decoding is performed. On the other hand, if the sign of the ADPCM word is “1” (subtraction of the difference value), the error counter value is added to the ADPCM word value in S621, and in this case also, the process is directed to suppress the change in the value after decoding. . Thus, the error counter value corresponding to the error weight is subtracted or added according to the sign of the ADPCM word.

  After S618, it is determined in S619 whether the ADPCM word is larger than “15” (hexadecimal “F”). If it is greater than “15”, the ADPCM word is set to “15” (hexadecimal “F”) in S620. After S621, it is determined in S622 whether the ADPCM word is smaller than “0”. If it is smaller than “0”, the ADPCM word is set to “0” in S623. In step S614, 1 is added to the variable j, and in step S615, it is determined whether or not the processing for all ADPCM words is completed. If the processing is completed, the received voice data processing is terminated.

  If the value of the error counter is “7” in S603, the error occurrence frequency is high and the error is severely errored. In such a state, the sound volume is attenuated by writing “15” (“F” in hexadecimal) indicating mute in all ADPCM words in S624.

  Thus, based on the error value set according to the error weight in the reception error processing (see FIG. 4) or subframe processing (see FIG. 5), the error counter value added for each error occurrence is The difference value input to the ADPCM decoding unit 36 as a correction value is converted into a small value according to the correction value, and the change width of the PCM data after decoding can be suppressed small. Therefore, since the audio data can be corrected according to the error weight, it is possible to cope with frequent occurrence of errors in a plurality of transmission frames. Accordingly, it is possible to cope with a reception error due to radio wave interference or the like, it is possible to prevent the occurrence of a popping sound, and it is possible to minimize a decrease in voice quality when a reception error occurs.

  Further, the correction in the received audio data (ADPCM word) sequence is performed by the correction interval value set in accordance with the error weighting degree in the reception error processing (see FIG. 4) and the subframe processing (see FIG. 5). By changing the interval of the time position for performing the correction, correction is performed at a frequency according to the degree of error, and excessive correction (abrupt change in output PCM data) can be prevented, so that the voice quality is reduced. Can be suppressed.

  Returning to FIG. 3, if no error has occurred in the determination by the error determination unit 34 in S300, the process proceeds to S700 to execute a reception error process.

  As shown in FIG. 10, in the reception error free process, first, it is determined whether or not the error counter is “0” (S701). If the error counter is not “0”, the subtraction value set in the error determination unit 34 is acquired in S702. In the present embodiment, this subtraction value is “3”. In step S703, the subtraction value “3” is subtracted from the error counter. In other words, when no error has occurred, the degree of correction is reduced by subtracting an error counter that is the degree to which the audio data is corrected. For example, when no error has occurred in the transmission frame, the error counter may be set to “0” in order to quickly recover the volume. However, depending on the audio data, there is a risk that a popping sound may be generated due to rapid volume recovery. In this embodiment, the error counter is subtracted by a predetermined value in order to suppress the occurrence of a pop sound while rapidly recovering the volume.

  In S704, it is determined whether or not the subtracted error counter is smaller than “0”. When the error counter is smaller than “0”, the error counter is set to “0” in S705.

  If the error counter is equal to or greater than “0”, the correction interval value is set to “4” in S706.

  For example, when there is no synchronization error and no control error, and an error has occurred in a subframe exceeding the subframe error threshold “3”, the correction interval value is “2” (No in FIG. 6). .9). Therefore, in this case, every two words of ADPCM words are corrected. However, by receiving once a transmission frame in which no error occurs (all subframes have no error), the correction interval is expanded from 2 words to 4 words as shown in FIG. 10 (S706), and the frequency of correction Lower. In step S707, the delay counter is set to “0”. By setting the delay counter to “0”, even when an error is detected in a subsequent transmission frame, information is maintained that it is not a continuous error following the previous transmission frame.

  If no error occurs in this way, the value of the error counter is decreased or the correction interval value is increased. If the situation improves from the state of error occurrence and there are consecutive frames without error from a certain time, the error counter value increased at the time of error occurrence is lowered, and the correction interval value is increased, so that the correction is relaxed. That is, in S504 of FIG. 7, when the delay counter is “0” (below the threshold), the error counter that determines the degree of correction is set to a relatively low value “1”, and the correction interval value is also relatively large “ 4 ". These values are maintained in a range where the delay counter is equal to or smaller than the threshold value, and the error counter is not counted up. Even if the situation deteriorates from the error-free state and errors continue to occur, the error counter is delayed until the delay counter reaches the threshold value. If an error-free transmission frame continues, the error counter is decremented to become an error counter “0” (S705), and the recovery of voice quality can be accelerated.

  When no reception error has occurred in this way, the degree of correction can be reduced by decreasing the value of the error counter by the process shown in FIG.

  As mentioned above, although embodiment of this invention has been described, this invention is not limited to the said embodiment. For example, in the present embodiment, the audio data in the transmission frame is divided into four subframes from the first subframe to the fourth subframe, but the number of divisions can be determined as appropriate. In that case, the number of subframe error thresholds is also changed according to the number of divided subframes. Note that the other threshold values and set values according to the present embodiment are examples, and other values may be selected as long as the magnitude relations of the related numerical values are maintained.

  Since the present invention can cope with a reception error due to radio wave interference or the like, it is possible to prevent the generation of a pop sound and to minimize a decrease in voice quality when a reception error occurs. This is suitable for an audio processing apparatus and an audio processing method of an ADPCM audio transmission system for transmitting quantized audio data.

DESCRIPTION OF SYMBOLS 1 Cordless telephone 2 Master unit 3 Slave unit 20 Transmitter 21 Telephone control unit 22 A / D converter 23 Quantization compression unit 24 ADPCM encoding unit 25 Transmission frame generation unit 26 Radio transmission unit 26a Antenna 30 Reception unit 31 Radio reception unit 31a Antenna 32 Reception frame separation unit 33 Error detection unit 34 Error determination unit 35 Audio processing unit 36 ADPCM decoding unit 37 Inverse quantization expansion unit 38 D / A converter 39 Amplification unit 39a Speaker

Claims (10)

An audio processing device of an ADPCM audio transmission system that transmits audio data differentially quantized by an ADPCM method,
A transmission frame is composed of a plurality of subframes.
An error detection unit that individually detects the presence or absence of an error for the plurality of subframes of the received transmission frame;
A voice processing unit that executes correction processing when an error is detected by the error detection unit;
Have
If the occurrence frequency of the subframe error determined by the error detection unit is less than the threshold, the audio processing unit performs correction at a first time interval, and the occurrence frequency of the subframe error is equal to or greater than the threshold. If there is, perform correction at the second time interval,
The second time interval is shorter than the first time interval;
A voice processing device for an ADPCM voice transmission system. The error detection unit first determines whether an error has occurred in the synchronization word, then determines whether an error has occurred in the control data, and then determines from the first subframe to the fourth subframe. The voice processing apparatus of the ADPCM voice transmission system according to claim 1, wherein it is determined whether or not an error has occurred in the voice data. When the occurrence frequency of the subframe error determined by the error detection unit is less than a threshold, the audio processing unit performs correction according to the error weighting degree indicated by the first value, and the occurrence of the subframe error If the frequency is greater than or equal to the threshold, correction is performed according to the error weighting indicated by the second value,
The error weighting degree indicated by the second value is more severe than the error weighting degree indicated by the first value.
The voice processing apparatus of the ADPCM voice transmission system according to claim 1. The voice processing unit suppresses a change in the value after ADPCM decoding by subtracting or adding a value corresponding to the error weight from the voice data in accordance with a code accompanying the voice data which is a difference value of ADPCM. Process direction,
The voice processing apparatus of the ADPCM voice transmission system according to claim 3 . The error detection unit detects an error in a synchronization word, control data, or audio data constituting the transmission frame,
The speech processing unit makes the speech data error heavier than the synchronization word and control data as the error weight.
The voice processing apparatus of the ADPCM voice transmission system according to claim 3 or 4 . The audio processing unit calculates an occurrence frequency of the subframe error by performing integration based on a value corresponding to the degree of weight of the error, and uses the correction value based on the occurrence frequency of the subframe error to calculate the audio data Correct,
The voice processing apparatus of the ADPCM voice transmission system according to any one of claims 3 to 5 . The speech processing unit corrects the speech data with a correction value according to the mildness when the error weight is mild, and performs correction with a correction value based on the occurrence frequency of the subframe error. Delay after receiving a predetermined number of transmission frames including
The voice processing apparatus of the ADPCM voice transmission system according to claim 6 . The voice processing unit counts the number of subframes in which an error occurs in one transmission frame, and sets an error value or a correction interval value according to the counted number of subframe errors.
The voice processing apparatus of the ADPCM voice transmission system according to claim 3 . An audio processing method for an ADPCM audio transmission system for transmitting audio data differentially quantized by an ADPCM method,
A transmission frame is composed of a plurality of subframes.
Determining the frequency of occurrence of subframe errors by detecting the presence or absence of errors individually for the plurality of subframes;
If the occurrence frequency of the subframe error is less than a threshold, perform correction at the first time interval;
If the occurrence frequency of the subframe error is equal to or greater than a threshold, correction is performed at a second time interval,
The second time interval is shorter than the first time interval;
Audio processing method for ADPCM audio transmission system. In the process of detecting the presence or absence of the error,
First, determine if there was an error in the sync word,
Next, determine whether an error has occurred in the control data,
10. The voice processing method of the ADPCM voice transmission system according to claim 9 , wherein it is determined whether an error has occurred in the voice data of the first subframe to the fourth subframe.

Images