JPH0281538A - Voice packet processing system - Google Patents

Abstract

PURPOSE:To realize talking voice with high quality by discriminating whether a section without arrival of a voice packet at the reception side is due to silent compression or due to packet abort. CONSTITUTION:A voice packet unarrival section at the reception side is discriminated as to whether or not a section in which the voice packet is to be reached substantially by monitoring at the receiver side a mark bit 202 or a sequence number 203 added to the voice packet transferred from the sender side. Moreover, the duration time of the section is measured by monitoring the arrival of the first voice packet after the production of a voice packet unarrival section by means of a timer 22. In this case, a threshold level distinguishing the silence compression section and the packet abort section in terms of timewise length is decided in advance and the duration time and the threshold level are compared. As the result of comparison, if the duration time is longer than the threshold level, it is discriminated to be silence compression section and when shorter, it is discriminated to be a packet abort section. Thus, the quality of talking voice is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to voice packet communication, and particularly to a voice packet processing method for realizing high-quality voice communication.

[Conventional technology]

In voice packet communication, in order to improve the efficiency of using a single line, a voice detector on the transmitting side distinguishes the voice signal into a sound part and a silent part, and audio information in the section determined to be a silent part is not transferred. There is.

Therefore, the audio packet in the section determined to be a silent section does not arrive at the receiving side. In addition, in voice packet communication, the buffer that temporarily stores packets in the packet switching network overflows under high traffic conditions, etc.
Packets may be discarded within the switching network, and voice packets in an interval determined to be active on the transmitting side may not arrive at the receiving side. Furthermore, in order to guarantee the real-time nature of one call voice even when packets arrive at the receiving side, packets exceeding a predetermined allowable delay time are discarded at the receiving side. In this case as well, there are sections where no audio signal actually arrives. In the above case, for example, JP-A-55-
As described in Japanese Patent Laid-open No. 21610 or Japanese Patent Application Laid-open No. 59-79658, there is a technique in which the receiving side creates pseudo noise by itself and interpolates the voice packet non-arrival period (
(hereinafter referred to as silent playback technology) is adopted.

[Problem to be solved by the invention]

However, the above-mentioned conventional silence reproduction technology is suitable for reproducing a silent part consisting of an audio signal with low audio power and a relatively monotonous audio waveform. Since the system does not distinguish between packet discards and interpolates intervals where voice packets have not arrived using pseudo-noise created by the system itself, there are problems in the following points.

If the voice packet non-arrival period is due to packet discard, the non-arrival period exists within the talkspurt. This talk spurt consists of an audio signal with high audio power and a complex audio waveform.For example, as shown in Figure 2, the pseudo noise created by the receiving side and the audio signal in the talk spurt are different. , the magnitude of the audio power and the complexity of the audio waveform differ greatly. These differences in the magnitude of the audio power and the complexity of the audio waveform cause interruptions, plosive sounds, etc. in the reproduced audio, and as a result, the quality of the call audio deteriorates significantly.

An object of the present invention is to identify the reason why a packet has not arrived on the receiving side in a system that transmits digital voice using a packet communication method, and to improve the quality of voice communication by applying an optimal interpolation technique to this voice packet. The goal is to realize a processing method.

[Means to solve the problem]

In order to achieve the above object, the present invention appropriately determines whether the period in which audio packets do not arrive at the receiving side is due to silence compression that does not transfer silent parts or due to packet discard, and based on the determination result, It is characterized by performing silent reproduction processing or discard compensation processing.

1 to determine the cause of the above voice packet non-arrival section.
In the present invention, there is a means for adding a mark bit to a voice packet to be transferred on the transmitting side, and a means for monitoring the mark bit of an arriving packet on the receiving side and determining a section in which a voice packet does not arrive based on the monitoring result. will be established. In addition, as another method for determining the cause of the voice packet non-arrival period, the present invention includes means for assigning a sequence number to a voice packet to be transferred on the transmitting side, and monitoring the sequence number of the arriving packet on the receiving side. Then, means is provided for determining the section in which voice packets do not arrive based on the monitoring results.

Still another method for determining the cause of the voice packet non-arrival period is to provide means for monitoring the arrival of voice packets using a timer on the receiving side, and means for determining the period in which voice packets do not arrive based on the monitoring result. It's okay.

[Effect]

According to the present invention, by monitoring the mark bits or sequence numbers attached to voice packets transferred on the transmitting side on the receiving side, it is possible to tell the receiving side that the interval in which the voice packet has not arrived is the interval in which the voice packet should have arrived. It can be determined whether or not there is.

Therefore, if the voice packet non-arrival period is a period in which a voice packet should originally arrive, the voice packet non-arrival period is determined to be a packet discard period, and if it is a period in which a voice packet does not arrive originally, the voice packet The unarrived section is determined to be a silent compression section.

Furthermore, by monitoring the arrival of the first voice packet after the occurrence of the voice packet non-arrival interval using a timer, it is possible to measure the duration of the interval. In this case, a threshold value for distinguishing a silence compression section and a packet discarding section by time length is determined in advance, and the duration time and the threshold value are compared.

As a result of the comparison, if the duration is longer than the threshold, it is determined to be a silence compression section, and if it is shorter, it is determined to be a packet discard section.

〔Example〕

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

FIG. 1 shows an embodiment in which a mark bit is used to determine an interval in which a voice packet has not arrived. 1st
In the figure, an analog audio signal input from an audio terminal 1 is converted into a digital signal by an A/D converter 2.

The digital signal is converted into an AD signal by a 32 kbps adaptive differential pulse code conversion (hereinafter abbreviated as ADPCM) encoder 3.
After being encoded into a PCM signal, the packet assembly unit 5 assembles it into a voice packet. In the packet assembling section 5, the digital signal is first divided into fixed length blocks by a signal dividing section 15. The signal dividing section 15 includes a voice detector 4
Only when the determination 101 is a voice determination, a timestamp, which is time information indicating the generation time of the block, is added to the block and sent to the mark bit adding unit 16. The mark bit adding section 16 performs the judgment 1 of the voice detector 4.
01, mark bits consisting of pits are given to the block 2.

The time stamp and mark bit are provided as shown in FIG. That is, time stamps are continuously assigned to the audio block 200, regardless of whether the audio block 200 is voiced or silent. The signal dividing section 15 adds a time stamp assigned to the block only to the block for which the judgment 101 of the voice detector 4 indicates that there is a sound, and sends the block to the mark bit adding section 16 . Therefore, as shown at 201 in FIG. 3, in the mark bit assigning section 16, the timestamp assigned to the block determined to be silent becomes a missing number. The mark bit adding unit 16 monitors the judgment 101 of the voice detector 4 and adds +i 1 u to the first bit of the mark bit area of the first sound block after changing from the silence judgment to the presence judgment. and sends it to the packet header adding section 17. Further, the city record mark/vic addition unit 16 sets 6'1'' in the second bit of the mark bit area of the block of the final voice determination when the determination 101 changes from the voice determination to the silence determination, and The mark bit of the sound block sent to the packet header addition unit 17 is sent to the header adding unit 17. Therefore, as shown in 202 in FIG. is 1” in the order of the 1st bit and the 2nd bit.
0”, and the final block is “0” 411 ?+
The other blocks are "0" and "O".

The packet header adding section 17 adds information such as a destination address to the sound block to create a voice packet, and sends this to the packet transmitting section 6.

FIG. 4 shows the format of the audio packet in this embodiment. As shown in FIG. 4, the audio packet 300 includes a packet header 301° mark bit 3
02. Timestamp 303 and audio data 304
It consists of The packet ladder 301 is a 4-byte area containing information such as a destination address. The mark bit 302 includes the first bit 302 (a) and the second bit 30
2 (b). The timestamp 303 has a 6-bit configuration and is assigned a numerical value from 0 to 63. The audio data 304 is composed of ADPCM signals of blocks determined to be audible, and has a data length of 31 bytes.

The voice packet is sent from the packet transmitter 6 to the packet switching network via the packet switching device 7.

Voice packets arriving at the packet receiving unit 8 from the packet switching network via the packet switch 7 are sent to the determining unit 9. In the determination unit 9, first, the delay jitter control unit 18 adjusts the variations in transmission delay time of each audio packet.

The processing in the delay jitter control section 18 will be explained using FIG. In the delay jitter control section 18.

A jitter control count value 403 is counted up. The jitter control count value 403 starts counting up 50 milliseconds after the sending time stamp 401 starts counting up. The jitter control count value 403
is an integer value between 0 and 63 like the timestamp 401, and is incremented by 1 every 25 milliseconds, which is the interval between voice packet occurrences. The audio packet is temporarily stored in the delay jitter control section 18, and the numerical value of the time stamp is read out. The delay jitter control section 18 includes:
The jitter control count value and the time stamp are compared, and if the time stamp value is smaller than the jitter control count value, as in packet 3 in FIG. 5, the audio packet is discarded. Therefore, the audio packet 3 whose delay time exceeds 75 milliseconds is discarded by the jitter control unit 18,
In the time stamp 404 of the arriving packet, number 3 is missing. In cases other than the above, the voice packet is not discarded, but is sent to the packet arrival monitoring unit 19 every 25 milliseconds.

Returning to FIG. 1, the packet arrival monitoring unit 19 issues an audio packet arrival notification 106 to the determination processing control unit 21 and sends the audio packet to the mark bit monitoring unit 20. After reading the mark bit of the audio packet, the mark bit monitoring unit 20 sends the audio packet to the packet disassembly unit 10. If the packet arrival monitoring unit 19 does not confirm the arrival of the audio packet, the packet arrival monitoring unit 19 sends the mark bit of the audio packet that arrived immediately before the section in which the audio packet did not arrive to the determination processing control unit 21. , and the instruction 107 to search for the mark bit of the next arriving audio packet.The 1 determination processing control unit 21 searches for the mark bit of the audio packet read out by the mark bit monitoring unit 20, and searches for the mark bit of the audio packet that arrives next. It is determined whether the reason why the packet does not arrive is due to silence compression, which deletes silent parts on the transmitting side, or because voice packets are discarded.

The procedure for the first determination will be explained below. When a section in which a voice packet does not arrive occurs, the determination processing control unit 21 performs a search 108 for the mark bit of the voice packet that arrived just before.If the first bit of the mark bit of the voice packet is II
If the second bit of QTt is "1", that is, if the voice packet is the last packet of the sound part, as shown at 202 in FIG.
determines that the period in which the voice packet has not arrived is a silence compression period. On the other hand, if the first bit of the mark bit of the voice packet is #OII and the second bit is "0", that is, as shown at 202 in FIG. or the first bit of the mark bit of the voice packet is “1”.
, if the second bit is "0", that is, if the audio packet is the first packet of the sound part, as shown at 202 in FIG. The arrival section is determined to be a packet discard section.

Further, the determination processing control unit 21 issues an instruction 109 to start the timer 22 when a packet discard period occurs,
If there is no packet arrival notification 106 within 75 milliseconds, the section from 75 milliseconds until the arrival of the next audio packet is determined to be a silence compression section. Next, the determination processing control unit 21 performs a search 108 for the mark bit of the first audio packet that arrived after the audio packet non-arrival period. The first bit of the mark bit of the audio packet is “1” and the second bit is “1”.
If the bit is “0”, that is, 202 in FIG.
If the voice packet is the first packet of the sound part, as shown in FIG. If the first bit of the mark bit of the audio packet is "0" and the second bit is "0", that is, as shown at 202 in FIG. , or the first mark bit of the voice packet is II Q N
, if the second bit is "1", that is, if the voice packet is the last packet of the sound part, as shown at 202 in FIG. It is determined that the first packet of the sound part following the section has been discarded.

Further, the determination processing control unit 21 controls unfavorable regeneration and disposal compensation based on the above determination result.

If it is determined that the voice packet non-arrival interval is a silence compression interval, the determination processing control unit 21 issues an instruction 102 to the silence reproduction unit 12 to generate pseudo noise. Further, the determination processing control unit 21 issues an instruction 105 to the selector 14 to select the output of the silence reproduction unit 12 at the timing when the decoding of the audio packet that arrived immediately before the audio packet non-arrival section is completed.

On the other hand, if it is determined that the voice packet non-arrival interval is a packet discard interval, the 1 determination processing control unit 21 instructs the discard compensation unit 13 to compensate for the packet discard interval.
Roll a 3. A method of interpolating the packet discard section will be explained with reference to FIG. The discard compensation section 13 includes the discard compensation control section 2
3, an audio waveform matching section 24, and a storage section 25.

In FIG. 6, the output of the selector 14 is stored in the storage section 25. The storage section 25 has a capacity to store three packets of audio signals, and the stored contents are updated sequentially. The discard compensation control unit 23 is.

When the discard compensation instruction 103 is sent, the audio waveform matching unit 24 issues an audio signal creation instruction 110 for compensating for the packet discard section to the audio waveform matching unit 24. The signal 111 and the audio signal input to the storage section 25 up to the packet discard section are read, an audio signal matching the audio waveform immediately before the packet discard section is created, and output to the selector 14.

The determination processing control unit 21 issues an instruction 104 to the 32 kbps ADPCM decoder 11 to stop the decoding process in the packet discard section. The contents of the decoding process stop instruction 104 will be explained using FIG. 7. The ADPCM decoder 11 includes an adaptive inverse quantizer 26. jt childization factor controller 27. Adaptive predictor 28. It consists of an adder 29. A detailed description of the functions of each unit is provided in CCITT Recommendation 0.721, so the explanation will be omitted. In Figure 7,
The quantization factor controller 24 issues a decoding process stop instruction 10
4, the update of the quantization factor is stopped. Furthermore, the adaptive predictor 28 stops updating the internal parameters used for adaptive prediction in accordance with the decoding process stop instruction 104. The adaptive dequantizer 26 does not receive any input from the packet disassembly unit 10, so it does not output anything to the adder 29. Similarly, the adder 29 does not receive any input from the dequantizer 26, so it outputs nothing to the adder 29. , nothing is output to the selector 14. Therefore, in the packet discard period, the quantization factor in the ADPCM decoder 11.

Both updating of the internal parameters and output to the selector 14 are stopped.

The determination processing control section 21 issues an instruction 105 to the selector 14 to select the output of the discard compensation section 13 at the timing when the decoding of the voice packet that arrived immediately before the packet discard section is completed.

When a new audio packet arrives, the determination processing control unit 21 performs the following operations at the decoding start timing of the audio packet: 32
Further, the determination processing control section 21 issues an instruction 105 to the selector 14 to select the output of the kbps ADPCM decoder 11. It issues instructions 102 and 103 to the discard compensation unit 13 to stop each process, and cancels the decoding process stop instruction 104.

As shown in FIG. 8, the audio signal obtained by interpolating the packet discard section by the above-mentioned discard compensation process matches other audio signals in the talk spurt in audio power and audio waveform.

The selector 14 selects the specified output to be sent to the determination processing control section 21 and outputs it to the D/A converter 2. The selector output is fed back to the discard compensator 13 for compensating for discarded packets, and is also converted into an analog audio signal by the D/A converter 2 and output from the audio terminal 1.

As described above, according to this embodiment, by searching the mark bit on the receiving side, it is determined whether the section in which a voice packet does not arrive is a silence compression section or a packet discard section, and then silence reproduction and discard compensation are performed. Because I try to do this,
Can reproduce high-quality call audio.

FIG. 9 shows another embodiment of the present invention in which a sequence number is used to determine an interval in which a voice packet has not arrived. In FIG. 9, the audio signal encoded into an ADPCM signal by the 32 kbps ADPCM encoder 3 is divided into fixed length blocks by the signal dividing section 15. The signal dividing section 15 assigns a time stamp expressed in 4 bits to the block and sends it to the sequence number assigning section 30 only when the determination 101 of the voice detector 4 is a voice determination. The sequence number assigning unit 30 includes the signal dividing unit 15
A sequence number consisting of 4 bits is assigned to the block sent from the block.

The timestamp and sequence number are given in the format explained in FIG. Timestamps are continuously assigned to the audio block 200 regardless of whether the audio block 200 is voiced or silent. The signal dividing unit 15 assigns a time stamp assigned to the block only to a block in which the judgment 101 of the voice detector 4 indicates that there is a sound, and sends the block to the sequence number assigning unit 30 . Therefore, 2 in Figure 3
As shown in 01, the timestamp assigned to the block determined to be silent by the sequence number assigning unit 30 becomes a missing number. The sequence number assigning unit 30 assigns a sequence number to the sound block and sends it to the packet header assigning unit 17 . As shown at 203 in FIG. 3, since the sequence number is a number assigned only to a sound block, the packet header adding section 17 has no missing sequence numbers.

The packet header adding section 17 adds information such as a destination address to the sound block to create a voice packet, and sends it to the packet transmitting section 6. FIG. 11 shows the format of the audio packet in this embodiment.

In FIG.

The audio packet 305 includes a packet header 301° sequence number 306. It consists of a timestamp 303 and audio data 304. The packet header 301 is
This is a 4-byte area containing information such as the destination address. The sequence number 306 has a 4-bit configuration, and is from 0 to 15.
A numerical value up to is assigned. The timestamp 303 has a 4-bit configuration and is assigned a numerical value from 0 to 15. The audio data 304 is composed of AI) PCMPJ numbers of blocks determined to be audible, and has a data length of 31 bytes.

The voice bucket I-, which has arrived at the packet receiving section 8 from the packet switching network via the packet switch 7, is sent to the delay jitter control section 18. The function of the delay jitter control section 18 is as follows:
Since it is the same as the delay jitter control section shown in FIG. 1, the explanation will be omitted.

The packet arrival monitoring unit 19 sends the voice packet arrival notification 106 sent from the delay jitter control unit 18 to the determination processing control unit 21, and sends the voice packet to the sequence number monitoring unit 31. After reading the sequence number of the audio packet, the sequence number monitoring unit 31 sends the audio packet to the packet disassembly unit 10. If the packet arrival monitoring unit 19 does not confirm the arrival of the audio packet, the packet arrival monitoring unit 19 sends the sequence number of the audio packet that arrived immediately before the section in which the audio packet did not arrive to the determination processing control unit 21; and an instruction 107 to search for the sequence number of the next arriving voice packet. The determination processing control unit 21 performs a search 112 for the sequence number of the audio packet read out by the sequence number monitoring unit 24, and determines whether the reason why the audio packet did not arrive is due to silence compression in which silent parts are deleted on the transmitting side. It is determined whether this is due to voice packets being discarded.

The determination procedure will be explained below. When a section in which no voice packet arrives occurs, the 1 determination processing control unit 21 searches 112 for the sequence number of the voice packet that arrived immediately before, and stores the sequence number. Further, the determination processing control unit 21 issues an instruction 109 to start the timer 22. 75 from the time when the voice packet non-arrival interval occurs
If the next audio packet arrives within milliseconds, the 1-determination processing control unit 21 searches 112 for the sequence number of the audio packet and compares it with the sequence number stored in the determination processing control unit 21 . If there is no missing number between the sequence number of the newly arrived voice packet and the stored sequence number, the determination processing control unit 21
determines that the audio packet non-arrival interval is a silence compression interval.

Conversely, if there is a missing number, it is determined that the audio packet corresponding to the missing number has been discarded. On the other hand, if there is no packet arrival notification 106 from the packet arrival monitoring unit 19 within 75 milliseconds, the 1 determination processing control unit 21 determines the interval up to 75 milliseconds as a packet discard interval, and after 75 milliseconds, the next The section until the voice packet arrives is determined to be a silence compression section.

Furthermore, the 1 judgment processing control section 21 searches 112 for the sequence number of the voice packet that arrived next, and compares it with the sequence number stored in the judgment processing control section 21 . If there is no missing number between the sequence number of the newly arrived audio packet and the stored sequence number, the determination processing control unit 21 determines that the packet at the beginning of the sound part following the silent compression section has not been discarded. I judge that. If there is a missing number, the 1 determination processing control unit 21 determines that the audio packet at the beginning of the sound part following the silent compression section and the audio packet corresponding to the missing number have been discarded.

Based on the above determination <32 kbps ADPCM decoder 11
, silent playback section 12. Disposal Compensation Department 13. and selector 1
The explanation of control No. 4 is the same as that of the embodiment shown in FIG. 1, so the explanation will be omitted.

As described above, in the case of this embodiment as well, by searching the sequence number on the receiving side, it is determined whether the section in which a voice packet does not arrive is a silence compression section or a packet discard section, and then silence reproduction and discard compensation are performed. This makes it possible to achieve high-quality call audio.

FIG. 10 shows still another embodiment of the present invention in which the determination of the interval in which a voice packet does not arrive is performed without using additional information such as mark bits and sequence numbers. In FIG. 10, the audio signal encoded into an ADPCM signal by the 32 kbps ADPCM encoder 3 is divided into fixed-length blocks by the signal division section 15. The signal dividing section 15 adds a time stamp expressed in 8 bits to the block and sends it to the packet header adding section 17 only when the judgment 101 of the voice detector 4 is that there is a sound.

The time stamp assignment method is the same as in the embodiments shown in FIGS. 1 and 9.

The packet header adding section 17 adds information such as a destination address to the sound block to create a voice packet, and sends it to the packet transmitting section 6. FIG. 12 shows the format of audio buckets 1 to 1 in this embodiment. In FIG.
．． It consists of a timestamp 303 and audio data 304. The packet header 301 is a 4-byte area containing information such as a destination address. timestamp 303
has a 1-byte structure, and is assigned a numerical value from 0 to 255. The audio data 304 is composed of ADPCM signals of blocks determined to be audible, and has a data length of 31 bytes.

Voice packets arriving at the packet receiving unit 8 from the packet switching network via the packet switch 7 are sent to the delay jitter control unit 18. The delay jitter control section 18 is the same as the delay jitter control section in the embodiment shown in FIG. 1, so a description thereof will be omitted.

The packet arrival monitoring unit 19 sends the voice packet arrival notification 106 sent from the delay jitter control unit 18 to the determination processing control unit 21, and divides the voice packet into the packet division unit 1.
Send to 0. If the packet arrival monitoring unit 19 does not confirm the arrival of a voice packet, the 1 determination processing control unit 21 issues an instruction 109 to start the timer 22. If the next voice packet arrives within 75 milliseconds from the time when the arrival of the voice packet was not confirmed 1
The determination processing control unit 21 determines the interval until the voice packet arrives as a packet discard interval. On the other hand, if there is no packet arrival notification 106 from the packet arrival monitoring unit 19 within 75 milliseconds, the determination processing control unit 21 determines the interval up to 75 milliseconds as a packet discard interval, and after 75 milliseconds, the next The section until the voice packet arrives is determined to be a silence compression section. In this embodiment, the 1-determination processing control unit 21 determines that at least the first packet of the voice packets in the sound portion following the silent compression section has not been discarded.

Based on the above determination <32 kbps ADPCM decoder 11
, silent playback section 12. Discard compensation section 13 and selector 1
The explanation of the control in step 4 is omitted because it overlaps with the explanation in the embodiment shown in FIG.

According to this embodiment, by monitoring the arrival of the first voice packet after the occurrence of a period in which no voice packet arrives, it is determined whether the voice packet non-arrival period is a silence compression period or a packet discard period. Therefore, there is no overhead of adding additional information, and since silent playback and discard compensation can be performed based on the determination, high-quality call audio can be realized.

Although the present invention has been described in detail above, the order of the audio packet format in the present invention is not limited to that shown in the embodiments, and the length of each field is also determined by the system or industry standard. However, this does not limit the present invention. Furthermore, the time for delay jitter control of voice packets and the storage capacity for discard compensation, etc. explained in the operation of the embodiment are examples for ensuring voice quality, and are not limited thereto. do not have.

In this embodiment, the encoder and decoder are equipped with 32 kbps ADP.
CM was used, but 64kbps PCM, 16kbps
This method is also effective in audio packet processing methods that use other codes and decoding methods such as APC-AB.

〔Effect of the invention〕

As is clear from the above description, according to the present invention, by determining whether the period in which audio packets do not arrive at the receiving side is due to silence compression or packet discard,
Appropriate silent playback or packet compensation can be performed, and high-quality call audio can be achieved.

[Brief explanation of the drawing]

1, 9, and 10 are system configuration diagrams showing embodiments of the present invention, FIG. 2 is a diagram showing the audio signal waveform when the packet discard section is interpolated according to the conventional example, and FIG. 3 is the system configuration diagram showing the embodiment of the present invention. Figures 4 and 11 are diagrams for explaining an example of a method for assigning timestamps, mark bits, and sequence numbers.
12 are explanatory diagrams of a packet format in an embodiment of the present invention, FIG. 5 is an explanatory diagram of a delay jitter control method in an embodiment of the present invention, and FIG. 6 is an explanatory diagram of an embodiment of a discard compensator. An explanatory diagram, FIG. 7 is an explanatory diagram of the control method of the ADPCM decoder in the packet discard section according to an embodiment of the present invention, and FIG. 8 is an explanatory diagram of an audio signal when the packet discard section is interpolated according to an embodiment of the present invention. It is a waveform. 4...Audio detector, 5...Packet assembly section, 9...
・Packet determination unit, 12... Silence playback unit, 13...
Discard compensation unit, 14... Selector, 15... Signal division unit, 16... Mark bit addition unit, 17... Packet header addition unit, 18... Delay jitter control unit, 19.
... Packet arrival monitoring unit, 20... Mark bit monitoring unit, 21... Judgment processing control unit, 22... Timer, 2
3... Discard compensation control unit, 24... Audio waveform matching unit,
25...Storage unit, 26...Adaptive inverse quantizer, 27
. . . Keiko controller, 28. Adaptive predictor, 29. Adder, 30. Sequence number assigning unit.
31...Sequence i″1″°
6

Claims (1)

[Claims] 1. Digitize an analog audio signal, divide it into blocks of a certain length, determine whether each block is sound or not, create audio packets based on the results, and send them to a transmission path; In audio packet processing methods that reproduce analog audio signals from audio packets that arrive at the receiving side,
When an interval in which a voice packet does not arrive at the receiving side occurs, a means for determining whether the interval is due to silence compression or discarding of the voice packet, and a means for determining whether the interval is determined to be a silence compression interval. 1. An audio packet processing method, characterized in that a silent reproduction is performed and means is provided for compensating for discarded packets in a section determined to be an audio packet discard section. 2. In claim 1, there is provided a means for adding additional information for performing the determination process to an audio packet (hereinafter simply referred to as an audio packet) in an interval determined to be a sound by the transmitting side, An audio packet processing method characterized in that the determination process is performed based on additional information. 3. In claim 2, bits (hereinafter referred to as mark bits) indicating the first voice packet and the last voice packet in consecutive voice packets (hereinafter referred to as talk spurts) are used as additional information adding means. means for adding a voice packet to a voice packet, and when a mark bit of a voice packet that arrived immediately before the voice packet non-arrival section indicates that the voice packet is the last voice packet of the talk spurt, the voice packet is Audio packet processing characterized in that the arrival section is determined to be a silence compression section, and if it is indicated that the audio packet is not the last audio packet of a talk spurt, the audio packet non-arrival section is determined to be a packet discard section. method. 4. In claim 2, means is provided for assigning a serial number (hereinafter referred to as a sequence number) assigned only to voice packets to voice packets, and the voice packets that arrived immediately before the voice packet non-arrival interval are provided. The sequence number of the voice packet that arrived next to the voice packet non-arrival section is searched, and if there is no missing number between the two sequence numbers, the voice packet non-arrival section is determined to be a silence compression section, and there is a missing number. If the audio packet does not arrive, the audio packet processing method is characterized in that the audio packet non-arrival period is determined to be a packet discard period. 5. In claim 3, there is provided means for monitoring a period in which voice packets do not arrive after the voice packet non-arrival period (hereinafter referred to as voice packet non-arrival period);
When a voice packet discard section is detected and the monitoring means detects that the next voice packet does not arrive within a predetermined time, it is determined that the last voice packet of the talk spurt has been discarded. Audio packet processing method. 6. In claim 4, means is provided to place emphasis on the period during which a voice packet has not arrived, and if the monitoring means detects that the next voice packet has not arrived within a predetermined time, the voice packet has not yet arrived. An audio packet processing method characterized in that an interval from an interval until the arrival of the next audio packet is determined to be a silent compression interval. 7. In claim 1, a means for monitoring a period in which no voice packets arrive after the voice packet non-arrival interval is provided as a determination means on the receiving side, and the means determines whether the next voice packet will arrive within a predetermined time. If it is detected that the voice packet has arrived, the section from the voice packet non-arrival zone until the voice packet arrives is determined to be a packet discard zone,
An audio packet processing method characterized in that when it is detected that the next audio packet does not arrive within the above-mentioned time period, the interval from the audio packet non-arrival interval until the next audio packet arrives is determined to be a silence compression interval. . 8. In claim 1, a means for monitoring a period in which a voice packet does not arrive after a voice packet non-arrival interval is provided as a determination means on the receiving side, and the means determines whether the next voice packet will arrive within a predetermined time. If it is detected that the voice packet has arrived, the section from the voice packet non-arrival zone until the voice packet arrives is determined to be a packet discard zone,
If it is detected that the next voice packet does not arrive within the above time, the period from the voice packet non-arrival period to the predetermined time period is determined to be the packet discard period, and the period from then on until the next voice packet arrives is determined as the packet discard period. An audio packet processing method characterized by determining a silent compression section. 9. The audio packet processing method according to claim 5, wherein if it is determined that the last audio packet of the talk spurt has been discarded, the audio packet non-arrival period is determined to be a silence compression period. 10. In claim 5, 6, 7, 8, or 9, the delay time of the audio packet arriving at the receiving side is monitored, and the delay time is set to a predetermined delay time. A voice packet processing method characterized by discarding voice packets exceeding 100 kHz on the receiving side.