JP2002185498A - Processing method for reproduction queue of voice packet, and absorbing apparatus of transmission-delay fluctuation in voice packet - Google Patents

Abstract

PROBLEM TO BE SOLVED: To absorb transmission-delay fluctuation in voice packets by using a simplified method and by performing their reproduction-queue control having a fixed queuing time, and to secure on the other hand the as early execution as possible of the reproduction of reception packets. SOLUTION: Correspondingly to a reception state, a packet to be the reference of a reproduction-queue control is altered dynamically. In a delay- fluctuation absorbing portion 7, based on the time-stump value included in a reception packet and its reception time, a predetermined operation is so performed as to inquire whether the reception interval between the voice packets is contracted in comparison with the transmission interval between the voice packets or not. When the reception interval is contracted, the uniform insertion of a queuing time is so interrupted as to pass to a reproduction processing the packet which has been the reference packet. Instead of that packet, the packet received at present is to adopted as a reference packet as to begin a new reproduction-queue control.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio packet reproduction queuing method and an audio packet transmission delay fluctuation absorbing apparatus.

[0002]

2. Description of the Related Art In recent years, a technology for efficiently transmitting a voice signal from a telephone terminal or the like via a packet communication network and having a conversation in real time has been put to practical use.

In a packet network represented by an ATM network or an IP network, even if voice packets are sent out at equal intervals on the transmitting side, the receiving side may have different delay times due to different delay times in the network for each packet. The reception interval of voice packets is not kept constant. Therefore, if the next packet has not yet arrived even after the reproduction of a certain audio packet is completed, the audio is interrupted, which causes deterioration of the audio quality.

[0004] As a countermeasure, a device for receiving a voice packet in a packet network inserts a waiting time between the reception of a voice packet and the start of decoding at a timing such as at the start of a call, so that delay fluctuations during the call are reduced. Even if it occurs, if the delay fluctuation is within the waiting time inserted on the receiving side, the audio signal will not be interrupted,
It enables continuous playback.

[0005]

As described above, if a waiting time is inserted from the reception of a voice packet to the start of decoding at a timing such as at the start of a call, continuous transmission is possible even if the transmission delay of each received packet is different. Reproduction becomes possible.

However, if the waiting time is uniformly inserted with reference to the first received voice packet, the reproduction processing of all the received voice packets is uniformly delayed by the waiting time. Such playback delays
In real-time voice communication, it may be a cause of deterioration of communication quality.

In other words, a request to obtain the effect of absorbing transmission delay fluctuations by performing reproduction queuing control and a request to reproduce a packet as soon as possible are conflicting. Have difficulty.

Conventionally, without fixing the playback waiting time,
There is also a technique of dynamically changing the control to perform control in accordance with the actual reception state of a packet.
No. 86886), such a dynamic control of the waiting time is performed gently by reflecting the actual communication state, has a shortage of immediate effect, and has a dynamic control of the waiting time. Is also complicated.

The present invention has been made on the basis of the above-described studies, and uses a simplified method, and
It is an object of the present invention to perform a fixed time reproduction queuing control to absorb transmission delay fluctuations, and at the same time, to ensure early execution of a received packet reproduction process as much as possible.

[0010]

According to the present invention, the playback waiting time itself is fixed, but the packet serving as the reference for the playback waiting is dynamically changed according to the reception situation.

In other words, on the receiving side, when a subsequent packet following the reference packet arrives earlier than the arrival time determined from the transmission interval (in such a situation, the reference packet is greatly delayed while the subsequent packet Can occur if the delay is small), it can be assumed that the packet is likely to arrive earlier than expected.
Even in such a case, uniformly waiting for the reproduction of the packet unnecessarily delays the reproduction. Therefore, the packet that has been used as a reference in the reproduction waiting process is immediately passed to the reproduction process, and the currently received packet is used as a new reference packet to perform the queuing control.

[0012] Thus, for example, subsequent packets, while clog the gap, in a situation to arrive one after another, the packets arriving earlier, since Ikeru Turn the reproduction processing one after another, resulting in Thus, the end point of the reproduction (decoding) of the received packet can be advanced. On the other hand, since the queuing control is performed as before, the delay fluctuation can be absorbed, and even if the transmission delay fluctuation varies for each packet, the problem that the reproduction is interrupted due to the fluctuation does not occur.

However, there is a possibility that the reproduction will be discontinuous with the change of the reference packet. Therefore, it is more preferable to perform such control only when a silent packet is received.

In one aspect of the audio packet reproduction wait processing method according to the present invention, when the audio packet transmitted from the transmitting side is received and the reproduction wait processing for absorbing the transmission delay fluctuation is performed, the reproduction wait processing is performed. It is determined whether or not the reception interval between the reference packet used as the reference and the subsequent reception packet received after the reference packet is shorter than the transmission interval on the transmitting side, and as a result of the determination, the reception interval is shorter. If it is found, the subsequent packet that has been determined is used as a new reference packet to start a new playback waiting process, and for the old reference packet, without waiting for the playback waiting time to expire. Start playback processing.

For example, equal intervals (predetermined intervals) from the transmitting side
If the transmission delay amounts of the respective packets sent in step (1) are the same, the packets should arrive at the receiving side at the same intervals (at equal intervals) as the transmitting side. However, for example, if the first arriving packet (reference packet) arrives with the largest delay for some reason, the delay of subsequent packets is smaller than the delay of the reference packet. The packet receiving interval on the receiving side is more congested than the packet transmitting interval on the transmitting side. In such a case, the necessity of waiting for the packet to be reproduced is reduced, so that the previously received packet is forwarded to the reproduction process without waiting for the expiration of the initial reproduction waiting time, and instead, the packet has just arrived. A new reproduction waiting process is performed based on the packet. As a result, it is possible to obtain both the effect of hastening the end point of packet reproduction and the effect of absorbing transmission delay fluctuation.

[0016]

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

FIG. 1 is a block diagram showing a main configuration of a voice packet transmitting apparatus according to an embodiment of the present invention.

In FIG. 1, a terminal interface unit 1 receives an audio signal from a terminal and transmits the audio signal to the terminal.

The terminal interface unit 1 is connected to a voice coding unit 2, a voice decoding unit 8, and a voice / non-voice determination unit 3.

The voice coding section 2 codes a voice signal received from the terminal interface section 1. A packet assembler 4 is connected to the audio encoder 2.
The packet assembler 4 receives the encoded audio signal encoded by the audio encoder 2 and assembles it into packets.

The voice / silence determining unit 3 is a section (voice section) containing significant voice information in the voice signal received from the terminal interface unit 1 or a section (voiceless section) containing no significant voice information. ) Is determined for each voice section to be packetized, the result is notified to the packet assembling unit 4, and the sound / non-sound information is set in the packet.

The packet assembling section 4 is connected to the transmission path interface section 5. Transmission line interface unit 5
Has a function of receiving a voice packet from the packet assembling unit 4 and transmitting the voice packet to the transmission path, and a function of transmitting the voice packet received from the transmission path to the packet decomposing unit 6.

The packet decomposing unit 6 decomposes an audio packet received from the transmission line interface unit 5 and extracts an encoded audio signal, a time stamp, and voice / non-voice information.
The packet decomposing unit 6 is connected to the delay fluctuation absorbing unit 7. The delay fluctuation absorbing unit 7 temporarily stores the encoded voice signal extracted by the packet decomposing unit 6, the time stamp, the voiced / unvoiced information, and the reception time of the voice packet notified from the timer 9, and the voice decoding unit 8 It has a function to insert a waiting time before output to the receiver and to absorb delay fluctuations of received voice packets.

The speech decoding section 8 decodes the encoded speech signal received from the delay fluctuation absorbing section 7 and sends it to the terminal interface section 1. The timer 9 is connected to the packet assembling unit 4 and the packet disassembling unit 6, and notifies the time. The packet assembler 4 sets the time notified from the timer 9 as a time stamp in the packet.

In the fluctuation calculator 10, the time stamp set in the received voice packet and the delay fluctuation absorber 7
A function of calculating the fluctuation time of the received voice packet from the reception time of the voice packet stored together with the coded voice signal, and controlling the transmission timing of the coded voice signal from the delay fluctuation absorbing unit 7 to the voice decoding unit 8 Is provided.

The portion surrounded by the dotted line (reference numeral 12) in FIG. 1 constitutes transmission delay fluctuation absorbing means.

Next, the operation of the voice packet transmitting apparatus of FIG. 1 will be described.

The audio signal input from the terminal interface unit 1 is encoded by the audio encoding unit 2 to become an encoded audio signal. The voice signal input from the terminal interface unit 1 is determined by the voice / non-voice determination unit 3 to be a section containing significant voice information (voice section) or a section containing no significant voice information (silence section). ) Is determined.

The header of the coded voice signal coded by the voice coder 2 together with the time (time stamp) notified from the timer 9 and the result of determination by the voice / non-voice determiner 3 by the packet assembler 4. They are added and assembled into packets.

When the voice signal is accommodated in the ATM network, the voice packet assembling section 4 assembles ATM cells as packets. When the audio signal is accommodated in the frame relay network, the audio packet assembling unit 4 assembles a frame of the frame relay as a packet.

Further, when the audio signal is accommodated in the IP network, the audio packet assembling section 4 assembles the IP packet as a packet. The packet assembled by the packet assembling unit 4 is transmitted to the transmission line via the transmission line interface unit 5.

On the other hand, the voice packet received from the transmission path via the transmission path interface section 5 is decomposed by the packet decomposing section 6 into an encoded voice signal, a time stamp, and sound / non-speech information.

The encoded voice signal, time stamp, voiced / silent information, and the reception time of the voice packet notified by the timer 9 are temporarily stored in the delay fluctuation absorbing unit 7.
When a voice packet is received in a state where there is no stored data in the delay fluctuation absorbing unit 7 at the start of a call or the like, a period from when the voice packet is received by the packet decomposing unit 6 to when it is output to the voice decoding unit 8 The waiting time is inserted for the set time.

The coded voice signal of the subsequent received voice packet is read out from the delay fluctuation absorbing unit 7 to the voice decoding unit 8 at regular intervals with a period corresponding to the time length of the voice signal included in one packet. It is. Thus, even if the reception interval of a voice packet during a call is longer than the processing cycle of the voice decoding unit due to the delay fluctuation, the delay fluctuation absorbing unit 7
As long as it is within the waiting time inserted in step (1), the audio decoding unit 8 can continuously decode the encoded audio signal.

The audio signal decoded by the audio decoder 8 is transmitted to the terminal via the terminal interface 1.

If the first received packet at the start of the call arrives with the shortest delay fluctuation, there is no problem in the above operation. However, if the first received packet at the start of a call arrives with the maximum delay fluctuation, the voice packets received during the call will not be able to shorten the reception interval between packets, based on the first received packet. Even if it does not grow.

Accordingly, in such a situation, the necessity of controlling the waiting time in the delay fluctuation absorbing unit 7 until reading of the encoded audio signal by the audio decoding unit 8 is reduced. In other words, the uniform insertion of the waiting time results in an increase in the end-to-end transmission delay time.

Therefore, in the present embodiment, control is performed so that queuing control is selectively performed based on packets that have received small delay fluctuations among packets received during a call. For this purpose, the fluctuation calculation unit 10 performs an operation based on the following expression (1) (and expression (2)), and determines the “relative fluctuation time” of each received packet with respect to the packet used as a reference for the queuing control. Ask for.

PDVn = PDVp + delay fluctuation with respect to previous reception packet (1) delay fluctuation with respect to previous reception packet = (reception time of current reception packet−reception time of previous reception packet) − (reception time of current reception packet (Time stamp value-time stamp value of previous received packet) (2) PDVn: delay fluctuation of the current received packet with respect to the reference packet PDVp: delay fluctuation of the previous received packet with respect to the reference packet PDVn: reception of the reference packet Sometimes 0.

Here, PVDn basically represents the difference between the transmission delay fluctuation received by the reference packet and the transmission delay fluctuation received by the subsequent packet of interest with respect to this reference packet. In other words, the above equation (1) is to accumulate the relative transmission delay fluctuation of the current received packet with respect to the immediately preceding received packet (previous received packet), thereby obtaining the relative value of the current received packet with respect to the reference packet. This means that transmission delay fluctuation is required.

The above equation (2) is essentially a subtraction of the transmission interval from the reception interval. If the reception interval is shorter than the transmission interval, the equation (2) takes a negative value. Become. Eventually, the expression (1) is obtained by accumulating the relative fluctuation amount obtained by the expression (2) up to the reference packet, so that the subsequent packet of interest (now received packet)
And the reference packet, the transmission interval on the transmitting side,
This means that it is determined which of the reception intervals on the receiving side is larger.

In equation (2), since the time stamp value attached to the packet is considered to indicate the transmission time point of the packet, the transmission interval is obtained using the time stamp. .

As a result of the operation based on the above equation (1), PDVn
Becomes a negative value, it means that the packet reception interval of the voice packet received this time is relatively shortened with respect to the packet in which the waiting time set in advance by the delay fluctuation absorbing unit 7 is inserted.

That is, in this case, it can be determined that the reference packet subjected to the queuing control has received a larger delay fluctuation than the currently received packet.

Further, it is expected that the reception interval of the packets to be received thereafter will be shortened.

Under such a reception situation, uniformly inserting the waiting time with the first packet as the reference packet results in delaying the reproduction (decoding) of the packet.

Therefore, in the above equation (1), if PDVn becomes a negative value, the voice packet received this time is newly changed to the reference packet. Then, a newly set waiting time is inserted into the changed reference packet, and thereafter, the encoded audio data is output to the audio decoding unit 8.

On the other hand, the original reference packet is immediately sent to the voice encoding unit 8 without waiting for the end of the waiting period.

In this way, a voice packet arriving after receiving a smaller delay fluctuation can be used as a reference packet for which the delay fluctuation absorbing unit 7 performs queuing control. That is, it is possible to dynamically change the reference packet according to the reception status.

Such a series of processing can be basically performed only by a simple operation using the time stamp value and the timer value, so that the existing apparatus configuration can sufficiently cope with the processing. Very simple. Therefore, realization is easy.

Hereinafter, the characteristic operation and effect of the apparatus according to the present embodiment will be specifically described with reference to FIG.

As shown on the left side of FIG.
Packets A, B, C, and D are transmitted. The transmission interval of each packet is “2 (relative value)”.

In FIG. 2, the delay amount (common delay amount) unique to the transmission path is set to “7 (relative value)”, and packets A to D
The delay fluctuation amount (relative value) is “4”, “
3 ”,“ 3 ”, and“ 2. ”In FIG. 2, a common portion of the transmission delay is indicated by a straight line, and a delay fluctuation portion of each packet is indicated by a dashed line.

A portion surrounded by a thick dotted line in the center of FIG. 2 is a portion indicating a receiving operation. In FIG. 2, a portion surrounded by a thick dotted line at the lower right is a portion indicating a reproducing operation according to the present invention, and a portion surrounded by a thick dotted line at the upper right is a portion indicating a reproducing example according to the conventional method. It is. In FIG. 2, the reproduction waiting time is indicated as “WT”.

Packet A is received at time t6 with the maximum delay fluctuation "4". The next packet B is received at time t7 with delay fluctuation “3”. In this case, the relative delay fluctuation is "3-4 = -1", and the relative reception interval between the packet A and the packet B is reduced by "1" from the transmission interval.

In this case, as indicated by the thick dotted line in the lower right part of FIG. 2, the packet A is immediately sent to the reproduction processing.

Then, the reference packet is changed to the currently received packet B, and the reproduction waiting control based on the packet B is newly started.

Subsequently, the packet C has a delay fluctuation “3”.
At time t8. In this case, since there is no difference in the delay fluctuation between the reference packet B and the packet C (that is, the relative delay fluctuation is “0”), the reproduction waiting control based on the packet B is continued.

Subsequently, the packet D arrives at time t10 with the minimum delay fluctuation "2". In this case, the relative delay fluctuation with respect to the previous received packet C is “−1”. Then, before the received packet C, the reference relative delay fluctuation for the packet B was "0", the packet D received now, relative delay fluctuation with respect to the reference packet B is cumulatively adding each fluctuation value And "0 + (-1)
= -1 ".

That is, as compared with the interval between the packet B and the packet D on the transmitting side, the interval on the receiving side is “
Therefore, in this case, the packet B is immediately forwarded to the playback processing without waiting for the expiration of the playback waiting time (time t10).

On the other hand, reproduction waiting control is started using packet D as a new reference packet.

Since the packet C has arrived before the packet C which has become the new reference packet, the time t12
Is sent to the regeneration processing step. Then, at time t13
At this time, the playback waiting time based on the packet D expires, and at this point, the packet D is sent to the playback processing step.

Thus, according to the method of the present invention,
At time t14, all of the received packets A to D can be sent to the reproduction process.

On the other hand, as shown in the upper right of FIG. 2, in the conventional method, reproduction waiting control is performed based on the first packet A, and at time t9, the packet A is sent to the reproduction processing. Then, at time t15, all of the received packets A to D can be sent to the reproduction processing.

[0065] As Comparing the upper right and lower right of Figure 2 is obvious, according to the method of the present invention, the time for feeding the whole of the received packet to the reproduction process, by the amount of time t15~ time t14, the conventional It is faster than the method.

However, when the method of the present invention is carried out, as in the period from time t8 to time t10 in FIG. 2, the reproduction of the packet may be discontinuous due to the change of the reference packet.

Therefore, the PD shown in the above equation (1)
The calculation of Vn and the queuing control for a new packet by the delay fluctuation absorbing unit 7 are executed only for voice packets whose voice / silence information extracted by the packet decomposing unit 6 indicates "silence". Most preferred.

In this case, it is possible to reduce the influence on the audibility of a break in sound that may occur when changing the reference packet of the queuing control. When the playback queuing control of the present invention is performed in a silent part, a silent part in a conversation is accurately inserted without a large delay, which is effective.

FIG. 3 summarizes the main points of the reproduction waiting control according to the present invention described above.

That is, the relative delay fluctuation (PVDn) of the currently received packet (currently received packet) with respect to the reference packet is obtained (step 200). Next, it is determined whether the value of the relative delay fluctuation (PVDn) is negative (step 201).

In the case of a negative value, the packet serving as the reference for the queuing control is changed to the currently received packet (step 20).
2). If the result of the determination in step 201 is not negative, the queuing control based on the previously received packet is continued (step 203).

The present invention can be variously modified and applied. In general, an encoded audio signal is output from the audio encoding unit at regular intervals, so that the time stamps set in the audio packets are also at equal intervals.

Therefore, on the receiving side, it is possible to detect the discarding of the packet in the network by monitoring the time stamp interval of the voice packet continuously received by the packet decomposing unit 6. Therefore, the packet decomposing unit 6 that has detected the discard of the packet outputs a dummy packet to the delay fluctuation absorbing unit 7 together with a flag indicating the information of the occurrence of the discard.

The voice decoding unit 8 that has read out the dummy packet to which the flag indicating the information indicating the occurrence of discarding has been added, sends to the terminal interface unit 1 the voice that has been silenced or that has been decoded by the frame erasure compensation process defined by the voice decoding algorithm. Output. In addition, packet discard detection is performed by RTP (Real
It can also be realized by monitoring the continuity of the sequence number of the (Time Protocol) header.

The calculation of the fluctuation time according to the equations (1) and (2) is not performed on the packet to which the flag indicating the information of the packet discard is added.

As described above, according to the present embodiment, the voice packet receiving apparatus can execute the queuing control for absorbing the delay fluctuation on the basis of the packet having received the smaller delay fluctuation. This can contribute to a reduction in to-End transmission delay time.

FIG. 4 shows the overall configuration of a system for performing communication via a wide area network such as the Internet. In FIG. 4, reference numeral 10 is a wide area network, reference numerals 20 and 30 are private networks (or home networks), and reference numerals 40 and 50 are
It is a gateway device.

Note that reference numerals 60 to 110 are computer terminals. Reference numeral 120 denotes a monitoring terminal.

The voice packet transmission device of FIG. 1 is built in gateway devices 40 and 50. As a result, the gateway devices 40 and 50 can have a function that can achieve both the absorption of the transmission delay fluctuation of the voice packet and the early reproduction process without complicating the configuration.

FIG. 5 is a diagram for explaining the relationship between a voice packet transmitting apparatus and a recording medium on which a program for executing the above-described procedure of the present invention is recorded.

A program for causing the main body 300 of the voice packet transmitting apparatus to perform the above-described characteristic operation may be recorded on a single recording medium 301,
It may be stored in a unique database 302 on the wide area network 10.

The recording medium 301 is used by being inserted into the main body 300 of the voice packet transmission apparatus. If the database 302 contains a program, the program is downloaded to the main body 300 of the voice packet transmission device via the wide area network 10 and used.

[0083]

As described above, according to the present invention, when a reception interval is shorter than a transmission interval and a reception is performed, an operation of dynamically changing a packet serving as a reference for reproduction wait control is executed. By doing so, it is possible to reduce the total time required to send the received packet to the reproduction process while absorbing the propagation delay fluctuation. Further, the method of the present invention is a simple method that does not require a special device configuration, and is therefore easy to realize.

[Brief description of the drawings]

Block diagram showing the configuration of a voice packet transmission apparatus according to an embodiment of the present invention; FIG

FIG. 2 is a diagram for explaining specific operations and effects of the voice packet transmission device of FIG. 1;

FIG. 3 is a flowchart showing a characteristic operation procedure in the voice packet transmission device of FIG. 1;

FIG. 4 is a diagram illustrating an overall configuration of an example of a communication system via a wide area network;

FIG. 5 is a diagram for explaining the relationship between the audio packet transmission device and a recording medium on which a program for executing the playback queuing control method of the present invention is recorded.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 terminal interface unit 2 audio encoding unit 3 voiced / silent determining unit 4 packet assembling unit 5 transmission line interface unit 6 packet decomposing unit 7 delay fluctuation absorbing unit 8 audio decoding unit 9 timer 10 fluctuation calculating unit

Claims (12)

[Claims] 1. A method for receiving a voice packet transmitted from a transmitting side and performing a reproduction queuing process for absorbing transmission delay fluctuation, comprising: a reference packet used as a criterion for the reproduction queuing process; It is determined whether or not the reception interval with the subsequent reception packet received after is shorter than the transmission interval on the transmitting side. If the result of the determination indicates that the reception interval is shorter, the determination is made. The reproduction waiting process is newly started using the subsequent packet targeted for the above as a new reference packet, and the reproduction process is started without waiting for the expiration of the reproduction waiting time for the old reference packet. An audio packet playback wait processing method. 2. A method for receiving a voice packet to which time stamp information is added, transmitted from a transmitting side, and performing a reproduction queuing process for absorbing transmission delay fluctuation. , The relative fluctuation time (PDV
n) is obtained based on the following (1) equation (fluctuation calculation processing), and it is determined whether or not the relative fluctuation time (PDVn) with respect to the reference packet becomes negative. The reproduction waiting process is newly started by using the currently received packet in the above formula (1) as a new reference packet, and the reproduction process of the old reference packet is started without waiting for the expiration of the reproduction waiting time. Play waiting processing method of voice packets. (Fluctuation calculation processing) PDVn = PDVp + delay fluctuation time with respect to the previous reception packet... (1) In the equation (1), PDVn is the delay fluctuation time of the present reception packet with respect to the reference packet, and is "PDVp".
Is the delay fluctuation of the previous received packet with respect to the reference packet. Further, “delay fluctuation with respect to the previous received packet” in the equation (1) is obtained based on the following equation (2). Delay fluctuation for previous received packet = (received time of current received packet-received time of previous received packet)-(timestamp value of current received packet-timestamp value of previous received packet) (2) 3. The method according to claim 1, wherein a series of processing for changing the reference packet is performed only for a voice packet in which voice / silence information obtained by decomposing a received packet indicates no voice. And a method of processing audio packet playback queuing. 4. Receiving a voice packet to which time stamp information is added, transmitted from a transmitting side, and performing a playback queuing process for absorbing transmission delay fluctuation.
A transmission delay fluctuation absorber for voice packets, a packet decomposer for decomposing received voice packets, and a delay for performing a reproduction queuing process by temporarily storing data obtained by decomposing by the packet decomposer. A fluctuation absorber, a time stamp obtained from the decomposed voice packet,
Using the reception time of the received voice packet, the received voice packet, a fluctuation calculation unit for calculating a relative fluctuation time with respect to a reference packet as a reference of the reproduction waiting process, comprising: From the calculation result, it is determined whether or not the reception interval between the reference audio packet as the reference for the playback waiting process and the audio packet of interest is shorter than the transmission interval of both audio packets on the transmitting side, and according to the determination result, A transmission delay fluctuation absorber for voice packets, wherein a timing of reading data from the delay fluctuation absorber is controlled. 5. The method according to claim 4, wherein a reception interval between the reference audio packet as a reference of the reproduction waiting process and the audio packet of interest is determined to be shorter than a transmission interval of both audio packets on the transmission side. The delay fluctuation absorbing unit newly starts the playback waiting process using the voice packet of interest as a new reference packet, and outputs the old reference packet without waiting for the playback waiting time to expire, thereby performing playback. A transmission delay fluctuation absorbing device for voice packets, characterized by being started. 6. The operation according to claim 4, wherein the operation of controlling the output timing of data from the delay fluctuation absorbing unit based on the calculation result in the fluctuation calculating unit is obtained by decomposing a received packet. An apparatus for absorbing transmission delay fluctuations of voice packets, wherein the apparatus is implemented only for voice packets in which voiced / silent information indicates silence. 7. An audio packet transmission apparatus for accommodating an audio signal in a packet network, comprising: an audio encoding unit for encoding an audio signal obtained via a terminal interface unit; and a receiving unit for receiving the audio signal from the audio encoding unit. A voice packet assembling unit for assembling the encoded voice signal into a voice packet and sending the voice packet to a transmission line interface unit, a timer for setting the voice packet assembling time to the voice packet, and a packet for decomposing the voice packet received from the transmission line interface A decomposing unit, a delay fluctuation absorbing unit that stores data decomposed by the packet decomposing unit, a fluctuation calculating unit that calculates a delay fluctuation of a received voice packet from a time stamp obtained from the decomposed packet and a reception time of the voice packet,
An audio decoding unit that decodes an encoded audio signal output by the packet decomposing unit, wherein reading timing from the delay fluctuation absorbing unit to the audio decoding unit is controlled based on a calculation result in the fluctuation calculating unit. A voice packet transmission device characterized by the above-mentioned. 8. An audio packet transmitting apparatus for accommodating an audio signal in a packet network, comprising: an audio encoding unit for encoding an audio signal obtained via a terminal interface unit; and a receiving unit for receiving the audio signal from the audio encoding unit. A voice packet assembling unit that assembles the encoded voice signal into a voice packet and sends the voice packet to the transmission line interface unit, a timer that sets the voice packet assembling time to the voice packet, and a voice / voice signal based on the voice signal received from the terminal interface unit. A voice / silence determining unit that performs a silent determination and sets the determination result in a voice packet, a packet decomposing unit that decomposes a voice packet received from the transmission path interface, and a delay that accumulates data decomposed by the packet decomposing unit. A fluctuation absorbing part,
A fluctuation calculator for calculating a delay fluctuation of a received voice packet from a time stamp obtained from the decomposed packet and a reception time of the voice packet, and a voice decoder for decoding an encoded voice signal output by the packet decomposer. , The sound of the received packet obtained from the packet decomposer /
An audio packet transmission device, wherein the timing of reading from the delay fluctuation absorbing unit to the audio decoding unit is controlled based on the operation result of the fluctuation calculating unit only when a silent packet is received based on the silent information. 9. The voice packet transmission device according to claim 7, wherein the network connected to the transmission line interface unit is any one of an ATM network, a frame relay network, and an IP network. . 10. The system according to claim 8, wherein the assembly time set in the voice packet is RTP (Real-time
An audio packet transmission device using a time stamp of Transport Protocol). 11. A gateway device comprising the voice packet transmission device according to any one of claims 7 to 10. 12. A reference packet used as a reference for a reproduction waiting process, comprising: a voice packet transmitting apparatus that receives a voice packet transmitted from a transmission side and performs a reproduction waiting process for absorbing transmission delay fluctuation; Means for determining whether or not the reception interval with a subsequent reception packet received after the packet is shorter than the transmission interval on the transmitting side; and, if the result of the determination indicates that the reception interval is shorter, Means for newly starting the reproduction waiting process using the subsequent packet as the target of the determination as a new reference packet, and for the old reference packet, starting the reproduction process without waiting for the reproduction waiting time to expire. And a program for causing the computer to function.