JP2001251342A - Packet receiver - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a packet receiver that can enhance sound quality in the case of insertion of a correction packet. SOLUTION: The packet receiver detects whether or not each communication packet is normally received in time series on the basis of an identifier given to each communication packet containing valid voice data by a transmitter side, and decides whether or not a prescribed correction packet not containing valid voice data is to be inserted to a received communication packet series. The packet receiver is provided with a communication packet size detection means that checks a size of a communication packet being that of the received communication packet and a correction packet insertion means that selects at least a size smaller than that of the communication packet as the size of the correction packet when the size of the communication packet exceeds a threshold value of the communication packet size set to the packet insertion means.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a packet receiving apparatus, and can be applied to, for example, a case where a packet transmitted containing an encoded voice signal is received, decoded, and voice output is performed. .

[0002]

2. Description of the Related Art In packet voice communication in which a packet containing voice data is transmitted and received, a packet arrival delay (called "jitter" or "fluctuation") at a receiving apparatus is called a packet during communication. Packet loss (or “packet loss”), which has a significant effect on call quality. An adverse effect common to jitter and packet loss is that the receiving buffer mounted on the receiving device becomes empty. If the next packet does not arrive before the data accumulated in the reception buffer is used up, the reception buffer becomes empty. When a packet loss occurs, the data that should have arrived is lost, so that the data amount of the lost packet is lost, and the data stored so far is consumed without storing the data in the reception buffer. For this reason, if the arrival of the next packet is delayed after a packet loss, as in the case of large jitter,
The receive buffer becomes empty. When the reception buffer becomes empty, there is no data to be reproduced, and in the worst case, the sound is interrupted.

There are other adverse effects on packet loss. For example, if the first packet to the third packet are transmitted, and the second packet is lost, the reception buffer stores the third packet after the first packet. appear. This skipping not only lowers the call quality, but also causes the information to be lost, so that the contents of the call become unclear.

As a method for solving these problems, Japanese Patent Laid-Open Publication No. Hei 10-285213 (Document 1) discloses a method in which a packet does not arrive even if a reception buffer becomes empty.
A method for reproducing a dummy packet having the same size as a transmission packet is shown. Document 1 describes that this can avoid interruption of sound.

[0005]

However, according to the method disclosed in the above document 1, dummy packets are reproduced only when the reception buffer becomes empty, so that even if a packet loss occurs, the reception buffer does not become empty. Causes sound skipping as in the prior art.

[0006] Depending on various conditions such as network traffic conditions and the capacity of the receiving buffer, the receiving buffer may be empty even if no packet loss occurs, but the receiving buffer does not become empty even if a packet loss occurs. It is possible.

[0007] Increasing the capacity of the receiving buffer can reduce the possibility that the receiving buffer will be empty even if no packet loss occurs, but increasing the capacity of the receiving buffer simultaneously reduces the packet loss. This means increasing the possibility that the receiving buffer will not be empty even if it occurs.

[0008]

In order to solve such a problem, the present invention provides a method for receiving a communication packet containing valid voice data in a time-series manner based on an identifier given by a transmission side to each communication packet. It detects whether each communication packet is normally received, and inserts a predetermined correction packet that does not contain valid voice data into the received communication packet sequence based on the detection result. In the packet receiving device for determining whether to perform the communication packet size, a communication packet size detecting means for checking a communication packet size, which is the size of the received communication packet, and a threshold value for the communication packet size are set, and at least the communication packet size is If the threshold value is exceeded, a correction packet for selecting a size smaller than the communication packet size as the size of the correction packet. Characterized in that it comprises a Tsu preparative insertion means.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (A) Embodiment An embodiment of a packet receiving apparatus according to the present invention will be described below.

The features common to the first to third embodiments are as follows.
If the size of the received packet (the number of frames included in one received packet) is small, an error packet having the same size as the packet is inserted. If the size of the received packet is large, an error packet smaller than the packet is inserted. By inserting a packet, the sound quality of the error packet portion is prevented from deteriorating.

(A-1) Configuration of First Embodiment FIG. 7 shows the configuration of a receiving apparatus 20 of the present embodiment.

In FIG. 7, a receiving apparatus 20 includes a receiving end 1.
, Packet monitoring unit 2, reception buffer 3, decoding unit 4
And an output terminal 5.

The receiving end 1 has a function of receiving a packet containing coded voice data transmitted from a transmitting side device (not shown) via a network and passing the packet to a packet monitoring unit 2. Hereinafter, a packet transmitted from the transmitting apparatus and received by the receiving apparatus 20 (or the receiving end 1) via the network is referred to as a received packet.

The packet monitoring unit 2 has a function of setting and monitoring a time-out period, a function of monitoring the reverse of the order of arrival of received packets, a function of generating and inserting an error frame, and a function of discarding received packets. It is shown.

In order to monitor the timeout period and the reverse of the order of arrival of the received packets, the packet monitoring unit 2 needs to identify each received packet as a prerequisite. For this identification, the transmitting side apparatus uses packet identification information given to each received packet. The packet identification information includes, for example, a sequence number (packet number) of a received packet, time stamp information indicating a time at which the received packet was transmitted from the transmitting device and a time at which the received packet was generated at the transmitting device. May be.

FIG. 1 shows the internal configuration of the packet monitoring unit 2.

(A-1-1) Internal Configuration of Packet Monitoring Unit 2 In FIG. 1, the packet monitoring unit 2 includes a reception normality detection unit 11, a packet size detection unit 12, a threshold setting unit 13, An error packet size selection unit 14 and an error packet insertion unit 15 are provided.

The reception normality detector 11 detects the packet transmitted by the transmitting apparatus based on the above-mentioned sequence number and time stamp without any loss or reversal of the packet.
In a portion for detecting whether or not a received packet is normally received within a certain timeout period, if a missing, reversed order, or timeout is detected in each received packet, an abnormal reception signal AS Is supplied to the packet size detector 12.

The purpose of the packet number, the timeout time, and the time stamp information is to monitor the loss of the received packet and the reversal of the order of arrival of the received packet, and any information can be used as long as they can be detected. For example, TCP / IP (Transmission
Control Protocol / Internet Protocol (Sequence Number) information, UDP / IP (User Datagram Protocol /
Internet Protocol (RTP) RTP (Realtime Transport Pr
otocol) can be used.

More specifically, the receiving side (the reception normality detecting section 11) monitors the sequence number of the received packet,
If the sequence number of the currently received packet is not the number obtained by incrementing the sequence number of the previously received packet (the sequence number of the previously received packet + 1), it is assumed that packet loss or inversion of the order of arrival of the received packet has occurred. It is conceivable that the reception normality detector 11 makes the determination.

In the case where the time stamp of the previous received packet is 16:52 and the time stamp of the current received packet is 16:40, even if the currently received packet is transmitted earlier than the previous received packet. Regardless, since the currently received packet arrives later, the reception normality detecting unit 1 determines that the order of arrival of the received packet has been reversed.
It is possible that 1 determines.

These methods are merely examples, and the present invention is not limited to this method. The transmitting apparatus adds unique detection information to each received packet and causes the packet to be transmitted. Needless to say, 11 may be performed.

When receiving the abnormal signal AS output from the reception normality detecting section 11, the packet size detecting section 12
It has a function of checking the size of each received packet being received and supplying the size to the error packet size selection unit 14 as size information SI.

The error packet size selection unit 14 receives the setting of the threshold value TH for the received packet from the threshold value setting unit 13, and determines the size by comparing the size information SI with the threshold value TH. This is a part that outputs the signal DS to the error packet insertion unit 15.

The error packet inserting section 15 is a section for inserting an error packet having a size defined according to the size determination signal DS into a received packet sequence.

An error packet is a collection of a plurality of error frames. An error frame is, for example,
A packet used in the concealment function of frame erasure, which will be described later, compensates for loss of frame data due to auditory characteristics, and compensates for loss in the amount of data (queue length) accumulated in the reception buffer 3 for subsequent reception. This is data created for the purpose of controlling the delay time (fluctuation in the receiving device 20, that is, the receiving packet interval) of the received packets to be stored.

For the structure of the error frame itself, for example, a method of creating error frame data by imitating white noise, colored noise, and data of the previous frame, and creating error frame data by inferring from the data of the previous frame Can be used. A method of imitating the data of the previous frame is described in, for example, G.11. 723.1 and G72
9 can be used. Further, as a method of imitating the data of the previous frame, various methods such as a linear prediction method, an adaptive prediction method, and a prediction using a neural network can be used. These imitation methods and prediction methods are not limited to those described here, and it goes without saying that the effects of the present embodiment can be obtained by a unique method.

The insertion of the error packet into the received packet sequence is performed, for example, after the last received packet (coded audio data) stored in the receiving buffer (first-in first-out type memory: FIFO) 3 in FIG. This is done by writing an error packet.

The decoding unit 4 has a function of decoding the audio data of the received packet (or error packet) read from the reception buffer 3 and a function of concealing the frame erasure.

The concealment function of frame erasure is that, when an error packet is detected in a packet sequence to be decoded, the error packet portion detects the error packet based on the audio data of the previous and next received packets. This is a function for synthesizing output speech by itself. With this feature,
The effect on the audio output that the valid audio data is lost in the error packet portion can be reduced as much as possible.

Here, the coding and decoding schemes of speech include, for example, linear PCM and ITU-T recommended G.264. 711
(Μ law PCM); 726 (ADPCM); 72
3.1, G. 729 (CS-ACELP), CELP (Code Excited Linear Prediction) coding, linear PCM, and the like. In addition, it goes without saying that the effects of the present embodiment can be obtained even with an encoding system of a unique specification, and the encoding / decoding system is not particularly limited.

The encoding and decoding methods affect the time of one frame during encoding and decoding. For example, in the above G.P.
In the 729 system, one frame is 10 ms (0.01 seconds)
G. In the 723.1 system, the frame is 30 ms.
(0.03 seconds). The number of frames included in one packet (called a packet size and a packet length) also varies, and this can be arbitrarily determined depending on the device used. However, one packet is usually formed by connecting coded audio data to a plurality of frames and adding a header which is information unique to the communication system. For this reason, as will be described later, if the packet size of the received packet is reduced, a higher communication speed is required instead of reducing the delay time. Conversely, if the packet size of the received packet is increased, the communication speed can be reduced. However, the influence of packet loss and the delay time tend to increase.

The output terminal 5 receiving the decoding result from the decoding unit 4 has a function of reproducing audio data in accordance with the decoding result, and may be, for example, a speaker for outputting audio.

The operation of the present embodiment having the above configuration will be described below.

(A-2) Operation of the First Embodiment The operation of the packet monitoring unit 2 is shown in the flowchart of FIG. This flowchart corresponds to a case where a timeout (Time Out) is detected as the reception normality, and includes steps S1 to S5. The process starts from the first step S1 each time the progress of the timeout period for each received packet is started.

The reception normality detector 1 in the packet monitor 2
1 constantly monitors the signal (S1) supplied from the receiving end 1 in accordance with the reception processing of each received packet (S2).

Within the timeout period, a signal indicating the arrival (reception) of the received packet is received by the reception normality detector 11.
In step S2, the branch on the No side indicating that the received packet is not received is continued, and in step S4, the branch on the No side is selected while recognizing that the timer has advanced until the timeout period. I do.

When a signal indicating that the received packet to which the target packet identification information has been received has been received from the receiving end 1 at any time within the timeout period (S1), step S2 is performed on the Yes side. Then, the received packet (or the encoded voice data contained in the received packet) is input to the reception buffer 3 (S3). At this time, the timer is reset and shifts to the monitoring state for the next timeout period.

On the other hand, if the signal indicating the reception of the target received packet has not been supplied to the reception normality detecting unit 11 before the timeout period, the step S2 does not branch to the Yes side within the timeout period. , The branch on the No side indicating that the received packet is not received is continued.

When detecting that the continuation state of the branch on the No side in step S2 has reached the time point of the elapse of the timeout period, the reception normality detection unit 11 supplies the abnormal reception signal AS to the packet size detection unit 12. , Step S
4 branches to the Yes side to execute the error packet generation processing in step S5.

In the error packet generation processing (S5), the following processing is executed.

First, the packet size detection unit 12 which has received the abnormal reception signal AS detects, for example, the size of a reception packet received immediately before the reception packet determined to be timed out, and determines a size corresponding to the size. The information SI is supplied to the error packet size selection unit 14.

Next, the error packet size selection unit 14 compares the size information SI with the threshold value TH supplied from the threshold value setting unit 13, and the size information SI is larger than the threshold value TH. If it is larger, the size of the error packet is determined based on the following equation (5). That is, at this time, the error packet size is equal to or larger than the size specified by TH and smaller than the size specified by SI.

TH ≦ error packet size <SI (5) On the other hand, if the size information SI is smaller than the threshold value TH, the error packet size is set based on the following equation (6). Is determined to be the same as the size specified by.

Error packet size = SI (6) Finally, a size determination signal D that specifies the error packet size determined based on these equations (5) or (6)
When S is supplied to the error packet inserting unit 15, the error packet inserting unit 15 inserts the error packet by writing the error packet of the size into the reception buffer 3.

At this time, the state of the size determination signal DS can be roughly divided into two states, a state corresponding to the equation (5) and a state corresponding to the equation (6). For the size determination signal DS to be performed, the error packet inserting unit 15 may set the size of the error packet to be inserted to be the same as the size of the received packet (corresponding to SI).

If the size of each received packet does not change during communication, the size of the error packet inserted by the error packet insertion unit 15 is determined at the start of communication, and can be changed for each communication. . Of course, if the size of the received packet is always constant, the size of the error packet may be determined at the time of installing the device equipped with the present method, and may be made constant thereafter. In this case, the threshold value TH of the packet size is referred to only when the device is installed.

The operation of the flowchart of FIG. 2 described above is repeated for each received packet and for each timeout period.

Finally, the threshold value TH having an important function for guaranteeing the quality of the audio output from the output terminal 5 will be described with reference to FIG.

FIG. 3 shows the relationship between the packet size SI and the sound quality evaluation QV when the receiving device 20 inserts an error packet having the same size as the packet size SI of the received packet.

In FIG. 3, when a large received packet such as the size SI2 is lost due to packet loss or the like and an error packet of the same size is inserted in that portion, the sound quality evaluation value becomes a low value such as QV2, and the sound quality is significantly deteriorated. It turns out that it becomes. This corresponds to the limit of the concealment function of the frame erasure described above, and the longer the portion without valid voice data (error packet section) becomes longer in time, the more the synthesized voice affects the human auditory characteristics, and the lower the quality is. It is also thought that this is because the degree of effect is significant.

On the other hand, in the case of a very small reception packet such as the size SI2, the sound quality evaluation value becomes an extremely good value such as QV1 even if an error packet of the same size is inserted.

However, if the size SI is too small, the ratio of overhead increases, so that the size S
In order to transmit the same amount of voice data within a unit time as before I was reduced, the required communication speed would increase.

This is because, in voice packet communication, a packet is formed by adding various header information to encoded voice data. However, since this header information is usually of a fixed size regardless of the packet size, This is because reducing the packet size is nothing less than reducing the proportion of voice data in all communication data.

For example, assuming that the header information size is 10, the packet size is 100, and the total audio data size to be transmitted is 9000, the required number of packets is expressed by the following equations (1) and (2). Thus, it becomes 100.

10 (header) + n (sound data amount) = 100 (packet size) (1) Thus, n = 90 (sound data amount included in one packet) Therefore, the required number of packets is 9000/90 = 100 (2) Under the same premise, if only the packet size is halved to 50, the number of packets required to transmit 9000 voice data is calculated from the following equations (3) and (4) as follows: It turns out that it is a packet.

10 (header) + n (sound data amount) = 50 (packet size) (3) Thus, n = 40 (sound data amount included in one packet) Therefore, the number of necessary packets is 9000/40 = 225 (4) Since the total data amount of the coded voice to be transmitted is the same for both packet sizes, the time for voice output on the receiving side is the same length.

Therefore, it is necessary to transmit and receive 100 packets of equation (2) and 225 packets of equation (4) at the same time. As compared with the case of communication, a communication speed of 1,125 times per data size and 225 times per packet number is required.

In general, an increase in communication speed is caused by a transmitting device,
In addition to increasing the cost of the receiving device and the communication network, the use of a communication network or the like for which a sufficient communication speed cannot be obtained is not preferable because it causes severe deterioration in sound quality.
In addition, since the increase in communication speed has a hardware limitation, the calculated design value cannot always be realized,
Even if it can be realized, there is a tendency that the faster the transmitting device and the receiving device, the larger and the more expensive the device.

As a result, it is difficult and disadvantageous to reduce the packet size of the received packet to prevent the communication quality deterioration phenomenon, and there is a certain limit.

Therefore, it is considered that it is better to select the threshold value TH near the illustrated size SIX in order to set no limit on the size SI of the received packet and to limit the size of the error packet.

Although the size SIX is considerably larger than the size SI1, the sound quality evaluation value QVX is good, and the level is almost the same as that of the QV1.

(A-3) Effect of the First Embodiment As described above, according to the present embodiment, even if the packet size of the received packet is reduced and the communication speed is not increased, the packet arrival order in the packet arrival order can be reduced. At the time of reversal, reception time-out, etc., it is possible to prevent remarkable sound quality deterioration due to skipping that occurred at the error packet insertion part, and to compensate for the loss of data amount, and to reduce the delay time of received packets received thereafter. It is also possible to control.

Further, in this embodiment, even if the reception buffer does not become empty, an error packet is inserted when the reception normality detection section outputs an abnormal reception signal. Increase.

(B) Second Embodiment In the following, only the points of this embodiment different from the first embodiment will be described.

In the first embodiment, when a normal reception state is continued, a certain reception packet and the next reception packet are sequentially transmitted according to the packet size of each reception packet received in time series. Is accumulated in the reception buffer 3 at the delay interval (reception packet interval), and forms a queue (queue). When the size determination signal DS corresponding to the equation (5) is output, it is inserted. Since the size of the error packet is smaller than the size of the original received packet, the interval between the received packet received after insertion and the received packet immediately before the inserted error packet becomes shorter than originally expected. .

This embodiment is characterized in that the accuracy of the delay time control function, which is not always sufficient in the first embodiment, is improved.

(B-2) Configuration and Operation of the Second Embodiment FIG. 4 shows the configuration of the receiving apparatus 21 of the present embodiment. This receiving device 21 is basically the receiving device 20 of the first embodiment.
It is a device corresponding to.

In FIG. 4, the receiving device 21 is
, Packet monitoring unit 2, reception buffer 3, decoding unit 4
, An output terminal 5 and a remaining frame control unit 6.

The functions of the components 1 to 5 other than the remaining frame control unit 6 are substantially the same as those in FIG.
It is the same as each part.

FIG. 8 shows the internal configuration of the packet monitoring unit 2 and the remaining frame control unit 6.

Here, the functions of the reception normality detection unit 11, packet size detection unit 12, threshold value setting unit 13, error packet size selection unit 14, and error packet insertion unit 15 constituting the packet monitoring unit 2 are substantially the same. Therefore, the same reference numerals are used as in FIG.

In FIG. 8, a portion surrounded by a broken line corresponds to the remaining frame control section 6.

That is, the remaining frame control unit 6 includes a size difference detection unit 30, an error frame continuation number determination unit 31,
An error frame insertion unit 32.

The size difference detection unit 30 compares the size determination signal DS received from the error packet size selection unit 14 with the size information SI received from the packet size detection unit 12 to determine the inserted error. A size difference (difference in the number of frames) between the packet size (corresponding to DS) and the packet size (corresponding to SI) of the received packet is detected, and the size difference information DD corresponding to the detection result is sent to the error frame continuation number determination unit 31. It is a part to supply.

That is, if the size difference to be obtained is set to Rem, the received packet size is set to Psize, and the inserted error packet size is set to Epsize, the size difference detection unit 30 performs the arithmetic operation represented by the following equation (7). .

Rem = Psize−Epsize (7) In other words, the size difference Rem indicates a delay time control error (DD) caused by error packet insertion performed by the error packet insertion unit 15.

In the error frame continuation number determining section 31, the size difference information DD (Rem) and the threshold value setting section 1
3 to determine the insertion pattern of the error frame with reference to the threshold value TH, the number of voice frames of the packet to be inserted, and the like. The inserted packet is a received packet arriving immediately after the error packet insertion unit 15 inserts an error packet (and a received packet arriving after that), and is a packet to which an error frame is to be inserted. The insertion operation is performed on the audio data part of the received packet arriving after the immediately following received packet when all the error frames cannot be completely inserted by only the immediately following received packet.

An error frame insertion unit 32 for performing an operation of inserting an error frame according to the number ME of consecutive inserted frames supplied from the number-of-continuous-error-frames determination unit 31 outputs an error frame of Perform an insert.

The error frame insertion unit 32 inserts the number of error frames specified by the number ME of consecutive inserted frames into the audio data portion of the packet to be inserted for each insertion operation, To accumulate. The insertion operation is performed once or multiple times for one inserted packet, and is repeated a number of times according to the size difference Rem.

The number ME of consecutive inserted frames needs to be set at least equal to or less than the threshold value TH from the viewpoint of sound quality, but need not be larger than the size specified by the size difference information DD. If the size specified by the size difference information DD (Rem) is larger than the threshold value TH, the number ME of consecutive inserted frames is set to the threshold value ME in order to eliminate the control error DD of the delay time as soon as possible. It may be set to the same value as the value TH.

When determining the insertion pattern, the error frame continuation number determining section 31 first detects the maximum number of consecutive insertion frames LE which is the maximum value of the number of error frames to be continuously inserted, and then detects the LE. The number ME of consecutive inserted frames may be determined from the following values.

The number ME of consecutive inserted frames may be the same value while the delay error DD caused by one error packet insertion is eliminated (one cancellation section). The value may be changed (however, the maximum value is LE) at each insertion.

If the packet to be inserted has many voice frames, the candidate positions (in the time axis direction) at which error frames can be inserted are large, and the delay error DD is relatively small, the LE is used. ME small enough
Can be continued over the one resolution section. However, if the number of candidate positions is small and the number of DDs is large, it may be necessary to continue the ME having the same value as LE over the one resolution section.

Usually, the positions on the queue where error frames can be inserted are limited (for example, the input terminal or the output terminal of the reception buffer 3, that is, the head or tail of the queue). In many cases, when trying to determine the insertion pattern of the error frame, it is not possible to completely eliminate the delay error DD. On the other hand, error frame insertion at the candidate position of the first packet to be inserted is not performed. There may be situations that must be done immediately. In such a case, the value of ME may be LE
To the same value as above, and from the time when the prospect of cancellation is established,
The value may be changed to a value smaller than LE.

Further, if necessary, it is possible to adopt a system configuration in which only a possible portion of all the delay errors DD is eliminated. Even in this case, the accuracy of the delay time control function is improved as compared with the first embodiment. For example, in a situation where the state of the network transmitting the received packet is bad and the error packet is frequently inserted in the receiving device 21, it is considered that such a system configuration has to be adopted.

In any case, the most preferable method for inserting the error frame in terms of speech quality is Rem.
This is to divide the amount of data for each frame and insert it into the next arriving packet as an error frame at arbitrary intervals or at equal intervals. This may be performed when there are many candidate positions into which the above-described error frame can be inserted and the delay error DD is relatively small.

The case where the Rem has 4 frames and the packet to be inserted has a packet size of 10 composed of 10 voice frames as shown in the following array String 1 will be specifically described as an example.

A, B, C, D, E, F, G, H, I, J... String1 Here, each alphabet of A to J indicates each audio frame contained in the inserted packet. I have.

Now, it is assumed that the number ME of consecutive frames to be inserted specifies one frame, each error frame to be inserted is (1), (2), (3), and (4).
Assuming that error frames (1) to (4) are inserted at equal intervals into ng1, the array String1 becomes the next array Stri
Converted to ng2.

(1) , A, B, C, (2) , D, E, F, (3) , G, H, I, (4), J... String2 Here, the inserted error frame (1) ) Is placed before A, which is the first audio frame, and contains an error frame.
(2) is inserted between successive audio frames C and D, error frame (3) is inserted between consecutive audio frames F and G, and error frame (4) is inserted
It is inserted between successive audio frames I and G.

This is an example in which error frames are inserted at equal intervals (three-frame intervals). However, the intervals do not necessarily have to be all equal intervals. By changing between frames H and I, a portion with a three-frame interval and a portion with a two-frame interval may be mixed.

The converted array String2 is accumulated in the reception buffer 3 in order from the leftmost frame (1), which is the first frame after the conversion, to form the tail of the queue.
After the last voice frame J, the voice frame (or error packet) of the next received packet is accumulated, and the same operation can be repeated thereafter.

In the array String2, the error frame (1) is arranged before the head audio frame A. However, in this insertion operation, the packet located immediately before the array String1 in the received packet sequence is an error packet. This is not preferable. This is because placing an error frame at this position is substantially equivalent to increasing the size of the error packet. However, if the packet located immediately before is the inserted packet and the last frame of the inserted packet is an audio frame (that is, Rem exceeds 4 and the array String2 is the second or later inserted packet). Packet case), there is no such problem.

Similarly, arranging an error frame after the last audio frame J in the array String2 is not a problem as long as normal received packets are stored immediately after that. If the following inserted packet is accumulated, restrictions are imposed on the size of the error packet and the insertion pattern of the error frame for the inserted packet.

In the array String2, since the number of consecutive error frames inserted is 1, as is apparent from FIG. 3, it is most preferable from the viewpoint of the sound quality of the sound output from the output terminal 5.

However, assuming that a three-frame interval is inserted, it may not be appropriate to execute such conversion when the Rem is too large. For example, if Rem is equal to or more than 5 frames, not all error frames can be inserted into the array String1.

In such a case, after the arrival of the received packet received next to the inserted packet containing the array String1, the error frame after the frame (5) is added to the audio data part of the received packet. Although it is conceivable to insert the threshold value, here, the threshold value TH is 2 or more and Rem is 5 to satisfy the above-described requirement of eliminating the control error of the delay time as soon as possible.
The array String1 is converted as in the following array String3, assuming that it is a frame.

A, B, C, (1), (2), D, E, F, (3) , (4), G, H, I, ( 5), J ... String3 In this array String3, an error occurs. Frames (1) and (2)
Is inserted between the audio frames C and D, the error frames (3) and (4) are inserted between the audio frames F and G, and the error frame (5) is inserted between the audio frames I and J. Only have been inserted. This array String
In No. 3, the insertion position of the error frame is also different from that of the array String2, but the insertion position may be appropriately changed according to the supply timing of the number ME of consecutive inserted frames.

In converting the array String1 into the array String3, the number ME of consecutive inserted frames received by the error frame inserting unit 32 from the number-of-continuous-error-frames determining unit 31 is represented by frames (1), (2) and (3). ), 2 when inserting (4), and 1 when inserting frame (5).

In the array String3, the number of consecutive error frames inserted is 2 or 1, but since the number of consecutive 2 is less than the threshold value TH, sound quality deterioration is suppressed to such an extent that the user hardly feels it.

Here, the converted arrays String2 and String3 are
The data is accumulated in the reception buffer 3 to form the tail of the queue.
You may make it supply to the decoding part 4. This is because the processing in the remaining frame control unit 6 only needs to be performed during the period in which the inserted packet is accumulated in the reception buffer 3, and is used for timing adjustment when the processing in the remaining frame control unit 6 takes time. It is advantageous.

In this case, too, the strings String1 to String2
Can be executed, and the conversion from array String1 to String3 can be executed.

(B-2) Effects of the Second Embodiment According to the present embodiment, the same effects as those of the first embodiment can be obtained.

In addition, in the present embodiment, the above-described control error of the delay time due to the insertion of the error packet can be completely eliminated, which is effective for preventing the audio interruption due to the exhaustion of the reception buffer and improving the sound quality deterioration. .

(C) Third Embodiment In the following, only differences between the present embodiment and the first and second embodiments will be described.

In the second embodiment, when inserting an error frame, no consideration is given to the contents of the audio frames before and after the insertion position. However, in the present embodiment, the quality degradation caused by the insertion of the error frame on the auditory characteristics is not considered. Focusing on the fact that the degree differs depending on the content of the adjacent audio frame, it is characterized in that the sound quality degradation at the error frame insertion portion is suppressed more effectively.

(C-1) Configuration and Operation of the Third Embodiment FIG. 5 shows the configuration of the receiving device 22 of the present embodiment. This receiving device 22 is basically the receiving device 2 of the first embodiment.
0, a device corresponding to the receiving device 21 of the second embodiment.

In FIG. 5, the receiving device 22 is
, Packet monitoring unit 2, reception buffer 3, decoding unit 4
, An output terminal 5, a remaining frame control unit 6, and a frame level identification unit 7.

The functions of the components 1 to 6 other than the frame level identification section 7 are substantially the same as those of FIG. 8 to which the same reference numerals are given.

FIG. 9 shows the internal configuration of the packet monitoring unit 2, the remaining frame control unit 6, and the frame level identification unit 7.

Components 11 to 15, 30, 31 in FIG.
Are substantially the same as those in FIG. 8 to which the same reference numerals are given.

In FIG. 9, a portion surrounded by a dashed line corresponds to the frame level identifying section 7.

That is, the frame level identification section 7 includes a silent frame detection section 40 and a sound pressure level detection section 41.

The silence frame detecting section 40 examines each voice frame contained in the received packet, detects the silence frame, and supplies the silence position NL indicating the position of the silence frame to the error frame insertion section 42. It is. A silence frame is a speech frame containing coded speech data that is valid as speech data but does not provide a significant speech output even when decoded.

As a function of determining a silence frame (silence portion), for example, a method of determining a root mean square of a waveform and determining that the value is smaller than a predetermined threshold value, or determining a variance value of the waveform and determining a variance value of the waveform, In such a case, there is a method of determining that there is no sound, a method of referring to a sound / silence identification code described in each frame data, and various other methods can be applied.

The last-mentioned sound / silence identification code described in each frame data is, for example, the above-mentioned G.100.
723.1 and G.P. In the case of encoding using a silence compression function (VAD function) in an encoding method such as 729, any of speech, silence, and transient sounds is described at the head of each frame data.

The sound pressure level detecting section 41 examines each sound frame contained in the received packet and detects a sound frame containing encoded sound data having a sound pressure level equal to or higher than a predetermined level. This is a part that supplies the large sound pressure position LL indicating the position of the audio frame to the error frame insertion unit 42. The sound pressure level is a loudness of an audio output obtained by decoding each audio frame.

The detection of the sound pressure level is performed, for example, by obtaining the mean square value of the waveform and comparing the value with a preset sound pressure level, in the same manner as the above-described determination function used for detecting the silent frame. be able to.

When an error frame is inserted into a silent section, a silent section is extended by the length of the inserted error frame, but it is practically impossible for a human to perceive it due to auditory characteristics.

When an error frame is inserted into a portion having a high sound pressure level, a masking effect by a large sound pressure immediately before the error frame makes it difficult for a human to perceive the sound in the error frame portion due to auditory characteristics. This masking effect occurs after a sound without sound pressure level (so-called small sound)
The effect is exhibited even when a sound with a high sound pressure level (a so-called loud sound) comes, and it is difficult for a sound with a high sound pressure level to be perceived as a sound without a previous sound pressure level.

Therefore, by selecting a position adjacent to a silent sound frame or a position adjacent to a sound frame of a high sound pressure level (high sound pressure level frame) as an insertion position of an error frame, deterioration of sound quality is reduced. It becomes difficult for the user to feel.

The function of the error frame insertion unit 42 is different from that of the error frame insertion unit 32 of the second embodiment described above. Detector 4
0 and the sound pressure level detector 4
Based on the large sound pressure position LL received from No. 1, an operation of inserting an error frame into the audio data portion of the above-described inserted packet is performed.

The frame level discriminating section 7 does not necessarily need to include both the silent frame detecting section 40 and the sound pressure level detecting section 41, and may include only one of them.

Further, since the frequency of occurrence of a silent frame or a high sound pressure level frame varies depending on the state of communication between users and the like, the high sound pressure level frame and the high sound pressure level frame in the voice frame accommodated in one inserted packet. The distribution state of silence frames is unknown. That is, in the audio frame accommodated in one inserted packet, there is a case where a high sound pressure level frame exists but no silence frame exists,
Conversely, there may be a case where there is a silence frame but no high sound pressure level frame, a case where neither a silence frame nor a high sound pressure level frame exists, or a case where both exist.

Here, the operation of this embodiment will be described by taking as an example a case where a silent frame exists but a high sound pressure level frame does not exist.

It is assumed that the inserted packet is composed of six audio frames as shown in the following array String4.

Silence 1, sound 2, sound 3, sound 4, sound 5, sound 6 ... String4 As a result of examining this array String4, the silence position NL output from the silence level detection unit 40 is silence 1 And the silence 4 is active, and the other positions are inactive. Further, in this example, since the sound pressure levels of the sound 2, the sound 3, the sound 5, and the sound 6 are not high, the large sound pressure position LL output from the sound pressure level detecting unit 41 has never been in the active state. It shall not be.

At this time, assuming that the number ME of consecutive frames to be inserted specifies 1, and that the size difference DD is 2 or more, the array String4 is changed to the next array String5 by the operation of the error frame insertion unit 42. Is converted to

Silence 1, sound 2, sound 3, (1) , sound 4, (2) , sound 5, sound 6... Trigger5 Here, one to the left and right of silence 4, a total of two Although error frames (1) and (2) have been inserted, it is also possible to insert any error frame on the right side of silence 1.

If the number ME of consecutive frames to be inserted is two, the array String4 can be converted to the next array String6.

Silence 1, sound 2, sound 3, sound 4, (1) , (2) , sound 5, sound 6... Trigger 6 Here, two consecutive inserted frames (1), (2) )
May be inserted to the left of silence 4 or to the right of silence 1.

In this embodiment, since a large sound pressure level such as a masking effect and the auditory characteristics of a silent part are used,
As compared to the second embodiment that does not use these, even if the value of the maximum consecutive number of inserted frames LE is increased, there is a high possibility that sound quality degradation will not be perceived. Therefore, the value of the maximum number of consecutive insertion frames LE may be determined by listening to the reproduced sound actually output from the output terminal 5.

It is to be noted that the converted strings String5 and String6 are accumulated in the reception buffer 3 to form the tail of the queue, as in the second embodiment.

As in the second embodiment, the converted arrays String5 and String6 may be directly supplied to the decoding unit 4 without being stored in the reception buffer 3. This is because the processing in the remaining frame control unit 6 and the frame level identification unit 7 only needs to be performed during the period in which the inserted packet is accumulated in the reception buffer 3.
This is advantageous for timing adjustment when the processing in the remaining frame control unit 6 and the frame level identification unit 7 takes time.

The receiving device 23 shown in FIG. 6 has a configuration corresponding to this case.

(C-3) Effects of Third Embodiment According to this embodiment, the same effects as those of the first and second embodiments can be obtained.

In addition, in the present embodiment, the error frame is inserted in consideration of the contents of the audio frame contained in the packet to be inserted. It is possible to eliminate the delay time control error (DD) more quickly by improving the second embodiment and by increasing the value of the maximum number of consecutive insertion frames (LE).

(D) Other Embodiments In the first to third embodiments, the functions of setting and monitoring the packet number, the timeout time, generating an error frame, and discarding the packet are integrated in the packet monitoring unit 2. However, all or some of the functions may be divided into a plurality of units.

In the first to third embodiments, many specific numerical values are shown for simplicity of explanation. However, these numerical values are merely examples, and the present invention is not limited to the numerical values described above. Can be widely applied.

Further, the configuration diagrams and operation flows shown in the first to third embodiments are merely examples for realizing the present invention. Therefore, in applying the present invention, it is sufficient that the element having the function described above is included somewhere.
Elements having functions other than these may be included.

In the first to third embodiments, the present invention is mainly realized by hardware, but the present invention can be realized by software.

[0143]

As described above, since the present invention detects whether or not each communication packet is normally received, it depends on whether or not the reception buffer becomes empty. Function to increase the possibility of preventing the occurrence of sound skipping and the like.

Further, according to the present invention, when the communication packet size is large, a correction packet smaller than the communication packet size is inserted. It is possible to suppress the deterioration of the sound quality at the insertion portion and maintain the communication quality.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of a packet monitoring unit of a receiving device according to first to third embodiments.

FIG. 2 is an operation explanatory diagram of the first to third embodiments.

FIG. 3 is a schematic diagram showing the relationship between packet size and sound quality evaluation when an error packet having the same size as the packet size of a received packet is inserted.

FIG. 4 is a block diagram illustrating a schematic configuration of a receiving device according to a second embodiment.

FIG. 5 is a block diagram illustrating a schematic configuration of a receiving device according to a third embodiment.

FIG. 6 is a block diagram illustrating a schematic configuration of a modified example of the receiving device according to the third embodiment.

FIG. 7 is a block diagram illustrating a schematic configuration of a receiving device according to the first embodiment.

FIG. 8 is a block diagram illustrating a schematic configuration of a main part of a receiving device according to a second embodiment.

FIG. 9 is a block diagram illustrating a schematic configuration of a main part of a receiving device according to a third embodiment.

[Explanation of symbols]

2 packet monitoring unit 3 reception buffer 4 decoding unit
6: remaining frame control unit, 7: frame level identification unit, 1
1 ... reception normality detector, 12 ... packet size detector,
13: threshold setting unit, 14: error packet size selection unit, 15: error packet insertion unit, 30: size difference detection unit, 31: error frame continuation number determination unit, 32, 42
... error frame insertion unit, 40 ... silence frame detection unit
41 ... Sound pressure level detection unit.

Continued on the front page F term (reference) 5K030 GA11 HA08 HB01 HB28 JT03 KA01 LA06 LB11 LC13 MA12 MB11 MB12 9A001 CC03 EE04 HH15

Claims (3)

[Claims] 1. A method for detecting whether or not each communication packet received in time series is normally received, based on an identifier assigned to each communication packet containing valid voice data on a transmission side. A packet receiving device that determines whether to insert a predetermined correction packet that does not contain valid voice data into a received communication packet sequence based on the detection result, Communication packet size detecting means for checking the communication packet size which is the size of
A packet receiving apparatus comprising: a correction packet inserting unit that selects a size smaller than the communication packet size as the size of the correction packet when at least the communication packet size exceeds the threshold. 2. The packet receiving apparatus according to claim 1, wherein, when the size of the correction packet inserted in the communication packet sequence is smaller than the communication packet size, a difference between the size of the correction packet and the size of the communication packet is reduced. A correction frame insertion unit for inserting a correction frame that does not contain valid audio data into a backward communication packet located after the correction packet in the communication packet sequence. Receiver. 3. The packet receiving apparatus according to claim 2, wherein the correction frame inserting unit checks valid audio data contained in the backward communication packet, and determines, from the audio data, a portion adjacent to the correction frame. And a correction frame for inserting the correction frame so that the inserted correction frame is adjacent to the non-visible data. A packet receiving device comprising an insertion control unit.