JPH03296346A - Voice packet recovery control method - Google Patents

Abstract

PURPOSE:To attain reproduction processing having a no problem in the voice quality by starting interpolation reproduction by predetermined packet number and reproducing a succeeding reception packet when a packet is received during the interpolation reproduction and discriminating it that the block is a non-voice block when the packet is not received till the end of time. CONSTITUTION:After one packet of an interpolation data 23 is transferred to a reproduction buffer 19, when a reproduction timing signal 18 is received from a reproduction timing generating section 17 and the packet is received, the reception packet reproduction processing is executed. When the reproduction timing signal 18 is received from the reproduction timing generating section 17, the interpolation reproduction by a predetermined packet number is finished and when no packet is received till that time, one packet for a non-voice block reproducing signal 25 generated in a non-voice block reproducing signal generating section 24 is transferred to the reproducing buffer 19 and the reproduction sequence number is incremented by 1.

Description

[Detailed description of the invention]

[Industrial Application Field] The present invention digitizes an analog audio signal on the transmitting side of a communication device, packetizes only the active sections of the digitized sample sequence, and sends the packets to the transmission path. Extract the sample sequence of the sound interval from the packet,
The present invention relates to an audio packet playback control method for converting this into an analog audio signal and inserting the signal into a silent section. [Prior Art] The voice packet method for performing voice communication over a packet communication network has the following advantages: voice communication and data communication can be economically integrated in one communication network, and because the network is asynchronous, the voice encoding speed can be reduced. There are no constraints on the selection, and D S I (D
1g1tal S peach Interpo
By applying the ration), there are advantages such as the ability to improve the usage bandwidth of the line. In audio packet communication, audio signals are divided into blocks at regular intervals, destination information (transfer header) is added to create audio packets, and the packets are sent to a transmission path. Since packet communication networks are basically asynchronous, the delay time for a packet to reach a destination terminal always fluctuates. On the receiving side, this is corrected and the audio is played back at a constant speed. Also, if a packet is lost due to a coding error in the transmission path, or
When the S technique is applied, the missing packets are reproduced as silence or pseudo-voice. In addition, in voice packet communication, packets may be lost due to coding errors in the transmission path or congestion in the packet switch, and packets with excessive delays may be discarded.
Furthermore, overruns and underruns occur. An underrun packet is a packet that does not arrive at the receiving side by the time when the next packet is to be played back, but arrives late. On the other hand, an overrun packet is a packet that arrives too early. In overrun, packet i+1 is played while packet i is being played back.
packet i+2 is received. In the case of an overrun, since packet i+2 is received when the reproduction of packet i ends, if there is a sequence number, it can be identified that packet 1+1 is missing. On the other hand, in underrun, the next packet i+1 is not received when the reproduction of packet 1 is finished, and packet i+1 is received after a certain period of time has passed. The occurrence of an underrun is not known until the playback of packet i is finished, and packet! It will only become clear when +1 is received. Note that voice packets are described, for example, in "Electronic Information and Communication Handbook," Volume 2, March 30, 1985, published by Ohmsha (C stock), pp. 2314-2315. [Problems to be Solved by the Invention] As mentioned above, on the receiving side, if the next packet is not received at the time when the reproduction of packet l is finished, it is difficult to determine whether the packet was discarded or whether an underrun occurred. Since it is not possible to determine whether a silent section has occurred, there is a problem in that it is unclear which processing to perform among discarded packet processing, silent section reproduction processing, or underrun processing. An object of the present invention is to solve such conventional problems, and to make it possible to perform playback processing without causing problems in voice quality when the next packet has not been received at the time when the playback of a received packet is finished. An object of the present invention is to provide a voice packet playback control method. [Means for Solving the Problems] In order to achieve the above object, the audio packet playback control method of the present invention provides a method for controlling audio packet playback of the present invention, in which, on the receiving device side, after adding a delay to the first packet of communication, the sequence number of the first packet is Each time a packet in a sound section is played back, each time one packet is interpolated and played back for discarded packets and underrun packets, and each time the packet assembly time elapses in a silent section, the playback sequence number on the receiving side is and compared the sequence number of the next playback candidate packet with the lowest sequence number other than the packet being played back out of the packets being received, and the playback sequence number of the packet being played back. If there is, play the next playback candidate packet following the packet being played back, and if the former is larger by 2 or more, play the packet after performing interpolation playback for the number of missing packets, If the two are equal or the former is smaller, discard the former packet, and if no packet has been received at the time of finishing playing the last packet among the packets received so far, Start interpolation playback for a predetermined number of packets, increment the playback sequence number by 1 every time interpolation playback is performed for one packet, and check if a packet is being received during interpolation playback and the sequence number of the received packet is As a result of comparing the playback sequence numbers, if the two are consecutive, the former packet is played following the currently playing packet, and if the former is larger, interpolation playback is performed for the number of missing packets. If the two packets are equal or the former is smaller, the former packet is discarded, and packets for a predetermined number of packets are received by the end time of interpolation reproduction. If not, start playing the signal for the silent section in units of packet length, and increase the playback sequence number by 1 every time the signal for the silent section is played for one packet. is being received, and as a result of comparing the sequence number of the received packet with the playback sequence number currently being played, if r4# is consecutive, the former packet will continue to the signal for the silent section being played. If the former is larger than 2 or more, play back the signal for the silent section for the number of missing packets, then play this packet, and if the two are equal or the former is smaller, then For example, the feature is that the former packet is discarded. [Operation] In the present invention, interpolation playback for a predetermined number of packets is started, and if a packet is received during interpolation playback, it is determined from the sequence number that a packet has been discarded or underrun. , the subsequent received packets are played back, and if no packets have been received by the end time of corrected playback for a predetermined number of packets, it is determined that a silent section has arrived, and a signal for the silent section is played back. . Thereby, even if the next packet has not been received at the time when the reproduction of the received packet is finished, reproduction processing can be performed without causing any problems in terms of audio quality.

【Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is an overall configuration diagram of a voice packet transmitting/receiving device showing an embodiment of the present invention, in which (a) is a transmitting side device;
) is the receiving device. First, the operation on the transmitting side will be described. Analog signal input terminal 1
The analog signal input from the low-pass filter section 2
After the high-frequency components are cut off, the signal is converted into a digital signal by an analog-to-digital converter (ADC) section 3. The converted digitized sample sequence 4 is transferred to a voice detection section 5 and a packet assembly section 6. The voice detection section 5 uses the sample sequence 4 to determine whether there is a sound section or a no-speech section, and transfers the determination result to the sequence number management section 8 and the packet assembly section 6 as a voice/non-sound determination signal 7. In addition, methods for determining whether there is a voiced section or a silent section include a method in which a voiced section is determined when the short-term average power of the voice exceeds a threshold value, and a method in which this is determined in combination with the number of zero crossings. etc. The packet creation timing generating section 9 outputs a packet sending timing signal 10 every time a packet creation time Δ elapses based on the time when the judgment result 7 of the voice detection section 5 changes from sound to silence after the start of communication. . The sequence number management unit 8 starts increasing the sequence number from zero from the time when the judgment signal 7 of the voice detection unit 5 becomes audible for the first time after the start of communication, and receives the packet sending timing signal lO from the packet creation time timing generation unit 9. The sequence number is incremented each time it is received, and the result is output to the packet assembling unit 6 as sequence number information 11. Here, it is assumed that the sequence number is 8 bits, and if the packet transmission timing signal 10 is received when the sequence number reaches 255, the sequence number is reset to zero. In the packet assembling unit 6, the packet sending timing signal 1
If the judgment result of the voice detecting unit 5 is that there is a sound at the time when 0 is received, the sample sequence is divided into voice blocks of a certain length, and then the sequence number information 11 received from the sequence number management unit 8 is added. A voice packet is created and sent to the transmission path 12. On the other hand, if the voice detection unit 5 determines that there is no sound at the time when the packet transmission timing signal 10 is received, the voice sample in the silent period is not transmitted. FIG. 2 is a processing timing chart on the transmitting side in FIG. 1(a). Here, the time required for internal processing is set to zero. Second
Figure (a) shows the output of the analog-to-digital converter 3, that is, a digitized sample sequence. FIG. 2(b) shows the output of the voice detection section 5, that is, the voice/silence determination output. FIG. 2(C) shows the output of the packet creation timing generator 9, that is, the packet sending timing signal. The interval between each signal is the packet creation time Δ. FIG. 2(d) shows a packet sent out from the packet assembly unit 6. FIGS. 5(a) and 5(b) are operation flowcharts on the receiving side. Next, the operation on the receiving side in FIG. 1(b) will be explained according to FIG. On the receiving side, packets received from the transmission path 13 are temporarily stored in the packet receiving buffer 14. If the received packet is the first packet of communication (step 503), the reproduction control unit 15 sets the delay fluctuation absorption time of the reproduction timing generation unit 17 using the time setting instruction signal 16 (step 507). The reproduction control unit 15 also reads the sequence number from the packet reception buffer 14 and sets the sequence number as the initial value of the reproduction sequence number (step 506).
Here, the playback sequence number is 8 bits as on the transmitting side, and if an increase request signal is received when the playback sequence number is 255, the playback sequence number is reset to zero. When the playback timing generation unit 17 receives the time setting instruction signal 16 from the playback control unit 15, it sets the instructed time in an internal timer, and controls the playback timing signal 18 to playback every time the set time elapses. Send to Department 15. After setting the delay fluctuation absorption time in the playback timing generation section 16, the playback control section 15 receives the playback timing signal 18 and then sets the packet creation time in the playback timing generation section 17 using the time setting instruction signal 16. at the same time,
The audio sample string of the first packet in the packet reception buffer 14 is transferred to the reproduction buffer 19. Next, after setting the packet creation time in the reproduction timing generation section 17, the reproduction control section 154 checks whether there is a received packet every time the reproduction timing signal 18 is received (step 501).
If there is a received packet (step 508), the received packet regeneration procedure (described later) is executed in step 50.
9) On the other hand, if there is no received packet, a correction reproduction process (described later) is executed (steps 502 and 504).
. The received packet reproduction process will be explained. If there is a received packet at the time when the reproduction timing signal 18 is received, the sequence number Sn of the next reproduction candidate packet is read from the packet reception buffer 14, this is compared with the current reproduction sequence number Sp, and reproduction processing is executed. . If the reproduction sequence number Sp and the sequence number Sn of the next reproduction candidate packet are consecutive (step 508), the audio sample string of the next reproduction candidate packet is transferred from the packet reception buffer 14 to the reproduction buffer 19, and at the same time,
The playback sequence number is also increased by 1. If the sequence number of the next playback candidate packet is 2 or more larger than the current playback sequence number (step 5
12) The interpolation instruction signal 21 is sent to the interpolation processing unit 20 to instruct the interpolation processing unit 20 to create interpolation data for the number of missing packets (Sn-Sp-1) (step 513). 31st! 1 is a reproduction timing chart of received packets in FIG. 1(b). Here, the time required for internal processing is assumed to be zero, and FIG. 3 shows a case where the sequence number of the next reproduction candidate packet is 8 at the time when the reproduction of the packet with the reproduction sequence number 5 is completed. The interpolation processing unit 20 constantly accumulates audio samples for a plurality of past packets. When the interpolation instruction signal 21 is received from the playback control section 15, a sequence of audio samples for interpolation is created for the number of packets instructed. As a method of creating a voice sample string for interpolation, there is a method of repeating the previous packet for a specified number of packets. When the interpolation processing is completed, the interpolation processing section 20 outputs an interpolation processing end signal 22 to the reproduction control section 15. Playback control section 15
When receiving the interpolation processing end signal 22, the interpolation processing section 2
1 of the sample strings (interpolated data 23) created in
The packet is sent to the playback buffer 19, and the playback sequence number is incremented by 1. The reproduction control unit 15 (Sn
-Sp-1) Each time the reproduction timing signal 18 is received from the reproduction timing generator 17, the sample sequence for the next l packets of the interpolated data 23 is stored in the reproduction buffer 19 until the interpolation for Sp-1) is completed. At the same time as the transfer, the playback sequence number is incremented by 1. The reproduction control unit 15
When the interpolation for (Sn-3p-1) pieces is completed, the audio sample string of the next reproduction candidate packet is transferred to the reproduction buffer 19. If the sequence number Sn of the next reproduction candidate packet is smaller than or equal to the reproduction sequence number Sp (step 510), among the received packets,
If there is a packet whose sequence number is consecutive to the playback sequence number Sp (sequence number Sp+1), this packet is transferred to the playback buffer 19, and at the same time, the playback sequence number is incremented by 1, so that the sequence number becomes the playback sequence number Sp. The following received packets are discarded (step 511). If the sequence numbers of all the received packets are less than or equal to the reproduction sequence number Sp, all the received packets are discarded (step 511).
11) Perform the following interpolation reproduction processing. Next, interpolation reproduction processing will be described. FIG. 4 is a packet reproduction timing chart of interpolation reproduction processing among the receiving side processing of FIG. 1(b). When the reproduction control unit 15 detects that among the packets received so far, there are packets that have not been received at the time when the reproduction of the last packet is finished (step 512), the reproduction control unit 15 generates interpolated samples for a predetermined number of packets. The interpolation processing unit 20 is instructed to create the following using the interpolation instruction signal 21 (step 513). FIG. 4 shows packet reproduction timing when the predetermined number of interpolated packets is two. When the interpolation processing unit 20 receives the interpolation instruction signal 21 from the playback control unit 15, the interpolation processing unit 20 generates the complementary 1SI for the number of packets instructed.
A voice sample string for month II is created, and when the creation is completed, an interpolation processing end signal 22 is inputted to the reproduction control section 15. When the reproduction control unit 15 receives the interpolation processing end signal 22, the reproduction control unit 15 reproduces the sample sequence (interpolation data 23) created by the interpolation processing unit 2o.
Of the packets, one packet is transferred to the playback buffer 19, and the playback sequence number is incremented by one. After transferring one packet of interpolated data 23 to the playback buffer 19, at the time when the playback timing signal 18 is received from the playback timing generator 17, if a packet is being received, the above-mentioned received packet playback process is performed. Execute(
Step 514). That is, the sequence number Sn of the next reproduction candidate packet is read from the packet reception buffer 14, and the reproduction process is performed by comparing this with the current reproduction sequence number Sp. Furthermore, after transferring the ] packet of interpolated data 23 from the interpolation processing unit 20 to the reproduction buffer 19, at the time when the reproduction timing signal 18 is received from the reproduction timing generation unit 17, no packet has been received (step 505),
In addition, if the predetermined number of packets have not been interpolated and reproduced (step 515), the next one packet worth of audio samples from the interpolation month sample sequence created by the interpolation processing unit 20 is transferred to the reproduction buffer. 19, and the playback sequence number is incremented by 1. When the playback timing signal is received from the playback timing generating section 17, interpolation playback for a predetermined number of packets has been completed (step 515), and if no packets have been received by this time, the silent section 1 of the silent section reproduction signal 25 generated by the reproduction signal generation section 24
Step 5: Transfer the packets to the playback buffer 19.
16), increase the playback sequence number by 1; After transferring one packet of the silent section reproduction signal 25 to the reproduction buffer 19, at the time when the next reproduction timing signal is received from the reproduction timing generation section 17, if no packet has been received (step 517), silence is generated. The next one packet of the section reproduction signal 25 is transferred to the reproduction buffer 19, and the reproduction sequence number is incremented by one. Further, after transferring one packet of the silent section reproduction signal 25 to the reproduction buffer 19, the reproduction timing generating section 17
At the time when the next playback timing signal 18 is received from , if the bucket has been received (step 5] 7), the sequence number Sn of the next playback candidate packet is read from the packet reception buffer 14, and the current playback sequence number is read from the packet reception buffer 14. S
If the sequence number Sn of the next reproduction candidate packet is smaller than or equal to p (step 520)
, among the received packets, the reproduction sequence number S
If there is a packet whose sequence number is consecutive to p (sequence number Sp+1), this packet is transferred to the playback buffer 19 and the playback sequence number is changed to 1.
, and discard received packets with sequence numbers less than or equal to Sp (step 521). If the sequence numbers of all the received packets are less than or equal to Sp (step 520), all the received packets are discarded (step 521), and the next one of the silent section reproduction signals 25 is discarded. The packet is transferred to the playback buffer 19, and the playback sequence number is incremented by 1. Further, if the reproduction sequence number Sp and the sequence number Sn of the next reproduction candidate packet are consecutive (step 518), the audio sample string of the next reproduction candidate packet is transferred from the packet reception buffer 14 to the reproduction buffer 19 at the same time. (Step 519), the reproduction sequence number is incremented by one. If the sequence number Sn of the next reproduction candidate packet is greater than the current reproduction sequence number Sp by 2 or more (step 522), the reproduction control unit 15 controls the number of missing packets (Sn-3p-141). Until the silent section reproduction ends, each time the reproduction timing generating section 17 receives the reproduction timing signal 18, it simultaneously transfers one packet worth of sample string of the silent section reproduction signal 25 to the reproduction buffer 19 (step 523). , increments the playback sequence number by one. When the playback of the silent section of (Sn-Sp-111 minutes) is completed, the playback control unit 15 transfers the audio sample sequence of the next playback candidate packet to the playback buffer 19 (step 52
4). The audio sample output of the playback buffer 19 is transferred to a digital-to-anlog converter (DAC) 26.
The unit 26 converts the sent sample string into an analog audio signal. This analog audio signal is sent to the low-pass filter section 27 and then output from the analog signal output terminal 28. In this way, in this embodiment, on the transmitting side. Each time a packet is assembled in a sound section, and each time the packet assembly time elapses in a silent section, the sequence number is incremented and added to the packet, and the receiving side delays only the first packet of communication. In addition, the playback sequence number is managed, and if no packet has been received by the time the playback of the last packet among the packets received so far is finished, interpolation playback for a predetermined number of packets is performed. If a packet is received during interpolation playback, it is determined from the sequence number that the packet has been discarded or there has been an underrun, and subsequent received packets are played back, and the number of packets determined in advance is If no packet has been received by the end time of interpolation playback of minutes, it is determined that a silent section has arrived, and the signal for the silent section can be reproduced. [Effects of the Invention] As explained above, according to the present invention, even if the next packet has not been received at the time when the reproduction of a received packet is finished, it is possible to perform reproduction processing that does not affect audio quality. Even if packets are discarded or underrun, high-quality voice communication can be achieved.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of a transmitting side and a receiving side showing an embodiment of the present invention, FIG. 2 is a processing timing chart of the transmitting side in FIG. 1, and FIG. 3 is a reproduction timing of received packets in FIG. 1. 4 is a packet reproduction timing chart of the interpolation reproduction process in FIG. 1, and FIG. 5 is an operation flowchart on the receiving side in FIG. 1. 1: Analog signal input terminal, 2: Low-pass filter, 3
:Analog-digital converter, 4: Sample sequence, 5:
Voice detection unit, 6: Packet assembly unit, 7: Speech/no-sound determination signal, 8 Number management unit, 9: Packet creation timing generation unit, IO: Packet sending timing signal, 1
1 sequence number information, 12, 13: transmission line, 14:
Packet reception buffer, 15: playback control section, 16: time setting instruction signal, 17: playback timing generation section, 18: playback timing signal, 19: playback buffer, 20: interpolation processing section, 21: interpolation instruction signal, 22: Interpolation processing end signal, 23: interpolation data, 24: silent section replay signal generation section, 25: silent section reproduction signal, 26: digital-to-analog conversion section, 27: low-pass filter section, 28: analog signal output terminal.

Claims (1)

[Claims] (1) After the transmitting device digitizes the analog audio signal and determines whether the digitized sample sequence is voiced or not, the packet assembly time elapses each time a packet is assembled in a voiced period and in a silent period. Each time, the sequence number is increased, and the sample string in the silent section is not sent, and the sequence number is added to the packet in the sound section.
By transmitting it to the transmission path, and converting the received sample string of packets in the sound section into an analog audio signal on the receiving device side, as well as interpolating discarded packets and underrun packets, and inserting signals into the silent section, In a voice communication system that performs communication, the receiving device side adds a delay to the first packet of communication, then starts increasing the sequence number from the first packet, and discards it every time a packet in a sound section is played back. Each time interpolation playback of packets and underrun packets is performed, and each time the packet assembly time elapses in a silent section, the playback sequence number on the receiving side is incremented, and among the packets currently being received, the packet being played back is As a result of comparing the sequence number of the next playback candidate packet with the lowest sequence number and the playback sequence number of the currently playing packet, if both are consecutive, the next playback candidate packet is If the former is larger than 2, perform interpolation playback for the number of missing packets, then play that packet, and if the two are equal or the former is smaller, then the former is If the packet is discarded and no packet has been received at the time when the last packet among the packets received so far is finished, interpolation playback for a predetermined number of packets is started, and the interpolation playback is performed. The playback sequence number is incremented by 1 every time 1 packet is performed, and if a packet is being received during interpolation playback, and the sequence number of the received packet is compared with the playback sequence number, if the two are consecutive. For example, play the former packet following the packet being played back, and if the former packet is larger, perform interpolation playback for the number of missing packets, then play this packet, and check whether the two are equal or not. If the former is smaller, the former packet is discarded, and if no packets have been received by the end time of interpolation playback for a predetermined number of packets, the silent section signal is played back in units of packet length. , and set the signal for the silent section to 1.
The reproduction sequence number is incremented by 1 for each packet, and the packet is received while the silent section signal is being reproduced, and the sequence number of the received packet is compared with the reproduction sequence number being reproduced. If the two are consecutive, the former packet is played following the signal for the silent section that is being played back, and if the former is larger by 2 or more, the signal for the silent section is played for the number of missing packets. A voice packet reproduction control method characterized in that after performing reproduction, this packet is reproduced, and if the two packets are equal or the former packet is smaller, the former packet is discarded.