JP2005197850A - Jitter absorbing method and apparatus for voice ip terminal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To overcome the problem of a conventional IP phone that sounded voice data are often expanded or reduced in the case of absorbing jitter resulting in causing remarkable deterioration in the speech quality because a jitter buffer 10 is used to absorb missing of voice packets or variations (jitter) in the arrival time of the packets and the jitter is absorbed by expanding or reducing the buffer. <P>SOLUTION: A voiced/silence state of a received packet 31 is identified (6), a flag is attached to the silence packet (7) and the resulting packet is stored in the jitter buffer 10. Data with the silence flag are monitored (12), a silence period is detected and the buffer is expanded/reduced (13, 14) during the silence period. Thus, it is possible to prevent deterioration in the speech quality. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a terminal (hereinafter referred to as a voice IP terminal) device that is accommodated in an IP (Internet Protral) network (hereinafter abbreviated as IP network) and performs a voice call. A novel method that enables voice communication with good voice quality using a jitter buffer that absorbs packet loss and packet arrival time delay that occur when a voice IP terminal transmits and receives voice packets; A device is provided.

  Voice IP terminals that perform voice calls over an IP network depend on the congestion level of the IP network because the time it takes to send and receive voice packets varies, causing packet loss and packet arrival time delays, resulting in degraded voice quality May occur. The voice IP terminal is provided with a jitter buffer to absorb the delay in order to prevent a time deviation from transmission to reception of the voice packet, that is, deterioration of voice quality when jitter occurs.

  FIG. 14 shows a circuit configuration of a conventional voice IP terminal device. Packets from the IP network are received by the packet receiver 5 via the interface (I / F) 2. The packet receiving unit 5 outputs the received packet 31 to the packet sound / silence identifying unit 6B. Receiving this, the packet sound / silence identifying unit 6B identifies whether the received packet 31 is a sound including a sound of a predetermined level or higher or a sound of a sound level lower than that level. The identification result is reported to the main processing unit 20B by a voice / silence identification signal 33B. Some received packets 31 include a next data pointer, BFI data, voice / silence identification flag, sequence number, and provisional silence data (for example, Patent Document 5).

  The received packet 35B after identification is sent to and stored in the jitter buffer 10B via the signals 51 and 52 under the monitoring of the address management unit 11 and the main processing unit 20B. The jitter buffer 10B receives the processing of the buffer contents via the signals 41, 42B, 43B, and 44B under the monitoring of the address management unit 11, the buffer monitoring unit 12B, the reduction processing unit 13B, and the expansion processing unit 14B, respectively. Yes. The jitter buffer 10B is a FIFO (First In First Out) type memory.

  The address management unit 11 manages the address of the audio data stored in the jitter buffer 10B. The buffer monitoring unit 12B monitors the contents of a plurality of buffers included in the jitter buffer 10B. The reduction processing unit 13B executes a reduction process for reducing the number of buffers included in the jitter buffer 10B. The decompression processing unit 14B increases the number of buffers included in the jitter buffer 10B, and executes decompression processing for storing pseudo silence data in the increased buffer.

  The pseudo silence data unit 15B generates silence data to be stored in the buffer increased by the decompression processing unit 14B and sends it to the decompression processing unit 14B as pseudo silence data 46B. The data reproducing unit 23B is notified. The jitter buffer 10B sends the output as the buffer output 36B to the data reproducing unit 23B.

  The data reproduction unit 23B sends reproduction data 49 including pseudo silence data to an interface (I / F) 3 that interfaces with a voice terminal. The timer 21 supplies the clock 40 to necessary units in the apparatus including the main processing unit 20B. The counter 22 counts the number of unstored buffers and the number of sound data included in the jitter buffer 10B via the main processing unit 20B, or stores the data in this buffer from the setting of the unstored buffer. The period until is counted, and the result is reported to the main processing unit 20B.

  FIG. 15 is a waveform diagram showing the principle of the conventional example shown in FIG. The audio data of the received packet 31 is shown as a waveform. (A) shows a case where all received packets 31 are normally received without jitter. Nos. 1 to 8 of the received packet 31 represent voiced data, and Nos. 9n, 10n, and 11n represent silent data. (B) shows a case where jitter is generated in the received packet 31, the buffer is expanded, and pseudo silence data 46B is inserted as In between Nos. 2 and 3 of the received packet 31.

  (C) shows a case where jitter is generated in the received packet 31 and data No. 3 in the received packet 31 is deleted to reduce the buffer. As apparent from the cases of (b) expansion and (c) reduction, when the expansion or reduction timing is performed during the period of a voice packet, an impulse change often occurs in the reproduced voice waveform.

  FIG. 16 is a buffer operation diagram showing an example of buffer expansion in the jitter buffer 10B (b). In state 1, an unstored buffer for accepting No. 5 received packet 31 (a) is set (BFI setting), and No. 1 (sound) reproduction data 49 (c) is output. However, the received packet 31 of No. 5 has arrived before State 1, and remains unstored even after waiting for arrival of State 2, State 3 and No. 5.

  Even if the received packet 31 (a) of No. 6 is received there, the unstored buffer (6) for accepting No. 6 is not set yet because the arrival is too early (BFI not set). Therefore, even if the arrival of state 5, state 6 and No. 6 is awaited, the unstored buffer remains. States 1 to 6 are all unstored buffers, and this situation is counted by the counter 22. In state 7, buffer a is added to the future side of buffer A (upper side of A) and expanded to No. 7. Waiting for arrival, No. 7 has already arrived, and this is also left unstored.

  If buffer b is expanded to the future side of buffer a in state 8 not shown, No. 8 can be stored. In this way, when a packet that is expected to be received at a certain period cannot be received at that period, BFI setting is performed (states 1, 4, and 7). When all the buffers in the jitter buffer 10B become unstored buffers, the counter 22 starts counting, and buffer expansion is performed when the value reaches a predetermined value (state 7 or FIG. 15 (b)). )).

  FIG. 17 is a buffer operation diagram showing an example of buffer reduction in the jitter buffer 10B (b). The state 1 is a state waiting for the No. 8 received packet 31 (a). Therefore, when No. 8 is received, state 2 is entered, and No. 8 received packet 31 is stored in packet A. In the state 3, the unstored buffer setting (BFI setting) for accepting the No. 9 received packet 31 is set, and simultaneously, the No. 1 (sound) reproduction data 49 (c) is output. Even when the state 4 is reached, the arrival of No. 9 is not made and the unstored buffer remains, but immediately after that, the No. 9 received packet 31 is received and stored in the buffer A.

  Therefore, when the state is shifted to the state 6, the BFI setting for accepting No. 10 is executed for the buffer A at the same time. The reproduction data 49 (c) of No. 2 (sound) is output. Even in state 4, No. 9 has not arrived and the unstored buffer remains, but waits until a predetermined period (clock) elapses.

  Here, as shown in the states 2 and 5, when the counter 22 counts and detects that all the buffers A to H are filled with the sound data without missing (a smaller number of buffers is sufficient), the buffers A buffer reduction process for deleting F, G, and H is performed to enter a state 8 (see FIG. 15C). In this state, the received packet 31 of No. 10 is received and stored in the buffer A. In the state 9, the number of buffers is smaller than in the state 1, and the time from the received packet 31 to the reproduction data 49 is shortened.

If the packet arrives smoothly within a predetermined period and there is no packet that has not arrived, all the buffers are filled with valid data. Therefore, the counter 22 counts that period to a predetermined value. When it reaches, the buffer reduction processing is executed (FIG. 15C).
JP 2000-286886 A JP 2000-349822 A JP 2002-64545 A JP 2002-185498 A JP 2002-271388 A

  In order to cope with delay variation (jitter) in the IP network, buffer expansion is performed on the future side of the jitter buffer (state 7 in FIG. 16), or the past side of the jitter buffer is reduced (see FIG. 17). State 8). As described with reference to FIG. 15, it is highly likely that the expansion or contraction timing is performed during the period of a voice packet, and at that time, there is often an impulse change in the reproduced voice waveform. This has a problem to be solved that the quality of reproduced sound is significantly deteriorated.

  The main feature of the present invention is to solve the above problems. Identify whether the voice data of the received packet received from the IP network is sound larger than a predetermined value or silence smaller than a predetermined value, and if it is identified as silence, a silence flag is added, When the buffer is required to be expanded or contracted, the expansion or contraction process is executed in the silence period in which the buffer with the silence flag exists.

  Since the expansion or contraction process in the jitter buffer is surely executed during the silence period, the reproduced voice has no sense of incongruity and an excellent call quality can be obtained.

  Identify whether the voice data of the received packet received from the IP network is voiced greater than a predetermined value or silence less than a predetermined value, and if it is identified as silence, a silence flag is added, When the buffer is required to be expanded or contracted, the expansion or contraction process is executed in the silence period in which the buffer with the silence flag exists.

    FIG. 1 shows a circuit configuration of a voice IP terminal device of the present invention. This figure corresponds to FIG. 14 showing a conventional example. Here, the same symbols are used for components corresponding to the components shown in FIG. Packets from the IP network are received by the packet receiver 5 via the interface (I / F) 2. The packet receiving unit 5 outputs the received packet 31 to the packet sound / silence identifying unit 6.

  Receiving this, the packet sound / silence identifying unit 6 attaches a packet number (No.) to the received packet 31 and at the same time, whether it is a sound including a sound of a predetermined level or higher, or a sound level lower than that level. Is identified. The identification result is applied to the silence flag adding unit 7 along with the received packet 32 having the packet number by the voice / silence identification signal 33. The silent flag adding unit 7 adds a flag to the received packet 32 identified as silent and sends it to the jitter buffer 10 as a received packet 35 with flag. At the same time, the silence flag adding unit 7 reports the addition of the flag to the main processing unit 20 by the silence flag addition notification 34.

  The flagged reception packet 35 is sent to and stored in the jitter buffer 10 via the signals 51 and 52 under the monitoring of the address management unit 11 and the main processing unit 20. The jitter buffer 10 receives buffer contents under the monitoring of the address management unit 11, the silence flag / buffer monitoring unit 12, the silence reduction processing unit 13, and the silence expansion processing unit 14 via signals 41, 42, 43, and 44, respectively. Have been processed. The jitter buffer 10 is a FIFO (First In First Out) type memory.

The address management unit 11 manages the address of audio data stored in the jitter buffer 10. The silence flag buffer monitoring unit 12 monitors the contents of a plurality of buffers included in the jitter buffer 10 in order to detect a silence period. The silence reduction processing unit 13 executes a silence reduction process for reducing the number of buffers included in the jitter buffer 10. The silent expansion processor 14 increases the number of buffers included in the jitter buffer 10 and executes a silent expansion process for storing pseudo silent data in the increased buffer.

  The pseudo silence data unit 15 generates silence data to be stored in the increased buffer by the silence extension processing unit 14 and sends it to the silence extension processing unit 14 as pseudo silence data 46. The data reproduction unit 23 is notified by a signal 45. The jitter buffer 10 sends the output to the data reproducing unit 23 as the buffer output 36.

  The data reproduction unit 23 sends reproduction data 49 including pseudo silence data to an interface (I / F) 3 that interfaces with a voice terminal. The timer 21 supplies a clock 40 to the main processing unit 20 and other necessary parts in the apparatus. The counter 22 counts the number of unstored buffers and the number of sound data included in the jitter buffer 10 via the main processing unit 20, or stores them in this buffer from the setting of unstored buffers. Or the period from the setting of the unstored buffer to the storage in this buffer is counted, and the result is reported to the main processing unit 20.

  FIG. 2 is a waveform diagram corresponding to FIG. 15 of the conventional example and showing the principle of the circuit configuration example of the present invention shown in FIG. The audio data of the received packet 31 is shown as a waveform. (A) shows a case where all received packets 31 are normally received without jitter. Nos. 1 to 8 of the received packet 31 represent voiced data, and Nos. 9n, 10n, and 11n represent silent data. (B) shows a case where jitter is generated in the received packet 31, the buffer is expanded, and pseudo silence data 46 is inserted as In into silence intervals of No. 9n and 10n of the reception packet 31.

  (C) shows a case where jitter is generated in the received packet 31 and No. 10n (silence) data of the received packet 31 is deleted to reduce the buffer. As is clear from the case of silence expansion in (b) and silence reduction in (c), the timing of silence expansion or silence reduction is performed during the period of a silence packet, which may cause a significant change in the reproduced speech waveform. It is possible to make high-quality calls without any discomfort.

  FIG. 3 is a buffer operation diagram showing the principle of silent expansion in the jitter buffer 10 (b). State 1 is a state in which an unstored buffer for accepting the No. 5 received packet 31 (a) is set in the buffer A. The No. 1 received packet 31 has a large delay in the IP network, and the buffer E is not yet used. The storage has passed. Then, the reception packet 31 of No. 1 (sound) is received and the state 2 is obtained.

  The address management unit 11 detects that the delay in the IP network is large. At this time, it is found from the monitoring by the silence flag / buffer monitoring unit 12 that the Nos. 3n and 4n of the silent reception packet 31 are stored in the buffers C and B. Therefore, the silent expansion processing unit 14 operates in the state 3 to insert the silent packet In between No. 3n and 4n of the silent received packet 31 and expand the buffer a to the future side. By this extension, the waiting time for No. 5 is extended by one buffer. At the same time, No. 1 audio data is output as reproduction data 49 (c).

  FIG. 4 is a buffer operation diagram showing the principle of silence reduction in the jitter buffer 10 (b). State 1 is a state of waiting for No. 5 received packet 31 (a). Therefore, when No. 5 is received, the state 2 is entered, and the received packet 31 of No. 5 is stored in the packet A. Packets B to E are also filled with received packets, and it is clear from the management of the address management unit 11 that the received packets 31 have arrived smoothly. This means that the time until sound reproduction can be shortened by reducing the number of buffers.

  As a result of monitoring by the silence flag / buffer monitoring unit 12, it is found that the No. 3n and 4n of the silent reception packet 31 are stored in the buffers C and B. Therefore, the silence reduction processing unit 13 operates in the state 3 to discard No. 3n of the silent reception packet 31 and reduce the buffer A. Due to this reduction, the time required to reproduce No. 5 is shortened by one buffer. At the same time, No. 1 audio data is output as reproduction data 49 (c).

  In this way, when the packets arrive smoothly within a predetermined period and there are no unarrived packets, all the buffers are filled with valid data (state 2 in FIG. 4). When the continuation period (only state 2 in FIG. 4) is counted by the counter 22 and reaches a predetermined value, the silence reduction processing of the buffer is executed.

  FIG. 5 is a time chart showing an example of the voice / silence discrimination operation of the voice / silence discrimination unit 6. (A) shows a case where the received packet 31 is normally received without jitter in the IP network. (B) shows the time of occurrence of jitter in which the received packet 31 is received in a state where jitter has occurred in the IP network. (C) shows the sound / silence determination result of the sound / silence discrimination unit 6 when jitter occurs. The reception interval of the normal reception packet 31 is, for example, 40 ms, and the time until reception from the transmission terminal via the IP network is, for example, 80 ms.

  When jitter occurs (b), Nos. 3 and 4 of the received packet 31 are delayed, and Nos. 5n and 8n are silent. No. 4 is received almost simultaneously with No. 7. The determination result of the voice / silence of each received packet 31 is as shown in (c), and the silence flag adding unit 7 adds a silence flag to the packet determined to be silent.

  Since it is only necessary to be able to discriminate between sound and silence, only the sound flag may be attached, or the presence or absence of the sound flag and silence flag may be used to identify the sound and silence. It will be clear from the purpose. In addition, in the transmission terminal, it is also possible to transmit the transmission packet with a voice / silence identification flag, and in this case, the packet voice / silence identification section 6 and the silence flag addition section 7 It is only necessary to refer to or check the flag.

  FIG. 6 is a time chart showing an example of the operation of judging whether or not reproduction is possible when jitter occurs in the address management unit 11. (A) shows a case where the received packet 31 is normally received without jitter in the IP network. (B) shows the time of occurrence of jitter in which the received packet 31 is received in a state where jitter has occurred in the IP network. (C) shows the result of judgment of whether or not reproduction is possible when jitter occurs. The reception interval of the normal reception packet 31 is, for example, 40 ms, and the time until reception from the transmission terminal via the IP network is, for example, 80 ms.

  When jitter occurs (b), Nos. 3 and 4 of the received packet 31 are delayed, and Nos. 5n and 8n are silent. No. 4 is received almost simultaneously with No. 8n. The result of determining whether or not each received packet 31 can be reproduced is as shown in (c), and the packet determined to be acceptable is stored in the jitter buffer 10. No. 4 of the received packet 31 has a significant delay and is determined to be no because it exceeds the predetermined reception period. As a result, since it is not stored in the jitter buffer 10, it is not reproduced.

FIG. 7 is a time chart showing an example of expansion operation during silence in the silent expansion processing unit 14. (A) shows a case where the received packet 31 is normally received without jitter in the IP network. (B) shows the time of occurrence of jitter in which the received packet 31 is received in a state where jitter has occurred in the IP network. (C) shows the determination result of the silence flag / buffer monitoring unit 12 when jitter occurs. Sound and silence are monitored to determine when to add a buffer. The reception interval of the normal reception packet 31 is, for example, 40 ms, and the time until reception from the transmission terminal via the IP network is, for example, 80 ms.

  When jitter occurs (b), Nos. 3 and 4 of the received packet 31 are delayed, and Nos. 5n and 8n are silent. No. 4 is received almost simultaneously with No. 8n. Received packets No. 1, 2, 3, and 5 of the received packet 31, but the counter 22 counted exceeding the period expected to receive No. 4, so a buffer should be added immediately after the silent No. 5n Is determined by the silence flag / buffer monitoring unit 12. As a result, the silence expansion processing work is executed in the silence expansion processing unit 14 of (d). By adding this buffer, the waiting time for the No. 4 received packet 31 is extended (40 ms). Silence expansion processing is possible up to the limit of the jitter buffer 10 if further required.

  FIG. 8 is a time chart showing an example of a reduction operation during silence in the silence reduction processor 13. (A) shows a case where the received packet 31 is normally received without jitter in the IP network. (B) shows the time of occurrence of jitter in which the received packet 31 is received in a state where jitter has occurred in the IP network. (C) shows the determination result of the silence flag / buffer monitoring unit 12 when jitter occurs. Sound and silence are monitored to determine when to delete the buffer. The reception interval of the reception packet 31 at normal time is, for example, 40 ms, and the time until reception from the transmission terminal via the IP network is, for example, 80 ms.

  When jitter occurs (b), Nos. 3 and 4 of the received packet 31 are delayed, but the amount of delay is very small, and Nos. 5n and 8n are silent. Since the counter 22 has counted that a predetermined period of time has passed to normally receive No. 1 to No. 5 of the received packet 31, the No. 6 buffer in which silence is followed by No. 5n and 6n is deleted. The silence flag / buffer monitoring unit 12 determines what should be done. As a result, the silence reduction processing unit 13 of FIG. This buffer deletion shortens the time from reception to audio reproduction (40 ms). The silence reduction process is executed in the subsequent silence interval if further required.

  FIG. 9 is a buffer operation diagram showing the operation of the jitter buffer 10. (A) has shown the case where the received packet 31 is received normally without the jitter in an IP network. (B) shows the time of occurrence of jitter in which the received packet 31 is received in a state where jitter has occurred in the IP network. (C) shows the stored contents of the jitter buffer 10. (D) shows the reproduction data 49. The reception interval of the reception packet 31 at normal time is, for example, 40 ms, and the time until reception from the transmission terminal via the IP network is, for example, 80 ms.

  When jitter occurs (b), Nos. 3, 4, 6, and 7 of the received packet 31 are delayed, and No. 5n is silent. In state 1, No. 2 received packet 31 is received and stored in buffer A. In the state 2, the buffer A is an unstored buffer for receiving the No. 3 received packet 31, but No. 3 could not be received.

  In state 3, buffer A is an unstored buffer for receiving the received packet 31 of No. 4, but No. 4 was not received, but No. 3 was received. Stored. Therefore, No. 0 audio data is output as the reproduction data 49. In state 4, buffer A is an unstored buffer for receiving No. 5n received packet 31. At the same time, No. 5n is received and stored, and No. 1 audio data is output as reproduction data 49. doing.

  Since No. 4 cannot be received even after state 4 has elapsed, In (silent insertion packet) is inserted into buffer A in state 5, buffer a is expanded to the future side, and No. 6 is not stored. A buffer. Since No. 4 is received at the same time, it is stored in the buffer C and No. 2 audio data is output as reproduction data 49. In state 6, the buffer a is an unstored buffer for receiving the received packet 31 of No. 7, but at the same time, No. 6 is received and audio data of No. 3 is output as reproduction data 49. .

  In state 7, buffer a is an unstored buffer for receiving No. 8 received packet 31, but at the same time, No. 8 is received and stored, and No. 4 audio data is output as reproduction data 49. doing. In state 8, the buffer a is an unstored buffer for receiving the received packet 31 of No. 9, but at the same time, No. 9 is received, and further, No. 7 is received by the buffer B and reproduced. No. 5n silent audio data is output as data 49.

  In state 9, buffer a is an unstored buffer for receiving No. 10 received packet 31, but at the same time, No. 10 is received and audio data of silent packet In inserted as reproduction data 49 is output. doing. In this way, in the reproduction data 49, the silent insertion packet In inserted after No. 0 to 4, 5n and No. 5n which is a silent packet is reproduced.

  10 to 12 are flowcharts showing the operation flow in the circuit configuration shown in FIG. When the reception operation is started, it is confirmed whether or not the packet receiving unit 31 has received a packet (S1, FIG. 10). If received (S1Y), the packet sound / silence identification unit 6 identifies whether the received packet 31 is sound with a sound level higher than a predetermined sound level or sound with a sound level lower than a predetermined sound level (S2).

  A flag for identifying the presence / absence of sound is added to the received packet 31, and as a received packet 32, a silent flag is added to the received packet 32 by the silent flag adding unit 7, and the flagged data 35 is stored in the jitter buffer 10 ( S3). If this storage has occurred, and if the packet receiver 31 has not received a packet (S1N), it waits for the clock 40, which is the audio data playback timing (S4). When the audio data reproduction timing comes (S4Y), the data stored in the buffer on the most past side of the jitter buffer 10 is sent to the data reproduction unit 23 as the buffer output 36 and reproduced as reproduction data 49 (S5, FIG. 11). ).

  At the same time, the sequence number of the unstored data (the sequence number of the received packet 31 to be stored) is stored in the most recent (future) buffer and the unstored buffer is set (S6). Therefore, the address management unit 11 checks whether the buffer expansion of the jitter buffer 10 is necessary (S7) or the buffer reduction is necessary (S8). If it is determined that buffer expansion is necessary (S7Y), a buffer expansion processing subroutine is entered (S10). If it is determined that buffer reduction is necessary (S8N), a buffer reduction processing subroutine is entered (S9).

  When the buffer reduction process (FIG. 12) is entered, the silence flag / buffer monitoring unit 12 checks whether or not there is silence data in many buffers included in the jitter buffer 10 (S21), and the silence data is stored in the buffer. Is present (S21Y), the silence reduction processing unit 13 operates to release and reduce (delete) the buffer containing the silence data (S22), and the buffer reduction process ends.

  When buffer expansion processing (FIG. 13) is entered, the silence flag / buffer monitoring unit 12 checks whether or not there is silence data in a number of buffers included in the jitter buffer 10 (S21), and the silence data is stored in the buffer. Is present (S31Y), the silence decompression processing unit 14 operates to decompress (add) the buffer after the buffer containing the silence data (S32), and the buffer decompression process is terminated. In this way, it is possible to make a call with good voice quality.

It is a circuit block diagram which shows one Example of this invention. It is a wave form diagram which shows the principle of this invention. FIG. 2 is a buffer operation diagram showing the principle of silence expansion in the circuit configuration shown in FIG. FIG. 2 is a buffer operation diagram showing the principle of silence reduction in the circuit configuration shown in FIG. 3 is a time chart showing voice / silence discrimination operation in the circuit configuration shown in FIG. 3 is a time chart showing whether or not reproduction is possible when jitter occurs in the circuit configuration shown in FIG. FIG. 2 is a time chart showing an extension example when there is no sound in the circuit configuration shown in FIG. 1. FIG. 3 is a time chart showing an example of reduction in silence in the circuit configuration shown in FIG. FIG. 2 is a buffer operation diagram showing the operation of the jitter buffer 10 in the circuit configuration shown in FIG. 2 is a flowchart showing an operation flow in the circuit configuration shown in FIG. FIG. 11 is a flowchart showing an operation flow in the circuit configuration shown in FIG. 1 together with FIG. 12 is a flowchart illustrating a subroutine for buffer reduction processing included in FIG. 11. 12 is a flowchart illustrating a subroutine of buffer expansion processing included in FIG. 11. It is a circuit block diagram which shows a prior art example. It is a wave form diagram which shows the principle of the prior art example shown in FIG. FIG. 15 is a buffer operation diagram illustrating an example of buffer expansion in the circuit configuration illustrated in FIG. 14. FIG. 15 is a buffer operation diagram illustrating an example of buffer reduction in the circuit configuration illustrated in FIG. 14.

Explanation of symbols

2,3 I / F (interface)
5 Packet reception unit 6, 6B Packet sound / silence identification unit 7 Silence flag addition unit 10, 10B Jitter buffer 11 Address management unit 12 Silence flag / buffer monitoring unit 12B Buffer monitoring unit 13 Silence reduction processing unit 13B Reduction processing unit 14 Silence expansion processing unit 14B Extension processing unit 15, 15B Pseudo silence data unit 20, 20B Main processing unit 23, 23B Data reproduction unit 31, 32 Received packets 33, 33B Sound / silence identification signal 34 Silence flag addition notification 35 Reception with flag Packet 35B Received packet 36, 36B Buffer output 40 Clock 41-45, 42B-44B Signal 46, 46B Pseudo silence data 49 Playback data 51, 52 Signal

Claims (8)

The received packet (35) after identification is identified by identifying whether the voice level of the voice data of the received packet (31) received from the IP network is voiced greater than a predetermined value or silence less than the predetermined value. Perform packet voice / silence discrimination processing (6, 7) to obtain
The received packet (35) after the identification is stored (10) in a first-in first-out type jitter buffer including a plurality of buffers.
When the jitter buffer needs to be expanded or contracted in order to absorb the jitter of the received packet (31), the silence in which the silence data of the received packet (35) identified as the silence is present. A method of absorbing a jitter of a voice IP terminal, which performs silent expansion / contraction processing (11 to 15) for expanding or reducing the jitter buffer during a period. The packet voice / silence discrimination processing (6, 7)
Silence flag for obtaining a received packet (35) after identification by attaching a silence flag for identification to the received packet (31) when the voice data is identified as silence that is smaller than the predetermined value The method for absorbing jitter of a voice IP terminal according to claim 1, further comprising an additional process (7). The silent expansion / contraction processing (11-15)
An address management process (11) for managing the storage status of the plurality of buffers included in the jitter buffer to determine whether the jitter buffer needs to be expanded or contracted;
A silence buffer monitoring process (12) for monitoring a silence period in which the silence data exists in the plurality of buffers included in the jitter buffer;
Silence expansion / contraction processing (13, 14) for performing the expansion or contraction operation in the silent period when the expansion or contraction operation is required in the jitter buffer including the plurality of buffers, and the The method for absorbing jitter in a voice IP terminal according to claim 1, further comprising: pseudo silence data processing (15) for generating pseudo silence data to be stored in a buffer to be inserted when performing an expansion operation. In the packet voice / silence identification processing (6, 7),
The received packet (31) received from the IP network is inspected to determine whether or not the voice data has already been marked to identify whether the voice data is voiced or silent. The method for absorbing jitter in a voice IP terminal according to claim 1, wherein the received packet (31) is the identified received packet (35). The received packet (35) after identification is identified by identifying whether the voice level of the voice data of the received packet (31) received from the IP network is voiced greater than a predetermined value or silence less than the predetermined value. Packet voice / silence identification means (6, 7) for obtaining;
Buffer storage means (10) for storing the identified received packet (35) in a first-in first-out jitter buffer including a plurality of buffers;
When the jitter buffer needs to be expanded or contracted in order to absorb the jitter of the received packet (31), the silence in which the silence data of the received packet (35) identified as the silence is present. And a silent expansion / contraction means (11-15) for expanding or contracting the jitter buffer over a period of time. The packet voice / silence identification means (6, 7)
Silence flag for obtaining a received packet (35) after identification by attaching a silence flag for identification to the received packet (31) when the voice data is identified as silence that is smaller than the predetermined value The jitter absorbing apparatus for a voice IP terminal according to claim 5, further comprising additional means (7). The silent expansion / contraction means (11-15)
Address management means (11) for managing the storage status of the plurality of buffers included in the jitter buffer and determining whether or not expansion or contraction is required in the jitter buffer;
Silence buffer monitoring means (12) for monitoring a silence period in which the silence data exists in the plurality of buffers included in the jitter buffer;
Silence expansion / contraction means (13, 14) for performing the expansion or contraction operation in the silent period when the expansion or contraction operation is required in the jitter buffer including the plurality of buffers; 6. A jitter absorbing apparatus for a voice IP terminal according to claim 5, further comprising pseudo silence data means (15) for generating pseudo silence data to be stored in a buffer to be inserted when the work for expansion is performed. In the packet voice / silence identification means (6, 7),
The received packet (31) received from the IP network is inspected to determine whether or not the voice data has already been marked to identify whether the voice data is voiced or silent. The jitter absorbing apparatus for a voice IP terminal according to claim 5, wherein the received packet (31) is the identified received packet (35).

Images