JP4454255B2 - Voice / fax communication system, voice / fax receiver, and fluctuation absorbing buffer amount control method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a voice / FAX communication system that performs communication by packetizing a voice signal or a facsimile (hereinafter abbreviated as FAX) signal and a communication control method thereof. The present invention is applicable to, for example, a multimedia packet switching system, a VoIP (Voice over IP) system for providing an IP (Internet Protocol) based voice communication service, and the like.
[0002]
[Prior art]
When voice data is transmitted in real time through a packet communication network, it is well known that the quality of voice communication is greatly influenced by the magnitude of packet delay. In order to avoid such deterioration in communication quality due to packet delay, various techniques for absorbing packet delay fluctuation have been proposed. For example, in Patent Document 1 (Japanese Patent Laid-Open No. 2000-224225), a jitter buffer for absorbing delay fluctuations of a received packet is provided, and a packet delay is calculated from time stamp information of the received packet, and the amount of delay is calculated. Accordingly, a technique for controlling the packet decoding timing has been proposed.
[0003]
However, there is a problem in that a sense of incongruity occurs when the buffer amount is varied during sound because the voice communication includes a voiced section and a silent section. For example, when the buffer amount is increased when there is sound, the reproduced sound is heard with a delay, and conversely when the buffer amount is decreased, the sound is clogged. Therefore, Patent Document 2 (Japanese Patent Application Laid-Open No. 2002-271391) discloses dynamic jitter that detects a silence pattern before decoding received audio data, and inserts or deletes the silence pattern during silence to vary the absorption buffer amount. Buffer control methods have been proposed.
[0004]
[Patent Document 1]
Japanese Unexamined Patent Publication No. 2000-224225 (abstract, description, paragraph number 0085, FIG. 1, FIG. 4)
[Patent Document 2]
JP 2002-271391 (paragraph numbers 0041-0042).
[0005]
[Problems to be solved by the invention]
However, in the technique described in Patent Document 2, since the silent pattern is unique to the system, the internationally standardized voice encoding technique cannot be used. In the case of a voice communication control system based on an internationally standardized voice coding system (for example, G.728, G.729, G.723.1), the encoder (transmission side) has a silence detection function. In general, silence is not a state where there is no sound at all, but generally there is background noise. Therefore, even if silence is detected, individual encoding results are not constant depending on the background noise at that time. If this is replaced with a certain silence pattern, a sense of incongruity will occur when the decoder (reception side) reproduces the sound. In particular, since a sense of incongruity occurs at the time of switching between voiced and silenced, the feeling of incongruity is as if a silence compression technique that is not so popular is used.
[0006]
Furthermore, in the technique described in Patent Document 2, it is necessary to determine whether there is silence or sound before the fluctuation absorbing buffer. That is, the silence detection performed in the decoder is also required in the packet analysis unit, and the same processing is performed twice.
[0007]
Further, in voice communication using a codec that does not have a silence compression function, FAX communication is permitted at a communication speed within the range of the voice compression rate. In general, after establishing a communication path by voice communication, the communication is switched to FAX communication. However, in some cases, voice communication and FAX communication may be alternately repeated.
[0008]
As is well known, since voice communication emphasizes immediacy (real-time performance) even if the quality is somewhat degraded, packets may be discarded when a delay occurs. On the other hand, in FAX communication, since certainty is emphasized even if a slight delay occurs, control is performed so that delayed packets are not discarded as much as possible. Even when voice communication and FAX communication having such reverse characteristics coexist, FAX communication has been conventionally controlled by the control method of voice communication. For this reason, data necessary for FAX communication has been lost and retransmission processing has occurred, resulting in unnecessary communication time, or a communication error has occurred and the FAX has to be sent again.
[0009]
SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide voice / fax communication that enables reproduction as a continuous smooth signal even when delay or fluctuation in the network occurs when voice / fax signals are packetized and communicated. A system, a voice / fax receiving apparatus, and a reception control method thereof are provided.
[0010]
Furthermore, another object of the present invention is to provide a voice / FAX communication system and its communication control that can reproduce a signal so that a procedure error and a FAX signal are not retransmitted even when voice communication and FAX communication are mixed. It is to provide a method.
[0011]
[Means for Solving the Problems]
According to the present invention, when voice / fax is packetized and communicated, the amount of fluctuation absorbing buffer is dynamically adjusted according to delay or delay fluctuation in the network. Further, the reference for controlling the absorption buffer amount is changed according to whether the voice communication (immediate importance) or the FAX communication (certainty).
[0012]
  According to the first aspect of the present invention, an audio / FAX communication system that transmits an audio / FAX signal via a packet communication network encodes an audio signal or FAX signal and whether encoded data is silent. A transmission device comprising: an encoding means for identifying the packet; and a packet generation means for generating a packet to be transmitted to the packet communication network by setting a silence flag in the encoded data when the encoded data is silent. , A fluctuation absorbing buffer that absorbs delay variation of packets received from the packet communication network, and after switching from voice communication to FAX communication, the encoded data in which the silence flag is setSilenceIn the section, the accumulated amount of the fluctuation absorbing buffer is changed from the standard for voice communication to the standard for FAX communication.And increasing the fluctuation absorption buffer accumulation amount by a predetermined amount using the silent period for each decoding cycle of the encoded data.And a reception device having a buffer amount control means.
[0013]
  The packet generation means of the transmission device sets a silence flag in a packet header and transmits the packet header, and the buffer amount control means of the reception device identifies a silence section of the encoded data by the silence flag of the reception packet header.It is desirable.
[0014]
  According to the second aspect of the present invention, there is provided a voice / fax receiving device that receives a packet including encoded data in which voice / silence is distinguished via a packet communication network. Fluctuation absorption buffer for absorbing delay fluctuations of the received packet, and after switching from voice communication to FAX communication, the amount of fluctuation stored in the fluctuation absorption buffer is determined based on the voice communication reference when the encoded data of the received packet is silent. Change to standardAnd increasing the fluctuation absorption buffer accumulation amount by a predetermined amount using the silent period for each decoding cycle of the encoded data.And a buffer amount control means.
[0015]
  The buffer amount control means includesSuitable for voice communicationFor voice communicationSuitable for the first reference table and FAX communicationFor FAX communicationIncluding the second reference table,It is desirable to switch between the first reference table and the second reference table depending on whether the received encoded data is an audio signal or a FAX signal.
[0016]
    According to the third aspect of the present invention, the buffer amount of the fluctuation absorbing buffer provided in the voice / fax receiving apparatus that receives a packet including encoded data in which voiced / silent is distinguished via a packet communication network The control method determines the silence / sound of the encoded data of the received packet, and after switching from voice communication to FAX communication, the amount of accumulation in the fluctuation absorbing buffer is calculated in the interval where the encoded data of the received packet is silent. Change from communication standard to FAX communication standardAnd increasing the fluctuation absorption buffer accumulation amount by a predetermined amount using the silent period for each decoding cycle of the encoded data.It is characterized by that.
[0017]
In particular, when the delay exceeds a predetermined range, it is desirable to forcibly change the amount of fluctuation absorbing buffer regardless of the presence / absence of sound.
[0018]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a block diagram showing a voice / fax communication system according to an embodiment of the present invention. In the voice / FAX communication system shown in FIG. 1, a plurality of voice packet communication devices (or VoIP gateway devices) # 1, # 2,... Are connected via a packet communication network. A telephone / fax that transmits and receives analog audio signals is connected. Here, the voice packet communication device 1 and the voice packet communication device 2 are connected through the relay line 3 of the packet communication network, and the telephone / FAX 41 is connected to the voice packet communication device 1 through the analog signal line 411, and the voice packet is transmitted. It is assumed that a telephone / FAX 42 is connected to the communication device 2 through an analog signal line 421.
[0019]
Hereinafter, the configuration of the voice packet communication device 1 will be described, but the internal configuration of the voice packet communication device 2 is basically the same. The analog signal line 411 to which the telephone / FAX 41 is connected is connected to the PCM codec 101, and the PCM codec 101 is connected to the transmission system and the reception system. The transmission system includes an encoder 111, a packet generator 121, and a transmission buffer 131, and the reception system includes a reception buffer 132, a packet analyzer 122, a fluctuation absorbing buffer 123, and a decoder 112. The WAN (relay unit) 141 is connected to the transmission buffer 131 and the reception buffer 132 through the data bus 133, transmits the transmission packet received from the transmission buffer 131 to the relay line 3, and receives the packet received from the relay line 3 as the reception buffer 132. To store.
[0020]
Similarly, in the voice packet communication apparatus 2, the analog signal line 421 to which the telephone / FAX 42 is connected is connected to the PCM codec 201, and the PCM codec 201 is connected to the transmission system and the reception system. The transmission system includes an encoder 211, a packet generator 221, and a transmission buffer 231, and the reception system includes a reception buffer 232, a packet analyzer 222, a fluctuation absorbing buffer 223, and a decoder 212. The WAN 241 is connected to the transmission buffer 231 and the reception buffer 232 through the data bus 233.
[0021]
The PCM codecs 101 and 201 in the present embodiment convert analog signals into 64K PCM codes (transmission signals) 1011 and 2011, respectively. Also, the encoders 111 and 211 compress the 64K PCM code, or extract the FAX modem signal and adjust it to the same band as the voice to generate encoded data 1112 and 2112, respectively. Furthermore, the encoders 111 and 211 in the present embodiment can determine whether the encoded data is a sound period or a silent period, and generate a silence instruction signal 1111 that is effective when the encoded data is silent.
[0022]
Decoders 112 and 212 decode the received encoded data into original 64K PCM codes (reproduced signals) 1012 and 2012, respectively. In addition, the decoders 112 and 212 in the present embodiment generate FAX communication display signals 1121 and 2121 that enable the decoding of the FAX modem signal from the start to the end, respectively.
[0023]
Assuming that the voice / fax signal transmission side is the voice packet communication device # 1, the analog signal 411 received from the telephone / fax terminal 41 is converted into a 64K PCM code 1011 by the PCM codec 101. The 64K PCM code 1011 is encoded by the encoder 111 every encoding period (for example, 10 ms), and the encoded data 1112 and a silence display flag 1111 indicating whether the encoded data is silent are sent to the packet generator 121. hand over. The packet generator 121 assembles a packet by adding encoded data 1 to N to a packet header to be described later, and further sets a silence display flag 1111 at a frame position corresponding to each of the encoded data 1 to N. The packet 1211 generated here is written into the transmission buffer 131 and makes a transmission request to the WAN 141. The WAN 141 takes a packet from the transmission buffer 131 via the data bus 133, generates a packet to be transmitted, and transmits it to the relay line 3. Details of the voice / fax signal transmission control will be described later.
[0024]
Assuming that the voice / FAX signal receiving side is the voice packet communication device # 2, the WAN 241 receives the packet from the trunk line 3, stores it in the reception buffer 232 through the data bus 233, and notifies the packet analyzer 222 of packet reception. .
[0025]
The packet analyzer 222 functions as a means for controlling the fluctuation absorbing buffer amount. The packet analyzer 222 reads the packet header 2221 from the reception buffer 232 and checks whether it is a correct packet. If it is not correct, the packet is discarded, and if it is correct, it is determined whether or not the current fluctuation absorbing buffer amount is appropriate. If the reception range is reproducible, the address of the reception buffer 232 storing the packet and the silence display flag of the packet are stored in the fluctuation absorbing buffer 223.
[0026]
The main process of decoding itself is conventionally known. The decoder 212 performs decoding every decoding cycle (for example, 10 ms). If there is a fluctuation in the fluctuation absorbing buffer amount, the buffer amount is changed as will be described later, and then data is extracted from the fluctuation absorbing buffer 223. To decrypt. The data 2012 decoded in this manner is returned to the analog signal 421 by the PCM codec 201 and reaches the receiving side telephone / fax machine 42. The reproduced frame is deleted from the reception buffer 232 and the fluctuation absorbing buffer 223.
[0027]
Also, the decoders 112 and 212 validate the FAX communication display signal 2121 when decoding the FAX signal. Depending on the state of the FAX communication display signal 2121, as described later, the packet analyzers 122 and 222 can switch the reference table of the fluctuation absorbing buffer control parameter to cope with a mixture of voice communication and FAX communication.
[0028]
Hereinafter, specific operation examples of the voice / fax communication system according to the present embodiment will be described in detail with reference to FIGS.
[0029]
Voice / FAX packet
FIG. 2 is a format diagram of a voice / FAX packet stored in the transmission / reception buffer of the voice packet communication apparatus according to the present embodiment. Here, it is assumed that a plurality of encoded data (frames) 1 to N are packetized.
[0030]
In FIG. 2, for example, a voice / FAX packet stored in one transmission buffer has a pointer 13111 indicating the start address of the next buffer, a command (packet transmission) number 13112, a packet length 13113, and a voice encoding method. Corresponding to each of the voice type 13114, packetized code data (hereinafter referred to as a frame), a silence display flag 13115 indicating whether the frame is voiced / silent, and a transmission sequence number 13116 indicating the packet transmission order. , A time stamp (hereinafter referred to as TS) 13117 indicating the encoding time of the packet, a transmission source identifier 13118, and encoded data (voice / fax signal) 13119. When the encoders 111 and 211 detect a silent section, a value indicating “silent” is recorded in the silent display flag 13115.
[0031]
The transmission buffers 131 and 231 operate in a pointer chain system indicating the head address of the next buffer, and packets sent to the trunk line 3 by the WAN are voice type 13114 to encoded data 13119 packets. The same applies to the reception buffers 132 and 232. When a packet of the audio type 23114 to the encoded data 23119 is received, a pointer 23111, a command (in this case, packet reception) number 23112, and a packet length 23113 are added and received. Stored in buffers 132 and 232, respectively.
[0032]
The command 13112/23112 and the packet length 13113/23113 are used as interface parameters between the voice / FAX signal processing unit and the relay processing unit. The parts that are actually transmitted in the packet format are the audio type 13114 to the transmission source identifier 13118 and the encoded data 13119 in the packet header. Parameters other than the silence display flag 13115 according to the present invention are given according to RTP (Real-time Transport Protocol) standardized by VoIP as usual. Since a plurality of frames 1 to N are packetized in one packet, the silence display flags 1 to n are required for the number of frames.
[0033]
Fluctuation absorption buffer
FIG. 3 is a schematic diagram showing the configuration and operation of the fluctuation absorbing buffer in the present embodiment. The fluctuation absorbing buffers 123 and 223 are a buffer 3001 for storing the address of the reception buffer in which the voice / FAX frame is stored, and a buffer for storing a silence display flag for displaying whether each of the voice / FAX frame is silent. 3002.
[0034]
The packet analyzers 122 and 222 include a playback pointer 3003 indicating a frame position to be played next, a playback frame time stamp (TS) 3004 indicating a sampling time of the playback pointer position, and a maximum delay of a packet that arrives later than the playback pointer ( An increase pointer 3006 indicating the position farthest from the playback pointer), a variable pointer 3007 indicating the maximum delay of the packet that arrived without delay after the playback pointer (position closest to the playback pointer), and a parameter for controlling the fluctuation absorbing buffer Using the table 3005, the fluctuation absorbing buffers 123 and 223 are respectively controlled as will be described later. Note that the initial values of the increase pointer 3006 and the variation pointer 3007 are both the same as those of the playback pointer 3003, and both are initialized when a fluctuation absorbing buffer amount described later is changed.
[0035]
The absorption buffer control parameter table 3005 includes an increase range parameter 30051 indicating a range in which the fluctuation absorption buffer should be increased, a dead range parameter 30052 indicating a range in which the fluctuation absorption buffer is not changed, and a decrease indicating a range in which the fluctuation absorption buffer is decreased. A range parameter 30053 is provided. Each parameter has two different values defined in advance depending on whether voice communication or FAX communication is being performed.
[0036]
In the example shown in FIG. 3, a range of 0 ms to 120 ms that is classified into an increase range, a dead range, or a decrease range in the case of voice communication is equally determined as an increase range during FAX communication. Thus, even when switching from voice communication to FAX communication occurs, it is always determined that the range is increased in FAX communication, and the buffer amount can be increased using a silent section as will be described later. Conversely, when switching from FAX communication to voice communication, for example, the insensitive range of FAX communication becomes the forcible decrease range of voice communication, and the buffer amount is forcibly used by using a silent section or at the timing of switching. Can be reduced. In this way, appropriate control can be performed by switching between the two parameter tables for immediacy and certainty.
[0037]
Next, the operation in the forced increase / increase / insensitive / decrease / forced decrease range will be briefly described.
[0038]
(a) Forced increase range: When a certain number of packets arrive in this range, the buffer amount is forcibly increased regardless of the voiced / silent period. Packets received during this time are discarded because the reproduction timing has passed.
[0039]
(b) Increase range: When a packet arrives within this range, the buffer amount for one frame is increased using a silent period. The received packet is loaded into the fluctuation absorbing buffer.
[0040]
(c) Insensitive range: When a packet arrives within this range, the buffer amount is not changed. The received packet is loaded into the fluctuation absorbing buffer.
[0041]
(d) Decreasing range: When a packet arrives within this range, the buffer amount for one frame is decreased using a silent period. The received packet is loaded into the fluctuation absorbing buffer.
[0042]
(e) Forced decrease range: When a packet arrives within this range, the buffer amount is forcibly decreased regardless of the voiced / silent period. The received packet is loaded into the fluctuation absorbing buffer.
[0043]
In order to absorb fluctuations, it is necessary to adjust the buffer amount toward a larger delay. It is also necessary to prevent an unstable operation such as increase in buffer amount → decrease → increase. Therefore, priority is given to processing in the increasing direction in the fluctuation absorption buffer amount fluctuation control.
[0044]
Voice / FAX signal transmission control
First, voice / fax signal transmission control in this embodiment will be described by taking voice packet communication apparatus # 1 as an example. Here, one packet is composed of N frames, and the position and number of frames are indicated by the frame count (see FIG. 2).
[0045]
FIG. 4 is a flowchart showing voice / FAX signal transmission control in the present embodiment. First, after performing predetermined initialization, the encoder 111 is activated at an encoding cycle (for example, 10 ms) (step S101), and the 64K PCM code 1011 output from the PCM codec 101 is encoded (step S102). The packet generator 121 receives the encoded data 1112 and the silence display flag 1111 of the encoded data from the encoder 111, and determines whether or not the number of encoded data to be incorporated into the packet, that is, the frame count is zero. (Step S103).
[0046]
If the frame count is 0 (YES in step S103), a packet header is generated to generate a new packet (step S104), and the frame count is increased by 1 (step S105). If the frame count is not 0 (NO in step S103), the frame count is increased by 1 without generating a packet header (step S105).
[0047]
Subsequently, the packet generator 121 sets the encoded data 1112 to a frame position according to the frame count (step S106), and similarly sets the silence display flag 1111 to a position according to the frame count (step S107). . Then, it is determined whether the frame count is equal to N (step S108). If the frame count has not reached N (NO in step S108), the process returns to step S101. That is, steps S101 to S107 are repeated until the frame count reaches N.
[0048]
When the frame count is equal to N (YES in step S108), a packet carrying N encoded data is completed, so this transmission packet 1211 is written to the transmission buffer 131 (step S109), and the frame count is calculated. Reset to 0 (step S110). Then, a transmission request is output to the WAN 141 and is transmitted to the trunk line 3 by the WAN 141 as described above (step S111). By repeating the above steps S101 to S111, transmission packets are sequentially transmitted.
[0049]
Voice / FAX signal reception control
Next, voice / fax signal reception control will be described in detail with reference to FIGS. 3 and 5 to 7, taking voice packet communication apparatus # 2 as an example.
[0050]
FIG. 5 is a flowchart showing packet analysis of voice / fax signal reception control in this embodiment, and FIG. 6 is a flowchart showing fluctuation absorption buffer amount fluctuation control in this embodiment. The packet analyzer determines the parameter table to be referred to by the FAX communication display signal 2121 as described above, and performs packet analysis, loading into the fluctuation absorbing buffer, and fluctuation control of the fluctuation absorbing buffer amount as described below. To do.
[0051]
(1) Packet analysis and loading into fluctuation absorbing buffer
In FIG. 5, when the WAN 241 of the packet communication apparatus # 2 receives the packet from the relay line 3 (step S201), the received packet is stored in the reception buffer 232 through the data bus 233 (see FIG. 2) and sent to the packet analyzer 222. Notify packet reception. Upon receiving the packet reception notification, the packet analyzer 222 reads and analyzes the packet header 2221 from the reception buffer 232 and determines whether the packet is a correct packet from the packet length 23113, the voice type 23114, the transmission source identifier 23118, and the like. (Step S202). As described above, if the packet is not correct, the packet is discarded. If the packet is correct, it is determined whether or not the current fluctuation absorbing buffer amount is appropriate as described below.
[0052]
First, using the playback pointer 3003, the playback frame TS3004, and the received packet time stamp (TS) 23117, the fluctuation absorbing buffer loading position of the received packet is calculated (step S203).
[0053]
For example, when the initial value of the received packet TS is 0, the TS is increased by 80 per frame in 64 KPCM sampling units, and one packet is composed of 3 frames, the TS of the packet received at the beginning of communication is “0”. , When the packet sequence number increases by “1”, the TS increases by “80 × 3 = 240”. In this case, if the reproduction frame TS is X and the reception packet TS is Y, the loading position can be calculated by (Y−X) / 80.
[0054]
In the buffer configuration shown in FIG. 3, when X = 2080 and Y = 2160, the loading position is “1”. 3 is loaded with a reception buffer address and a silence display flag. If it is “during voice communication” in FIG. 3, it is received in “increase range”. If X = 2080 and Y = 1840, the loading position is “−3”, so that the reception buffer address is loaded into # 125 to # 127 of the reception buffer address 3001. In this case, if it is “during voice communication”, it means that it has been received in a “forced increase range”, that is, outside the increase range (reproduction impossible range).
[0055]
As described above, the absorption buffer control parameter table referred to for determining the forced increase / increase / insensitivity / decrease / forced decrease range is that the FAX communication display signal 2121 is “voice communication in progress” or “FAX communication”. It is determined depending on which of “in communication” is indicated.
[0056]
In this way, the packet analyzers 122 and 222 determine whether the calculated loading position is in the forced increase range or any other range (step S204).
[0057]
Packets received in the forced increase range will not be played because the playback time has already passed, but if this state continues for a long time, the sound will not be played and the sound will be silenced, so the fluctuation absorbing buffer will be changed to the adaptive position at an early opportunity There is a need to. Therefore, when the loading position falls within the forced increase range (step S204), the packet analyzer determines whether or not the loading position is smaller than the increase pointer 3006 (step S205). If the loading position is smaller than the increase pointer 3006, that is, the delay is larger (YES in step S205), the increase pointer 3006 is updated to the currently calculated loading position (step S206), and the out-of-range counter is incremented by 1. (Step S207), it is determined whether or not a packet of a certain value or more has arrived in the forced increase range (step S208).
[0058]
If it is determined that a packet of a certain value or more has arrived in the forced increase range (S208), the forced increase flag is set to ON (step S209). As a result, as will be described later, the fluctuation absorbing buffer can be increased regardless of sound / silence. Since a packet received outside the forced increase range is in time for the playback time, there is no urgency in changing the fluctuation absorbing buffer to the adaptive position. In this case, the packet analyzer determines whether the loading position calculated in step S203 is smaller than the variation pointer 3007 in order to vary the buffer amount at a silent timing as described later (step S210).
[0059]
If the loading position is smaller than the fluctuation pointer 3007, that is, if the delay is larger (YES in step S210), the fluctuation pointer 3007 is updated to the currently calculated loading position (step S211), and the reception in which the frame to be reproduced is stored is received. The buffer address and the silence display flag of the frame are loaded into the fluctuation absorbing buffer (step S212). Note that the initial values of the increment pointer 3006 and the variation pointer 3007 are the same as those of the playback pointer 3003, and are initialized when the fluctuation absorbing buffer amount is changed.
[0060]
(2) Fluctuation absorption buffer amount fluctuation control
As described above, the packet analyzer refers to the parameter table determined depending on the FAX communication display signal 2121 and controls the fluctuation absorbing buffer amount. A more detailed control operation is shown in FIG.
[0061]
In FIG. 6, when the decoding cycle of the decoder starts (step S301), the packet analyzer decrements the cycle counter by 1 (step S302), and whether the forced increase flag set in step S209 described above is on. It is determined whether or not (step S303).
[0062]
If the forced increase flag is on (YES in step S303), the out-of-range monitoring counter exceeds a certain value, so it is necessary to forcibly increase the buffer amount. Therefore, the playback pointer 3003 and the variation pointer 3007 are changed to the values of the increase pointer 3006 (step S304), the out-of-range monitoring counter is reset to 0 (step S305), and the period counter is reset to the initial value L (step S304). S306), a received packet reproduction process described later is executed (see FIG. 7). Note that the packet received during this time is discarded because the reproduction timing has passed.
[0063]
If the forced increase flag is off (step S303), it is determined whether or not the period counter is 0 (step S307). If the period counter is not 0 (NO in step S307), the received packet reproduction described later The process is executed (see FIG. 7). If the cycle counter = 0 (YES in step S307), it is determined whether or not the fluctuation pointer 3007 is in the forced decrease range (step S308). If the fluctuation pointer 3007 is smaller than the forced decrease range (that is, less than the decrease range), the silence display flag corresponding to the address of the fluctuation absorbing buffer determined by the current period counter is checked (step S309).
[0064]
If the corresponding silence display flag is 1 (silence), it is determined whether or not the counter outside the increase range = 0 (step S310). If the counter outside the increase range = 0 (YES in step S310), Since the loading position is not within the forced increase range, the existence range of the variation pointer 3007 is determined (step S311).
[0065]
If the variable pointer 3007 is in the increase range, the buffer amount is increased by one frame by decreasing the playback pointer 3003 by one (step S312), and the process proceeds to the above-described step S306. Therefore, the buffer amount can be increased little by little (here, one frame) only in the silent period. Also when the fluctuation pointer 3007 is in the insensitive range, the process proceeds to step S306 described above. If the variation pointer 3007 is in the decreasing range, the process proceeds to step S314 described later.
[0066]
If the silence display flag is 0 (sound) in step S309, the period counter is incremented by one (step S313), and a received packet reproduction process described later is executed (see FIG. 7).
[0067]
If the variation pointer 3007 is within the forced decrease range in step S308, or if the variation pointer 3007 is in the decrease range in step S311, the packet analyzer extracts information for one frame from the fluctuation absorbing buffer (step S311). In step S314, the corresponding reception buffer and fluctuation absorbing buffer are released (steps S315 and S316). Then, the buffer amount is decreased by one frame by increasing the reproduction pointer 3003 by one. These steps S314 to S317 are repeated until the reproduction pointer 3003 matches the fluctuation pointer 3007 (step S318). Therefore, the buffer amount for the frame corresponding to the difference between the playback pointer 3003 and the variation pointer 3007 can be reduced. When the playback pointer 3003 matches the fluctuation pointer 3007, the increase pointer 3006 is set to the value of the fluctuation pointer 3007, and the process returns to step S306.
[0068]
In this way, the buffer amount is forcibly reduced when the variable pointer 3007 is within the forced decrease range in step S308, and the silent section is used when the variable pointer 3007 is within the decrease range in step S311. it can.
[0069]
During periodic monitoring, it is assumed that both the increasing direction and decreasing direction may be monitored, but in order to absorb fluctuations, it is necessary to adjust the buffer amount to the larger delay, and the buffer amount increases. In order not to generate unstable processing such as → decrease → increase, priority is given to processing in the increasing direction, and when the buffer amount is changed, all monitoring is cleared.
[0070]
In addition, when the fluctuation is stable in the insensitive range, when switching from voice communication to FAX communication, the table to be referenced changes, so the insensitive range of voice becomes the increase range of FAX communication, and the silent section is It is possible to periodically increase the buffer amount by using it. On the contrary, when switching from FAX communication to voice communication, the insensitive range of FAX communication is decreasing in voice communication, so if the buffer amount is small, use the silent section or the moment of switching Forcibly, the buffer amount can be reduced.
[0071]
(3) Voice / FAX packet playback control
FIG. 7 is a flowchart showing the voice / fax packet reproduction control in this embodiment. However, as described above, the inlet terminal C in the flowchart of FIG. 7 is connected to the connection terminal C in the flowchart of FIG.
[0072]
The main processing of decoding in the voice packet communication apparatus is substantially the same as that conventionally known. Decoding is performed by the decoder 212 every decoding cycle (for example, 10 ms). As described above, when the fluctuation absorbing buffer amount varies, the reproduction pointer 3003 is changed before taking out data to be decoded.
[0073]
In FIG. 7, the packet analyzer first extracts information for one frame from the fluctuation absorbing buffer (step S401), and the decoder decodes and reproduces the corresponding encoded data (steps S402 and S403). The reproduced frame is deleted from the reception buffer (reception buffer release: step S404), and is also deleted from the fluctuation absorption buffer (fluctuation absorption buffer release: step S405). Subsequently, the playback pointer is increased by one (step S406), and the playback frame TS is increased by 80 (step S407). The data 2012 decoded in this manner is returned to the analog signal 421 by the PCM codec 201 and sent to the receiving telephone / fax machine: 42.
[0074]
As described above, according to this embodiment, smooth voice reproduction can be obtained even when a delay occurs in the network during a call and the packet reception interval fluctuates. Further, since silence detection on the transmission side can be omitted, the processing on the reproduction side can be made more efficient. In addition, by switching the parameter table according to the FAX communication display signal, a procedure error and retransmission of the image signal are less likely to occur even if a delay occurs in the network during the FAX communication and the packet reception interval fluctuates.
[0075]
Although the present invention has been described above with reference to the embodiments, the specific processing and numerical values in the voice packet communication device according to the embodiments are merely examples and do not define the present invention, and realize the intended function. Of course, different procedures and different parameters may be adopted if possible. An integrated voice packet communication device including a telephone / fax terminal (voice device) may be used.
[0076]
【The invention's effect】
According to the present invention, even if there is no sound, the background noise is utilized and the fluctuation absorbing buffer amount is changed in the silent period. Therefore, even if a delay occurs in the network during a call and the packet reception interval fluctuates, the voice can be smoothly reproduced. It has the effect.
[0077]
Also, by identifying that the FAX communication is being performed, the fluctuation absorbing buffer amount is increased from the buffer amount at the time of voice communication, and the change of the buffer amount is periodically increased little by little using a silent period. Therefore, even if a delay occurs in the network during FAX communication and the packet reception interval fluctuates, it becomes possible to continuously reproduce without destroying the signal necessary for FAX communication, and it is difficult for procedure errors and retransmission of image signals to occur. .
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a voice / fax communication system according to an embodiment of the present invention.
FIG. 2 is a format diagram of a voice / FAX packet stored in a transmission / reception buffer of the voice packet communication device according to the present embodiment.
FIG. 3 is a schematic diagram showing the configuration and operation of a fluctuation absorbing buffer in the present embodiment.
FIG. 4 is a flowchart showing voice / FAX signal transmission control in the present embodiment.
FIG. 5 is a flowchart showing packet analysis of voice / fax signal reception control in the present embodiment.
FIG. 6 is a flowchart showing fluctuation absorption buffer amount variation control in the present embodiment.
FIG. 7 is a flowchart showing voice / fax packet reproduction control in the present embodiment.
[Explanation of symbols]
1, 2 Voice packet communication equipment
3 trunk lines
41, 42 Telephone / FAX terminal
101, 201 PCM codec controller
111,211 encoder
112,212 decoder
121,221 packet generator
122,222 packet analyzer
123, 223 Fluctuation absorption buffer
131, 231 Transmission buffer
132,232 receive buffer
133,233 Data bus
141,241 Relay Unit (WAN)

Claims (7)

In a voice / fax communication system that transmits voice / fax signals via a packet communication network,
An encoding means for encoding an audio signal or a FAX signal and identifying whether the encoded data is silent; and when the encoded data is silent, a silence flag is set in the encoded data and the packet A packet generation means for generating a packet to be transmitted to the communication network,
A fluctuation absorbing buffer that absorbs delay variation of packets received from the packet communication network, and accumulation of the fluctuation absorbing buffer in a silent section of encoded data in which the silence flag is set after switching from voice communication to FAX communication A buffer amount control means for changing the amount from a voice communication reference to a FAX communication reference, and increasing the fluctuation absorption buffer storage amount by a predetermined amount using the silent period for each decoding cycle of the encoded data ; A receiving device comprising:
A voice / fax communication system comprising:   The packet generation unit of the transmission device sets a silence flag in a packet header and transmits the packet header, and the buffer amount control unit of the reception device identifies a silence period of the encoded data by the silence flag of the reception packet header. The voice / fax communication system according to claim 1. In a voice / fax receiving apparatus that receives a packet including encoded data in which voice / silence is distinguished via a packet communication network,
A fluctuation absorbing buffer for absorbing delay variation of the packet received from the packet communication network;
After the voice communication is switched to the FAX communication, the amount of fluctuation stored in the fluctuation absorbing buffer is changed from the voice communication reference to the FAX communication reference when the encoded data of the received packet is silent, and the encoded data is decoded. A buffer amount control means for increasing the fluctuation absorbing buffer accumulation amount by a predetermined amount using the silent section for each period ;
A voice / fax receiving apparatus comprising:   4. The voice / fax reception according to claim 3, wherein a silence flag is set in a header of the received packet, and the buffer amount control means identifies a silent section of the encoded data by a silence flag in the received packet header. apparatus.   The buffer amount control means includes a first reference table for voice communication suitable for voice communication and a second reference table for FAX communication suitable for FAX communication, and the received encoded data is a voice signal and a FAX signal. 5. The voice / fax receiving apparatus according to claim 3, wherein the first reference table and the second reference table are switched depending on which one is selected. In a buffer amount control method of a fluctuation absorbing buffer provided in a voice / FAX receiving apparatus that receives a packet including encoded data in which voiced / silent is distinguished via a packet communication network,
Determine the silence / sound of the encoded data of the received packet,
After the voice communication is switched to the FAX communication, the amount of fluctuation stored in the fluctuation absorbing buffer is changed from the voice communication reference to the FAX communication reference when the encoded data of the received packet is silent, and the encoded data is decoded. Increasing the fluctuation absorption buffer accumulation amount by a predetermined amount using the silent period for each period ,
A fluctuation absorbing buffer amount control method characterized by the above.   7. The fluctuation absorbing buffer amount control method according to claim 6, wherein when the delay of the received packet exceeds a predetermined range, the fluctuation absorbing buffer amount is forcibly changed regardless of whether there is sound or silence.

Images