JPH03109840A - System and apparatus for voice packet communication - Google Patents

Abstract

PURPOSE:To minimize the deterioration in voice by adding information such as a packet sequence number, voiced state, discrimination of end of voice and noise and noise level information to a voice packet from a sender. CONSTITUTION:A voice packet 1 consists of a header part 2 and an information part 4. A coded voice signal is stored in the information part 1 and the header part 2 stores additional information representing distinguishment of voiced period/unvoiced period and its state. Sequence indication and voice identifier are added for the voiced period and a packet 1 including unvoice identifier is sent at the end of the voiced period and when a noise level changes, a packet 1 including noise identifier is sent. When the packet including the identifiers is received at the receiver side, the voiced period, the unvoiced period and noise are reproduced with fidelity according to the identifiers and the missing of packets for the voiced period is interpolated. Thus, the packet for the voiced/ unvoiced periods is reproduced similarly to the case with the sender side.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a voice packet communication method and an apparatus thereof.

(Prior art) In conventional audio packet communication systems, when processing an audio signal on the transmitting side, it is determined whether the audio signal is a sound section or a silent section, and the audio signal in the sound section is divided into sections with a predetermined length or less. It was divided into blocks and stored in the information section of the packet, a header containing destination information etc. was added to the blocks, and the packets were sent out to the communication network. A series of sequence numbers is added to the packet.

When the receiving side receives the packet, it restores it to an audio signal and reproduces the audio in the sound section. At this time, in principle, noise is reproduced in the portion where no packets have arrived as a silent section. More specifically, after receiving a packet, if packets are not continuously received in a plurality of packet sections, it is determined that a silent section has arrived and noise is reproduced. If a missing number is detected in the sequence number of a packet received when it is supposed to be a voice section, this means that the packet with the missing number has been discarded within the network. Interpolate the missing parts of the audio. This also applies when no packet is received for one packet section after receiving a packet. In other words, in the conventional audio packet playback method, if one packet is not received during a sound interval, it is determined that the audio signal is missing and the audio signal is interpolated, and if no packet arrives in two or more packet intervals, there is no sound. It is determined that the section has been entered and the noise is played back.

(Problem to be Solved by the Invention) In this way, in the conventional method, in the first packet section in which no packet arrives following the arrival of a packet, audio interpolation is performed by discarding the packet, and in the subsequent packet interval in which no packet arrives, voice interpolation is performed. Noise is reproduced in the silent section only after entering the packet section.

Therefore, for example, when the transmission of a series of packets is truly interrupted for two or more packet intervals on the transmitting side, that is, when a silent interval is entered, the receiving side adds a voice by interpolation to the beginning of the silent interval. Put it away. Furthermore, silence in one packet section sandwiched between sound sections is treated in the same manner as packet discard, and interpolated audio is reproduced. Furthermore, if two or more consecutive packets are discarded within one communication network, the receiving side has the disadvantage that the second packet section is determined to be a silent section even though it is a sound section.

SUMMARY OF THE INVENTION An object of the present invention is to eliminate these drawbacks and provide an audio packet communication system and apparatus for reproducing sound/silent sections in the same manner as on the transmitting side.

(Means for Solving the Problems) In order to solve the above-mentioned problems, the present invention provides an audio packet communication system that transfers audio packets. , and assembles the audio signal of the voice section into a voice packet, and when assembling the voice packet, adds a voice identifier indicating that the voice packet is a voice packet and a sequence display indicating the order of the voice packets to the voice packet. including,
When a silent section is detected, a packet containing a silence identifier indicating the end of a sound section is sent, and when the level of noise in the silent section changes significantly, a packet containing a noise identifier that requests a change in the level of noise to be generated is sent. When the receiving side receives one of these packets, it disassembles it and uses the voice and silence identifiers included in the disassembled packet to distinguish the voice and silence intervals. During the sound period, the audio signal included in the audio packet is played back, and if a gap is detected in the sequence display order during playback of the audio signal, the audio signal of the missing part is interpolated, and the audio signal is played back during the silent period. When a noise identifier is detected, the level of generated noise is changed.

1 According to the present invention, there is provided a packet communication system that implements such a voice packet communication method, and its transmitting device and receiving device.

(Function) According to the present invention, a sequence display and a sound identifier are added to the packet as additional information in a sound section, and a packet including a silence identifier is sent at the end of the sound section, and the noise level changes. Then, it sends out a packet containing the noise identifier. On the receiving side, when a packet containing these identifiers is received, the sound section, silent section, and noise are faithfully reproduced according to the identifiers, and missing packets in the sound section are interpolated.

(Example) Next, an example of the audio packet reproducing method and apparatus according to the present invention will be described in detail with reference to the accompanying drawings.

In the audio packet communication system according to the present invention, an identifier is contained in addition to the sequence number in an additional information section such as a header section of the packet. This identifier basically includes a voice identifier, a silence identifier, and a noise identifier, and on the transmitting side of the packet, a voice identifier is added in a voice section, and a silence identifier is added at the beginning of a silent section. If the noise level changes significantly in a silent section, a noise identifier is added. These identifiers allow the receiving side to faithfully reproduce the audio signal in the active period and the noise in the silent area, and to appropriately interpolate packet loss in the active area.

FIG. 2 shows an example of the structure of the audio packet 1 applied to this embodiment. The audio packet l is composed of a header part 2 and an information part 4. The information section 1 accommodates encoded audio signals. The header section 2 stores additional information indicating whether the section is a sound section or a silent section and its status. In this embodiment, this additional information includes routing information RT, sequence number SQ, packet type TYP, and noise level NL. Routing information RT includes the own address, destination address, etc. for passing through the packet communication network.
3 Mmm. The sequence number SQ is an order indication that defines the order of packets l, and in this example takes a 4-bit cyclic number from 0 to 15. Packet type TYP is 2 in this embodiment.
It is the identifier of the bit, and "00" indicates the sound section■,
"Ol" is the end display E of the sound section, and r IIJ is the display N (FIG. 3) of the silent section, that is, the noise section. Na, i
rlOJ is an unused code. Noise level NL is 2
This is an identifier that indicates a noise level in four levels from 0 to 3 in bits.

An embodiment of an apparatus for implementing the voice packet communication system according to the present invention will be described with reference to FIG. Apparatus 1 shown in FIG.
0 is a signal processing device that is located between lines 12TJ and 121 (which transmit and receive PCM audio data 16 with a cycle of 8 kH7, and packet lines 14T and 14R which transmit and receive audio packet 1), and which assembles and disassembles packet 1. The packet lines 14T and 14R are accommodated in a packet communication network (not shown).The packet communication network is a communication network that transfers the packet l according to the information carried by the header section 2 of the packet. Due to congestion, there is a possibility that 4 baquerats will be discarded.

The signal processing device 10 has a transmitting circuit 18 and a receiving circuit 20. The former 18 is a circuit that is connected between the transmission lines 1.2T and 1.4T and assembles audio data received from one line 12T into a packet 1 and transmits the packet 1 to the packet line 14T. The latter, that is, the receiving circuit 20 is connected between the receiving lines 14R and 12R, and decomposes the audio packet l received from the packet line 14R and sends it as an audio signal to the line L2.
This is the circuit that sends the signal to R. For the purpose of explaining the invention, the signal processing device 1 does not have to include both circuits 18 and 20. Send 4 to the transmission line 14T at one end of the packet communication network.
It is sufficient for the explanation of the present invention that eight circuits 18 are accommodated and a receiving circuit 20 is connected to another receiving line 14R at the other end.

The transmitting circuit 18 has a packetizing circuit 22, which is connected to the transmitting line 12T and divides the audio data 16 received therefrom into blocks of a predetermined data length or less, and sends the blocks to the packet assembling circuit 24.

Line 12T is level detection circuit 26 and presence/silence determination
5 is also connected to a circuit 28, the former 26 being a circuit for detecting the signal level of the audio data 16, especially the level of background noise in a silent state. The presence/silence determination circuit 28 has an input 1
This is a circuit that determines whether the signal of 2T audio data 16 is sound or silent.

The output 27 of the level detection circuit 26 and the output 29 of the presence/silence determination circuit 28 are connected to the input of the header generation circuit 30. The header generation circuit 30 is a circuit that generates data to be inserted into the header portion 4 of the audio packet l based on the level detected by the level determination circuit 26 and the result determined by the presence/silence determination circuit 28.

The packet assembling circuit 24 is a circuit that assembles the packet 1 by storing the audio data manually generated by the packetization circuit 22 in the information section 4 and inserting the header information generated by the header generation circuit 30 into the header section 2. . The assembled packet l is transmitted from the 1 assembly circuit 24 to the packet line 14T.

The receiving circuit 20 has a packet disassembly circuit 32. The packet disassembly circuit 32 is connected to the packet line 14R, and when it receives the audio packet l from it, it disassembles the packet 1 and sends the data in the header part 2 to the header analysis circuit 34, and also sends the audio data in the information part 4 to the header analysis circuit 34. This circuit sends the packet-time storage circuit 36 to the packet-time storage circuit 36. Packet-time storage circuit 3
6 is a buffer circuit that temporarily stores this audio data.

The header analysis circuit 34 is a circuit that analyzes the data in the header section 2 to determine if the packet 1 is missing, whether the packet is a sound section or a silent section, and determines the noise level to be inserted into the silent section. Depending on the determination result, the header analysis circuit 34 controls the selector 38 through its control output 46 and also sets the noise level in the noise generation circuit 40. Noise generation circuit 4
0 is a circuit that generates noise at a level set by the header analysis circuit 34, and its output 42 is sent to the selector 3.
8 is connected. The readout output of the packet-time storage circuit 36 is connected to the other input 44 of the selector 38.

The selector 38 selectively selects the two inputs 42 and 44 from the line 12 in response to the control output 46 of the header analysis circuit 34.
This is a switch circuit connected to H. According to the analysis result of the header analysis circuit 34, the selector 387 transfers the packets read out from the packet-time accumulation circuit 36 in a first-in, first-out manner to the line 12R during the sound section, and retransmits the immediately preceding packet for missing packets. and perform interpolation. In the silent section, the selector 38 selects the noise output 42 of the noise generation circuit 40 under the control of the header analysis circuit 34;
The generated noise is sent to line 12R. This noise level is set in the noise generation circuit 40 according to the header analysis result of the header analysis circuit 34.

A voice packet communication method using the signal processing device IO will be explained with reference to FIG. In this figure, waveforms are drawn according to the direction in which the signal is transmitted or received, that is, from right to left in the figure.

In the transmitter audio waveform fal shown in the figure, sections 50 and 53 indicate the waveform of the audio signal, and sections 51, 52 and 54 indicate the waveform of the background noise. This waveform (al) is the PCM audio data input to the human power line 12T of the transmitting circuit 18 converted into an analog waveform.The noise level is different between sections 51 and 852. The results of signal processing and determination by the presence/silence determination circuit 28 are shown in tb+ of the same figure.Sound detection results 55 and 57 and silence detection results 56 and 58 shown here.
is output from the output 29 of the determination circuit 28.

The header generation circuit 30 uses this result to determine the packet type TYP.
The noise level NL data is generated according to the detection level of the level detection circuit 26. These helag data are given to the packet assembly circuit 24,
The circuit 24 adds a sequence number SQ, a packet type TYP, and a noise level NL as a header section 2 to an information section 4, as shown in the transmitting side sending packet fcl in the figure. In the same figure (c), the numbers 1 to 4 in the audio packet 1 indicate the sequence number SQ. As the packet type TYP, a voice display V is added in sound detection sections 59 and 64, and a time point 60 when sound is changed from sound to silence is added.
and 65, an end indication E is added. Furthermore, at a time point 62 when the noise level changes as shown in the figure between the silence detection sections 6I and 63, a noise indicator N is added as the packet type TYP, and a value representing the noise level is added as the noise level NL. Ru. In this way, audio packet 1 with header 2 added is transmitted from packet line 14T.

This voice packet I is transferred through a packet communication network (not shown) and reaches the receiving circuit 20 via the destination receiving line 14R. In the receiving side received packet (dl) shown in FIG. 3, in sections 66, 68 to 74, the state of the received voice packet l corresponds to the state of the packet l sent out on the transmitting side. However, in the section 67, the communication This means that the packet was discarded within the network.

In the receiving circuit 18, in the process of sequentially reproducing the received packets fdl, the header analysis circuit 34 obtains the reception unique/silent section identification result tel. In the analysis circuit 34, the voice display ■ included in the header part 2 of the received packet l
Sound sections 75 and 77 are identified correspondingly. In addition, packet discard is detected by monitoring the order of sequence numbers SQ in the header analysis circuit 34, and for example, 0, the immediately preceding packet stored in the packet-time storage circuit 36 is detected. Interpolate by using the signal of The silent section 76 is identified by detecting the end indication E by the header analysis circuit 34. Note that there is a possibility that a packet with an end indication is discarded within the packet network. As a countermeasure, the transmitting circuit 18 may be configured to redundantly transmit a plurality of packets having the end indication E.

The audio waveform reproduced by the receiving circuit 20 from the received audio signal (Cd) is converted into an analog signal and is shown in FIG. As will be seen from now on, audio signals 79.8o and 83 are reproduced for the sound sections 75 and 77 based on the audio data of the voice display ■ packet. For the discarded packet 67, in this embodiment, interpolation 8o is performed using the previous audio data 79. Other interpolation methods may be used. Noise reproductions 81, 82 and 84 are performed for the silent sections 76 and 78. The level of the noise to be reproduced is set in the noise generation circuit 4o based on the data of the noise level NL of the header section 2 analyzed by the header analysis circuit 34. Therefore, in section 81, the noise level N of the packet with noise indication N received previously
Noise is generated at a level corresponding to the value of L, and in section 82, noise is generated at a level corresponding to the value of the noise level NL of the packet of noise indication N received at that time. In this way, the noise level changes adaptively. Through the above processing, the receiving circuit can detect whether there is a sound or not.
The audio waveform of the silent section is played back.

The reproduction sequence of the sound/silence section in the receiving circuit 20 will be explained with reference to FIG. The receiving circuit 2° waits for packet 1 to be received during the packet receiving process.
ol, if packet l does not arrive, it is determined that it is a silent section (1,02), and the selector 38 selects the noise generation circuit 40.
is connected to the four-wire 12R and reproduces noise (104).

If there is a packet 1, the header analysis circuit 34 checks whether the type TYP of the header part 2 is voice display ■. [1061° If it is not voice display V, then checks whether the type TYP is noise display N. Examine (1081° If the noise display is not N, silent interval processing ■ returns to 12; if the noise display is N, the noise level of the noise generation circuit 40 is set to the value indicated by the noise level NL, and the process returns to silent interval processing 102.

f10 if the type TYP of header part 2 is voice display V
61, load its sequence number Sa and fl12+
, determines that it is a sound section (1141), and reproduces the audio data included in the information section 4 of the audio packet l (116
1° After a predetermined time from the arrival of packet 1 in step 100, the standard sequence number of the header analysis circuit 34 is set to 1.
Advance the unit (1181, check the presence or absence of packet 1 (
1201° If there is no corresponding packet 1, it is interpolated and reproduced (122). If the packet l exists, it is checked whether the type TYP is voice display V or not. If this is not the voice display ■, if the type TYP is the end display E, the process returns to the silent section process 102, and if it is not the end display E, the process changes to the interpolation playback process 122 (f126). Type TY
If P is a voice display V and the sequence number SQ is consecutive to the previous one, the process returns to the voice packet playback process 116. However, if there is discontinuity, that is, three packets are missing, the process returns to the interpolation reproduction process 122.

(Effects of the Invention) As explained above, in the present invention, information indicating the packet sequence number, voice, end of voice, noise, and noise level is added to the transmitted voice packet at 0111, and the receiving side Using these additional information, the voice packet is reproduced by faithfully identifying the sound section and the silent section. Thereby, audio deterioration can be minimized. Furthermore, since the sound interval can be accurately identified, packet discards can be appropriately identified.

[Brief explanation of drawings]

FIG. 1 is a functional block diagram showing an example of the configuration of a signal processing device for voice packet communication according to the present invention, and FIG. 2 is a diagram showing an example of the configuration of a voice packet applied to the signal processing device shown in FIG. 1. 3 is a diagram showing an example of an audio packet reproduction sequence in the same embodiment, and 4. FIG. 4 is a flow diagram of an audio packet reproduction operation in the receiving circuit of the same embodiment. 2. 10. 18. 20. 24. 26. 8 0 2 34°38. OL SQ. TYP. Major. Description of ``'''' Audio packet header section Signal processing device Transmission circuit Receiving circuit Packet assembly circuit Level detection circuit Presence/silence judgment circuit Header generation circuit Packet disassembly circuit Header analysis circuit Selector Noise generation circuit Noise level Sequence number Packet type 5

Claims (1)

[Claims] 1. In a voice packet communication system that transfers voice packets, the transmitting side of the voice packet detects a sound section and a silent section of the voice signal, and converts the voice signal in the voice section into a voice packet. Assembling, when assembling the voice packet, a voice identifier indicating that the voice packet is a voice packet and a sequence indicator indicating the order of the voice packet are included in the voice packet, and when the silent section is detected, the voice packet is voiced. A packet containing a silence identifier indicating the end of a sound interval is transmitted, and when the level of noise in the silence interval changes significantly, a packet containing a noise identifier requesting a change in the level of noise to be generated is transmitted, and these packets are When receiving one of these packets, the receiving side disassembles it, identifies a sound interval and a silent interval using the voice identifier and silence identifier included in the disassembled packet, and determines the voice interval. Then, the audio signal included in the audio packet is played back, and when a loss is detected in the order of the sequence display during playback of the audio signal, the audio signal of the missing part is interpolated, and a noise identifier is detected in the silent section. Then, the voice packet communication method is characterized by changing the level of generated noise. 2. In the system according to claim 1, on the transmitting side,
When transmitting a packet including a noise identifier, a noise level identifier defining the level of noise to be generated is added, and the receiving side generates noise at a noise level corresponding to the noise level identifier. Voice packet communication method 3: A voice packet communication system that transfers voice packets, the system comprising: a transmitter that assembles voice signals into voice packets and transmits the voice packets; receives and decomposes the voice packets; a receiving device that reproduces an audio signal, and the transmitting device includes: a first detection device that detects a sound section and a silent section of the audio signal; and a second detection device that detects a noise level in the silent section. , a packet assembling means for assembling the voice signal of the voice section into a voice packet and transmitting the voice packet, and a voice presence identifier indicating that the voice packet is a voice activity in response to the first and second detection means; additional information generating means for generating additional information including a silence identifier indicating the end of a section, a noise identifier requesting a change in the level of noise to be generated, and a sequence display indicating the order of the audio packets; includes a sound identifier and a sequence indication in the audio packet when assembling the audio packet, and when the first detecting means detects a silent section, the silence identifier is included in the packet, and the second detecting means detects a silent section. detecting a significant change in the level of noise in the packets, the noise identifier is included in packets, the packet assembling means assembles and transmits these packets, and the receiving device receives and disassembles these packets. additional information identifying means for identifying the additional information included in the disassembled packet; playback means for reproducing the audio signal included in the disassembled audio packet; noise generating means for generating a sound, and the additional information reproducing means identifies a sound section and a silent section using a sound identifier and a silence identifier included in the additional information, and in the sound period, the noise generating means generates a noise that is not included in the voice packet. When a omission is detected in the order of the sequence display during reproduction of the audio signal, the reproduction means interpolates the audio signal of the missing part, and a noise identifier is set in the silent section. A voice packet communication system characterized in that, upon detection, the level of noise generated by the noise generation means is changed. 4. A voice packet transmitting device that assembles voice signals into voice packets and transmits the voice packets, the device comprising: a first detection means for detecting a sound section and a silent section of the voice signal; a second detecting means for detecting a voice signal; a packet assembling means for assembling a voice signal in a sound section into a voice packet and transmitting the voice packet; additional information that generates additional information including a sound identifier indicating the end of a speech section, a silence identifier indicating the end of a speech section, a noise identifier requesting a change in the level of noise to be generated, and a sequence indication indicating the order of the voice packets. generating means, the additional information generating means includes a voiced identifier and a sequence indication in the voice packet when assembling the voice packet, and when the first detection means detects a silent section, the additional information generating means includes a voiceless identifier and a sequence indication in the voice packet. and when the second detecting means detects a significant change in the level of noise in the silent section, the noise identifier is included in the packet, and the packet assembling means assembles and transmits the packets. Packet transmitter. 5. In an audio packet receiving device that receives an audio packet containing an audio signal, decomposes it, and reproduces the audio signal, the device receives a packet to which additional information is added and decomposes it. additional information identifying means for identifying the additional information included in the disassembled packet; and reproducing means for reproducing the audio signal included in the disassembled audio packet, the additional information comprising: The voice packet includes a voice identifier indicating that the voice packet is a voice packet, a silence identifier indicating the end of a voice section, a noise identifier requesting a change in the level of noise to be generated, and a sequence display indicating the order of the voice packet. , the device further includes a noise generating means for generating noise in a silent section, and the additional information reproducing means identifies the sound section and the silent section based on a sound identifier and a silent identifier included in the additional information, During the sound period, the reproduction means reproduces the audio signal included in the audio packet, and when a omission is detected in the order of the sequence display during reproduction of the audio signal, the reproduction means reproduces the audio signal of the missing part. The voice packet receiving device is characterized in that the level of noise generated by the noise generating means is changed when a noise identifier is detected in a silent section by interpolation.