JPS61156949A - Packetized voice communication system - Google Patents

Abstract

PURPOSE:To minimize deterioration of the reproduced voice by providing a data-stuff controlling part that inserts a pseudo data formed by extracting and interporating the data among those in the memory of which influence on the reproduced voice is minimum when the data quantity reaches the upper or lower limit value. CONSTITUTION:When the accumulation reaches the upper limit value, the data stuff part retrieves and removes one among all the data stored in the reception memory of which influence on the reproduced voice due to the removal is the least. Therefore, the reproduced voice positions between the upper and lower limit values. In case the voice data reaches the lower limit value, the data stuff part inserts the copy of the said data or a dummy data formed by interpolating the said data in the data cycle of which influence on reproduced voice due to the insertion is the least. Because the removal and insertion occurs being limited on one among the data of received-voice LCU stored in the temporary memory of which influence on reproduced voice is the least, the deterioration of voice is reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a continuous voice transmission system using a packet-switched network.

BACKGROUND OF THE INVENTION Conventional voice packet communication systems are aimed at real-time communication of telephone voice. In the case of telephone audio, when looking at the audio on the time axis, more than 50% of the audio is silent. Therefore, when the input audio is digitized and packetized on the transmitting side, silence is determined and silent data is not output to the network. By this,
A conventional voice packet communication system that reduces traffic will be described with reference to FIG. Figure 4 shows a packet voice line control device (hereinafter referred to as voice LC).
Call it U. ), and suppose that one voice line can be exchanged using this device.

The input analog audio is digitized in synchronization with the local clock 1o in the audio LUU in the A/D section 11,
12 transmission FiF○ (first.

The signal is input to the sound-in, 7-foot-out memory) and blocked to a fixed length, and is also input to the silence determination section 13. If the data input to the F i FO is determined to be more active, it is input to the 16 working memories by the 14 controllers. Here, bucket 17) After adding a header and chain and making it into a packet, 16 network I/F
is output to the packet switching network via.

The output voice packet is received via the network 1111
1 voice arrives at LCU. Here, for convenience, the operation of the receiving side audio LCU will be explained using the receiving section of FIG. 4.

In the receiving section, the received audio packet is received into 16 working memories via 16 network I/Fs, the header and chain are removed by 14 controllers, and the received audio packets are transferred to 17 receiving FiFs.
Enter O. In the FiFO, reproduction of received data is delayed for a certain period of time.

- This is to absorb variations in packet arrival times that occur in packet switching networks. After this, FiFO
The output is converted into analog audio by the local cross of the receiving side audio LCU in the 18 D/A sections.

If there is a silent period on the transmitting side, the packet will not be output to the network. Therefore, on the receiving side, the receiving FiFO waits for the packet to arrive until the packet arrival variation absorption time, and if the packet does not arrive (if the receiving FiFO is empty), the transmitting side determines that the silent period has entered. Background noise is reproduced from a noise generator (for example,
21610).

In the above system, the A/D section sampling clock of the transmitting side audio LCU and the D/A section reproduction clock of the receiving side audio LCU are local clocks of each audio LCU, and are not synchronized. For this reason, a data slip occurs between both audio LCUs, but since the data slip occurs during a silent period according to the algorithm, it does not cause a quality deterioration of the call voice.

Problems to be Solved by the Invention In such an audio packet communication system, when transmitting continuous audio such as music, there are almost no silent periods, and it is also necessary to transmit minute level signals in terms of audio quality. , the transmitting side audio LCU does not perform silence determination. At this time, because there is a speed difference between the local clocks of the transmitting and receiving audio LCUs, the receiving audio LCU receives a slip in the received data (a state in which the audio data cannot be played back completely on the receiving end, or a state in which the audio data is lost). ) occurs, resulting in noise in the playback audio. Moreover, since it is uncertain where the audio data slip occurs, there is a problem in that noise occurs when the audio level is high.

Means for Solving the Problems The present invention solves the above problems by
a reception memory for received audio data, a function to monitor the amount of audio data occupied in the memory, a function to search for stored data, and an upper limit value for the amount of received audio data to occupy in the memory in advance; A lower limit value and a margin area up to the lower limit value are set, and data playback is performed sequentially starting from the first data when data is accumulated to the middle of the upper limit value and lower limit value, and when the amount of data reaches the upper limit value or the lower limit value. If you reach
A data stuff control section is provided for inserting pseudo data by removing or interpolating data in the memory that has the least influence on reproduced audio.

Effect of the Invention With the above-described configuration, even if a data slip occurs due to a frequency error between both the clocks of the transmitting and receiving audio LCU, the most suitable for the reproduced audio out of the received audio data accumulated in the receiving memory up to the time when the data slip occurs. - It is also possible to select data that does not have any influence as slip data and remove or insert it, thereby reducing the deterioration of the reproduced audio.

Example.

FIG. 1 is a block diagram showing an embodiment of an audio LCU according to the present invention. In FIG. 1, 1 is a local clock within the audio LCU, and in synchronization with this, the input analog continuous audio is digitized by the A/D section 2. The output from 2 is converted into a fixed length block in the transmitting FiFO 3,
It is input into the working memory 4. This data is packetized by the controller 5 and output to the packet switching network via the network I/F 6. In the case of continuous audio, silence determination is not performed. The sent voice packet is the destination voice L
Arrive at CU. The operation of the destination voice LCU will be explained using the receiving section shown in FIG.

At the destination voice LCU, from the packet switching network, the network i/F6
Voice packets arrive via This packet is input to the working memory 4, the packet is disassembled by the control unit 6, and only the audio data is input to the reception memory. After a certain amount of input data is accumulated in the reception memory, it is output to the D/A unit 8 in order from the oldest data, and the original analog audio is reproduced using the local clock of the destination LCU. At this time, the data stuffing unit 9 monitors the amount of received data in the reception memory, and performs control so that the amount of data occupied in the reception memory is approximately constant, as shown below.

Figure 2 shows the internal state of the logical memory. Continuous audio such as music is a one-way communication, and telephone calls do not require real-time communication, so it is possible to have a large receiving memory capacity. It is. This reception memory also has a function of absorbing variations in packet arrival times that occur in a packet switching network.

When the received audio data in the reception memory is accumulated to the middle point between the upper limit and the lower limit in FIG. 2, the data is reproduced from the earliest data under the control of the data stuff control section, similar to FiFO. FiFO position of data output to D/A section
) Name it Knob.

Next, the processing of the data stuff section 9 will be explained using FIG. 3, which shows changes in the amount of received audio data in the reception memory.

After a certain amount of time has elapsed since the start of audio playback,
An error occurs between the amount of data input into the reception memory (determined by the sampling clock of the transmission audio LCU) and the amount of data to be reproduced (determined by the reproduction clock of the reception audio LCU), and the amount of accumulated data is at the upper limit. approaching the value or lower limit. If it approaches the upper limit, transmit audio LC
When the clock of U is slightly faster than that of the received audio LCU, it approaches the lower limit value when it is the opposite. The case where the amount of accumulated audio data reaches the upper limit value is shown in Figure 3a. Shown below. At this time, the data stuff section searches all the data in the reception memory for data that is unlikely to affect the playback sound quality by data removal (for example, data at a level of silence or near silence) and removes it. . As a result, the position of the reproduced audio data (
FiFO) is between the upper limit and lower limit.

FIG. 3 shows a case where the accumulated amount of audio data reaches the lower limit value. At this time, the data stuff section inserts data from among all the data in the reception memory, as shown in FIG. Insert dummy data obtained by copying or interpolating the prefix data.

Effects of the Invention As described above, according to the present invention, data sling on the receiving side that occurs during continuous voice transmission over a packet network can be eliminated from among the voice data stored in the temporary memory of the received voice LCU. Since the generation is limited to those that have the least influence on the reproduced audio, deterioration in audio quality can be reduced.

[Brief explanation of drawings]

FIG. 1 is a block configuration diagram of a voice LCU in a voice packet communication system according to an embodiment of the present invention, FIGS. 2 and 3 are explanatory diagrams of temporary storage operation of the received voice LCU for explanation, and FIG. 4 is a conventional FIG. 2 is a block diagram of an audio LCU of FIG. 2...A/D converter, 3...Transmission Fi
F○, reception memory, 8...D/A converter, 9.
...Data staff control section. Name of agent: Patent attorney Toshio Nakao and 1 other person
1 Figure q 2nd Figure 1 1.

Claims (1)

[Claims] In a system that transmits continuous audio via a packet switching network between a pair of packet audio line controllers each operating with an independent clock, the receiving packet audio line controller includes a receiving memory for receiving audio data. , has a function of monitoring the amount of data occupied in the reception memory and a function of searching the contents of the accumulated data, and has a function of monitoring the amount of data occupied in the reception memory, and determines in advance the upper limit, the lower limit, and the margin for the lower limit of the amount of data occupied in the reception memory. is set, and from the time when the data is accumulated to the middle of the upper limit value and the lower limit value, the first-arrived data will be played back sequentially, and when the amount of accumulated data reaches the upper limit value, the data in the reception memory will be played back. When the data that has the least influence on playback is searched for and removed, and the amount of accumulated data reaches the lower limit, the data section that has the least influence on playback is searched for among the data in the reception memory, and this is removed. A voice packet communication method that controls the insertion of pseudo data.