JP4406382B2 - Speech coding selection control method - Google Patents

Description

  The present invention relates to a voice coding selection control method for controlling a coding calculation amount in voice packet communication in an IP network, a voice packet transmitting device, a receiving device, a program, and a recording recording the same. It relates to the medium.

In the Internet, video conferencing combined with voice communication services and image communication using VoIP (Voice over Internet Protocol) technology (hereinafter simply referred to as voice packet communication) that transmits voice signals in real time using IP (Internet Protocol) packets Although services are spreading, there is a problem that sound interruption occurs during communication.
As a cause of sound interruption in voice packet communication, the currently widely used Internet is a best-effort IP network, and there is no guarantee that the packet will surely arrive at the destination. May be lost in the middle of communication (packet loss), and packets may not arrive at the receiving side and sound to be reproduced cannot be created, which may cause sound interruption. As a technique for suppressing packet loss due to such network congestion, it is effective to use low bit rate speech coding during network congestion. For example, Patent Document 1 proposes a method for preventing packet loss by evaluating a network state based on a packet loss rate or the like and switching an encoding method according to the network state.

In order to reproduce the audio at the receiving terminal in real time, it is necessary to transmit the audio data from the transmitting terminal at a fixed time interval (hereinafter referred to as a frame, which is often a time interval of about 10 to 20 ms). It is necessary to finish the encoding process within the time interval of the frame. However, since low bit rate speech coding requires a precise analysis of the signal and requires a large amount of calculation processing, for example, in the situation where a program that simultaneously encodes image data and performs transmission processing cooperates with the image data. As the amount of processing increases, the load on the transmitting terminal tends to increase. As a result, the encoding process of the audio data cannot be performed at the time interval of the frame, and the packet storing the audio data (hereinafter referred to as the audio packet in this specification) is transmitted from the transmission side at the time interval of the frame. As a result, packet loss occurs due to the inability to perform playback, and even when the network is free, there is a problem in that real-time playback cannot be performed on the receiving side and sound interruption occurs.
Japanese Patent Laid-Open No. 2002-300274 “Gateway Device and Voice Data Transfer Method”

  The present invention has been made in view of the above-described problems, and an object of the present invention is to realize high-quality voice packet communication by suppressing packet loss caused by a load on a communication terminal when performing voice packet communication. .

The essence of the present invention is that, in voice packet communication in an IP network, the calculation time required for voice encoding at the transmission terminal is measured, and when the calculation time is a predetermined time or more, the load on the transmission terminal is large. A speech coding selection control method for suppressing packet loss due to an increase in the load on the terminal by performing control for selecting a low-load speech coding method and preventing sound interruption, and using this control method A voice packet transmitting device and a receiving device that operate in this manner are proposed.
The speech coding selection control method proposed in the present invention receives discrete speech musical tone samples as input, encodes discrete speech musical tone signal samples, packetizes the encoded digital code, and transmits the packets. A voice coding selection control method for dynamically switching at least two kinds of voice coding systems having different calculation amounts necessary for coding a voice music signal sample, and voice coding a voice section every predetermined time. The method includes a step of measuring time, and a step of selecting a speech coding method with a small amount of calculation when a time required for speech coding becomes longer than a predetermined time.

According to the present invention, the speech coding selection control method further includes the step of returning to the speech coding system with a small amount of information to be transmitted when the time required for speech coding is shorter than a predetermined time. .
In the speech coding selection control method according to the present invention, in the step of selecting a speech coding method with a small amount of computation, when the computation time is longer than a certain threshold, the image coding processing that is simultaneously operating is temporarily stopped. A step of performing control .

  According to the present invention, when voice communication is performed in an IP network, the amount of encoding calculation varies depending on the calculation time required for voice encoding in order to generate packet loss due to an increase in the load on a terminal that causes sound interruption. Control can be performed by performing control for switching the audio encoding method and control for switching between operation and non-operation of image encoding.

  The voice packet transmitting apparatus that executes the voice coding selection control method according to the present invention and the voice packet receiving apparatus that receives the voice packets transmitted by the voice packet transmitting apparatus, decodes them, and reproduces the voice are constructed entirely by hardware. be able to. However, in order to more simply realize this, the voice packet transmission program and the voice packet reception program proposed in the present invention are installed in the computer, the CPU provided in the computer decodes these programs, and the program is executed. An embodiment in which the device functions as a voice packet transmitting device and a voice packet receiving device is the best embodiment.

In order for a computer to function as the voice packet transmitting apparatus according to the present invention, the computer has a coding time measuring unit that measures the calculation time required for voice coding of a voice section every predetermined time, and the coding time measuring unit measures the calculation time. Based on the calculated computation time, when the computation time required for speech encoding is longer than a predetermined time, a coding method selection unit that selects a speech coding method with a small amount of computation, and a code representing the selected speech coding method A voice packet transmission unit is configured to add the encrypted identification signal to the voice coded signal and send it as a packet signal.
In order for a computer to function as the voice packet receiving device according to the present invention, the computer has a voice packet receiving unit, a voice coded sequence separating unit that separates a voice coded signal and a coded identification signal, and codes having different calculation amounts. Decoding operation in which at least two types of decoding units that respectively decode encoded data encoded by the encoding method and a decoding unit selected by the encoded identification signal divided by the audio encoded sequence separation unit Build a selector.

  Hereinafter, embodiments of each part will be described in detail with reference to the drawings.

FIG. 1 shows a first embodiment of a voice packet transmitting apparatus 100 according to the present invention. The voice packet transmitting apparatus 100 according to the present invention includes a timer 101, an encoding unit 102, an encoding time measuring unit 103, an audio encoding method selection unit 104, an encoding method selection unit 105, and an audio packet creation unit 106. The voice packet transmission buffer 107 and the voice packet transmission unit 108 are provided.
The timer 101 outputs time data every certain time (for example, 1 ms). The time data may be in a format in which the relative time between two times is known. For example, time data that is incremented by 1 every certain time starting from 0 (for example, 1 ms) is output.

  The encoding unit 102 incorporates a plurality of encoding methods having different calculation amounts per unit time required for encoding the input speech signal. In the present embodiment, a case will be described in which the encoding method A of the encoding means 102A is ITU-T G.729 and the encoding method B of the encoding means 102B is ITU-T G.711. The characteristics of these encoding methods are shown in FIG. As is apparent from FIG. 4, the encoding scheme ITU-T G.729 has a characteristic that the amount of calculation for encoding is large, instead of a small amount of information to be transmitted. Also, the encoding scheme ITU-T G.711 has a characteristic that the amount of computation for encoding is small, instead of the large amount of information to be transmitted. Here, a configuration having two types of speech encoding has been described, but three or more types of encoding may be applied.

An input speech signal (a digitized discrete speech tone signal sample) is input to the encoding means 102A or 102B and encoded according to the selection of the encoding method selection unit 105. In a steady state, it is assumed that the encoding unit 102A that performs an encoding operation with an encoding (G.729) with a small amount of information to be transmitted in consideration of the influence on the transmission path is initially set to be selected and operated. In the steady state, the encoding scheme selection unit 105 inputs the input speech signal to the encoding unit 102A and operates the encoding unit 102A.
The encoding time measuring unit 103 measures the time required for encoding by the encoding unit 102 using the time data output from the timer 101. The time required for encoding is, for example, when the input audio signal frame length (the length for dividing a digital code string obtained by digitizing an audio signal into predetermined time intervals) is 20 ms, N samples of data existing in the 20 ms are Refers to the time to encode. The encoding time measuring unit 103 measures the time required for encoding by the encoding unit 102A or 102B operating in the encoding unit 102, and outputs the measurement result to the speech encoding method selection unit 104. The computation time required for speech coding is obtained by computation time (= (coding end time) − (start time)).

  The speech coding scheme selection unit 104 outputs a coding identification signal according to the measurement time given from the coding time measurement unit 103, and switches the state of the coding scheme selection unit 105 according to this coding identification signal. FIG. 5 shows an example of a coding identification signal output from the speech coding method selection unit 104. In the example shown in FIG. 5, when the calculation time is greater than 5 ms, 1 theory is assigned to the encoded identification signal, and when the calculation time is less than 5 ms, 0 logic is assigned to the encoded identification signal. This encoding identification signal may be anything as long as the encoding method prepared on the transmission side and the reception side can be designated.

By inputting the coding identification signal to the coding method selection unit 105, the coding method selection unit 105 determines the supply destination of the input speech signal according to the coding identification signal. That is, in this example, encoding is performed with G.729, which has a small amount of information to be transmitted when the calculation time is less than 5 ms. The state in which the input speech signal is input to the encoding means 102A is maintained.
On the other hand, when the computation time is longer than 5 ms, the state is switched to a state where the input speech signal is supplied to the encoding means 102B having a small computation amount.
The voice coded sequence data coded by the coding unit 102 and the coded identification signal given a delay of one frame by the one-frame delay unit 109 are input to the voice packet creation unit 106. The voice packet creation unit 106 receives voice coded sequence data as in the conventional VoIP, and adds a packet header including a packet number, a coding identification signal, time information, etc. for distinguishing the packets. Output data. This is shown in FIG.

The voice packet transmission buffer 107 receives the voice packet data, buffers the voice packet data as in the case of real-time voice communication in a normal IP network, and outputs the voice packet data at the same time interval as the frame. The voice packet sending unit 108 receives voice packet data as input and sends the voice packet data to the network.
FIG. 2 shows an embodiment of a voice packet receiving apparatus according to the present invention. In this embodiment, a case where the voice packet receiving device 200 is configured by the voice packet receiving unit 201, the voice packet receiving buffer 202, the voice coded sequence separating unit 203, the decoding selecting unit 204, and the decoding unit 205 is shown. Show.

The voice packet receiving unit 201 receives voice packet data from the network and outputs the received voice packet data.
The voice packet reception buffer 202 receives voice packet data, sorts the packets according to the packet number attached to each packet, and has the packet number transmitted earliest among the stored voice packets. Output data.
The speech coded sequence separation unit 203 receives speech packet data as input, and separates and extracts the packet header and speech coded sequence data from the speech packet data in the format shown in FIG. Further, the encoded identification signal is extracted from the packet header, and this encoded identification signal is input to the decoding selection unit 204. The decoding selection unit 204 determines the supply destination of the speech encoded sequence data input from the speech encoded sequence separation unit 203 according to the encoded identification signal.

  That is, the decoding unit 205 includes decoding means 205A and 205B for decoding encoded string data encoded by encoding methods with different calculation amounts, and according to the encoded identification signal, these decoding means 205A and The encoded string data is input to any one of 205B, and the decoding operation is executed. Here, if the coded sequence identification signal is 0, the coding means 205A is selected, and G.729 decoding with a small amount of information to be transmitted is executed. If the encoding identification signal is 1, the decoding unit 205B is selected, decoding based on G.711 that requires a small amount of calculation for encoding is performed, and an audio signal is output.

  As described above, the voice coding selection control method according to the present invention and the voice packet transmitting apparatus that operates using this control method always perform coding operation using the coding scheme G.729 with a small amount of information to be transmitted. Although the communication is performed with a small amount of transmission information, the time required for encoding per unit time exceeds a predetermined value due to an increase in the calculation processing of the computer on the transmission device side, and the limit of occurrence of sound interruption occurs. When approaching, encoding system selection control is executed so that the system immediately switches to the encoding system G.711, which requires only a small amount of calculation. As a result, while the addition of computers on the transmission side is increasing, encoding is performed with an encoding method with a small amount of calculation, and the transmission interval of voice packets is maintained at a normal interval, thereby preventing sound interruptions. it can.

  When the computation time for speech coding returns to a state shorter than a predetermined value (5 ms), the coding system is also restored to the coding system that requires less information to be transmitted. In addition, although the case where G.729 of an encoding method having a large calculation amount compared with G.711 is applied as an encoding method with a small amount of information to be switched when the time required for speech encoding is small has been described. The present invention is not limited to the method, and other encoding methods can be applied. The point is that when the time required for speech encoding is small, the amount of calculation of the encoding method to be switched does not matter.

6 and 7 show a second embodiment of the present invention. This embodiment shows an embodiment in which the present invention is applied to a voice packet transmitting apparatus and receiving apparatus having a transmission state and receiving function of image packet data. FIG. 6 shows an audio packet transmitting apparatus having an image packet data transmitting function, and FIG. 7 shows an embodiment of the receiving apparatus.
Reference numeral 100 shown in FIG. 6 denotes the voice packet transmitting apparatus according to the present invention described with reference to FIG. Therefore, also in this embodiment, the voice packet transmitting apparatus 100 includes the timer 101, the coding unit 102, the coding time measuring unit 103, the voice coding method selection unit 104, the coding method selection unit 105, and the voice packet creation unit. 106, a voice packet transmission buffer 107, a voice packet transmission unit 108, and a one-frame delay unit 109, and execute a voice packet transmission operation similar to that of the embodiment of FIG.

The configuration characteristic of the second embodiment is that the image encoding selection unit 604 and the image encoding control signal output from the image encoding selection unit 604 stop the encoding operation of the image encoding unit 611 as necessary. The point is that an encoding operation control unit 610 for controlling the control is provided.
The image coding selection unit 604 receives the measurement result of the coding time measuring unit 103 as input, and inverts the state of the image coding control signal when the calculation time for speech coding reaches a predetermined value or more. The encoding unit 611 stops the encoding operation.
That is, the speech coding scheme selection unit 104 selects a coding scheme according to the rules shown in FIG. Here, when the computation time for speech coding becomes longer than 5 ms, for example, the speech coding method (for example, G.711) with a small amount of computation processing is switched. By switching to speech coding with a small amount of calculation processing, the calculation time changes in the direction of shortening. Even if the speech coding method operates with a method that requires a small amount of computation (for example, G.711), the computation time for speech coding becomes longer again as the load on the computer constituting the transmitter increases more and more. Change direction.

  In the present invention, when the time required for speech coding (time for coding speech data for one frame) exceeds a predetermined value in spite of operating in a coding method with a small amount of calculation, the image data Control to stop the encoding operation is performed. FIG. 8 shows a control rule for performing stop control of image data. In the example shown in FIG. 8, when the calculation time for speech encoding exceeds 15 ms, the logic of the image encoding control signal is inverted to 1, and the image encoding operation control unit 610 is turned off by this image encoding control signal. The supply of the input image signal to the image encoding unit 611A is cut off, and the encoding operation of the image encoding unit 611A is stopped.

When the encoding operation of the image encoding unit 611A is stopped, the load on the computer is reduced by the amount corresponding to the encoding operation, and at least the transmission state of the voice packet is maintained. When the computation time for speech encoding returns to the steady state (state of 15 ms or less), the transmission of the image packet is resumed. Further, when the computation time for speech encoding further recovers and returns to a state of 5 ms or less, the speech encoding method is returned to the G.729 method with a small amount of information to be transmitted.
The image data encoded by the image encoding means 611A is input to the image packet creation unit 612. The image packet creation unit 612 adds the packet header shown in FIG. Is output. The image packet output from the image packet creation unit 612 is buffered by the image packet transmission buffer 613, and the image packet packetized at the earliest time among the image packets existing in the buffer is output, and the image packet transmission unit 614 is sent to the network.

  In the voice packet receiving apparatus 200 shown in FIG. 7, either one of the decoding means 205A or 205B is selected and decoded in accordance with the coded identification signal sent by being added to the voice packet in the same manner as described with reference to FIG. Operates and plays audio signals. At the same time, the image packet receiving apparatus 700 receives the image packets addressed to itself from the network by the image packet receiving unit 701, rearranges them in the order of transmission by the image packet receiving buffer 702, and converts the encoded image data by the image encoding / separating unit 703. And the packet header are separated, and the image encoded string data is sent to the image decoding unit 704 to decode the image signal.

According to the operation of the image packet transmitting apparatus 600 described with reference to FIG. 6, the image packet receiving apparatus 700 has a situation in which the reception of the image packet is temporarily interrupted. By maintaining the display of the image signal, interruption of the reproduced image can be avoided.
The voice packet transmission program and the voice packet reception program of the present invention are each described in a computer-readable program language and recorded on a computer-readable recording medium such as a magnetic disk or a CD-ROM. It is installed in a computer from a recording medium or through a communication line. A program installed in the computer is decoded by a CPU provided in the computer, and the computer is caused to function as the voice packet transmitting device and the voice packet receiving device described with reference to FIG. 1, FIG. 6, FIG. 2, or FIG.

  The voice coding selection control method and voice packet transmitting apparatus and voice packet receiving apparatus operating using this control method according to the present invention are used in the field of communication conference systems.

The block diagram for demonstrating the 1st Example of the voice packet transmitter by this invention. The block diagram for demonstrating the 1st Example of the voice packet receiver by this invention. The figure for demonstrating the data structure of the audio | voice packet demonstrated in FIG. The figure for demonstrating the characteristic of the encoding system used for the Example demonstrated in FIG. The figure for demonstrating the control rule by the encoding identification signal used for the Example demonstrated in FIG. The block diagram for demonstrating the 2nd Example of the audio | voice packet transmission apparatus of this invention. The block diagram for demonstrating the 2nd Example of the voice packet receiver of this invention. The figure for demonstrating the image coding control signal used for the Example shown in FIG.

Explanation of symbols

100 voice packet transmitter 600 image packet transmitter
101 timer 604 image coding selection unit
102 encoding unit 610 encoding operation control unit 102A, 102B encoding unit 612 image packet creation unit
103 Coding time measurement unit 613 Image packet transmission buffer
104 Voice encoding method selection unit 614 Image packet transmission unit
105 Coding method selection unit 700 Image packet receiver
106 Voice packet creation unit 701 Image packet reception unit
107 Voice packet transmission buffer
108 Voice packet transmission unit 702 Image packet reception buffer
200 Voice Packet Receiver 703 Image Coding Separator
201 voice packet receiving unit 704 image decoding unit
202 Voice packet reception buffer
204 Decryption selection unit
205 Decryption unit

Claims (3)

The length of the digital code string obtained by digitizing the speech signal is set to the frame length, the discrete speech tone sample is input, the discrete speech tone signal sample is encoded, and the encoded digital code is packetized and transmitted. A voice coding selection control method for dynamically switching coding processing with different amount of computation required for coding discrete voice musical tone signal samples every predetermined time , in frame length units ,
Measuring the time for speech encoding a speech interval for each predetermined time;
A step of performing a coding process with a small amount of calculation when the time required for the speech coding is longer than a predetermined time;
A speech coding selection control method comprising: 2. The speech coding selection control method according to claim 1, further comprising a step of performing coding processing with a small amount of information to be transmitted when the time required for speech coding is shorter than a predetermined time. Speech coding selection control method. 3. The speech coding selection control method according to claim 1, wherein the speech coding processing with a small amount of computation is performed simultaneously when the computation time is longer than a certain threshold value. A speech coding selection control method comprising a step of performing control to temporarily stop an image coding process being performed.

Images