JP2002135322A - Voice coding automatic selector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a voice coding automatic selector that can select a voice coding system optimum to a transmission state of a present network in real time even when a parameter for a transmission characteristics of the communication network is not recorded. SOLUTION: The voice coding automatic selector of this invention is provided with a Packet Detection section 5 that receives a packet of a voice signal via a packet communication network, a Decoder section 8 that decodes a voice output from the received packet, a Loss section 3 that obtains a packet loss rate from the received packet, a Delay section 4 that obtains a delay time, and a management table 6 that selects a voice coding rule on the basis of the packet loss rate and the delay time, and also with an Encoder section 1 that decodes the voice signal by the voice input, a Packet Generation section 2 that assembles the voice signal into packets and transmits the packet to a packet communication network and an Enc Typ section 7 that controls decoding timing of the voice signal from the Encoder section 1. The voice coding automatic selector preferably uses a grade of a voice service for a parameter.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used in a gateway or other relay system for transmitting voice information through a packet communication network, and allows an edge portion of the relay system to select a voice coding rule suitable for the transmission characteristics of the communication network. The present invention relates to an automatic speech coding selector.

[0002]

2. Description of the Related Art Conventionally, there have been known various voice coding systems for decomposing voice into a plurality of packets while coding the voice, and transmitting the packets through a packet communication network. Such a speech coding system is required to be able to transmit telephone and other speech to a remote receiving station while maintaining high transmission quality by using a high-speed coding scheme.

[0003] For example, Japanese Patent Application Laid-Open No. 04-349735 discloses a "speech encoder control system" for transmitting speech information with high quality through an ATM (asynchronous transmission mode) communication network. In this control method, the coding rate, cellization, and the presence / absence of echo cancellation are efficiently selected by a predetermined algorithm according to the transmission mode of the voice signal as it is, low-speed and high-speed modem modes.

[0004]

However, when using the voice coding system according to the above-mentioned conventional example, it is attempted to transmit voice signals while maintaining high quality over a packet communication network comprising a wide area switching network. The following problems have arisen.

[0005] First, when encoding a voice signal using a packet communication network instead of the conventional circuit switching network,
In order to maintain high voice quality, it is necessary to know the transmission parameters of the corresponding line in advance. Such transmission parameters include a packet loss of a packet due to a transmission path and a degree of delay of the packet as to the voice quality, and these parameters are necessary when optimizing the selection of the voice coding method. .

However, the packet communication network has a Q
There are various communication networks for which the OS (Quality of System) is not guaranteed, and when transmitting data through such a communication network, specific parameters regarding the transmission characteristics are often not recorded. For this reason, speech coding cannot always be performed depending on these transmission characteristic parameters. In particular, this is a remarkable problem in a communication network based on IP (Internet Protocol).

Second, even in the case of a communication network in which transmission parameters are specifically recorded, when a user's telephone or the like is busy, the speech coding method for each individual packet is changed as needed. It cannot be changed to something. For this reason, it may not be possible to select an optimal speech coding method for each communication path of a communication network through which each packet is transmitted. As a result, high-quality calls cannot be reliably promised to a wide range of users using the communication network, and it has been an important issue to link and solve these problems.

[0008] Therefore, an object of the present invention is to solve the above-mentioned problems, and to provide a speech coding scheme that is optimal for the current transmission state of a network, even if transmission characteristics parameters are not recorded in the communication network. An object of the present invention is to provide a speech coding automatic selector which can be selected in real time.

[0009]

An automatic speech coding selector according to the present invention receives a speech information packet via a packet communication network, decodes a speech signal from the received packet, and outputs a speech signal. Restoring means, transmission parameter calculating means for calculating parameters in the transmission path of the packet based on the contents of the header information of the packet, and coding scheme selection for selecting one of a plurality of coding schemes based on the calculated parameters. Means for coding a voice input by a coding method selected by a coding method selecting means for a voice input corresponding to a voice output by a voice decoding stage while monitoring a call state based on the content of the voice information. And voice transmission means for transmitting the data via the Internet.

In this automatic speech coding selector, a parameter that affects the speech quality of the transmission characteristics of the packet communication network is defined in advance, and a packet of the communication network is received while calculating the parameter in real time.
Then, from the parameters obtained by the calculation, select a speech coding method suitable for the current transmission characteristics,
Also, by monitoring the state of the call, a packet of voice information that has been subjected to voice coding when not in a call is transmitted to the communication network. As a result, voice information such as a telephone call is relayed while making the voice quality follow the routing by the communication network, and voice coding can always be optimized for the transmission path.

An automatic speech coding selector according to claim 2 is
The packet communication network is configured to be a communication network conforming to the Internet protocol. Thereby, the voice information is relayed in the network layer connected to the Internet.

An automatic speech coding selector according to claim 3 is
The transmission parameter calculation means is a means for calculating a packet loss or a delay of the packet originating from the transmission path of the packet as a parameter. As a result, packet loss due to packet switching and packet delay are defined as parameters.

An automatic speech coding selector according to claim 4 is
The encoding method selecting means is means for searching for and selecting one encoding rule from one or more parameters by using a table in which a plurality of encoding rules are arranged in combination with the values of unique parameters. . This allows
An appropriate speech coding rule is selected by a simple table search process.

An automatic speech coding selector according to claim 5 is
The transmission parameter calculation means is configured to calculate a service level, which is a priority when a voice communication service is provided, as a parameter. Thus, the service level associated with the communication service is defined as a parameter without depending on the transport layer or the upper layer, or in order to support each of these layers.

Preferably, the coding method selecting means always selects a high-priority coding rule for a packet to which a high-priority service level is set, and gives a priority to a packet to which a low-priority service level is set. This is a means for selecting a coding rule having a low degree. Thereby, when traffic of the communication network is deteriorated, the speech coding is adjusted based on the priority of the service level, so that the margin of the frequency band in the transmission path is maintained. Therefore, for a call having a higher priority, the cause of deterioration of voice quality due to the packet is reduced.

[0016]

Preferred embodiments of the present invention will be described below with reference to the accompanying drawings. An example of the first embodiment of the present invention will be described with reference to FIGS. A packet communication network using an example of the automatic speech coding automatic selector shown in FIG. 1 is provided with another automatic speech coding automatic selector (not shown) having the same function as the automatic speech coding automatic selector (hereinafter referred to as “opposite”). , The opposite device) are connected via a communication network,
Each constitutes each node of the present packet communication network. Hereinafter, a case of voice communication conforming to the Internet protocol (VoIP: Voice over IP) will be described as an example.

The automatic speech coding selector includes a PacketDetection unit 5 for receiving a packet of a speech signal via a packet communication network (Packet Network), a Decoder unit 8 for decoding a speech output from the received packet, and A loss section 3 for calculating a packet loss rate, a delay section 4 for calculating a delay time, and a management table 6 for selecting a speech coding rule based on the packet loss rate and the delay time are provided.

Further, an Encoder unit 1 for decoding an audio signal by an audio input, a PacketGeneration unit 2 for organizing the audio signal into packets and transmitting the packets through a packet communication network,
An EncTyp unit 7 for controlling the decoding timing of the audio signal of the encoder 1 is provided, and these units 1 to 8 constitute an automatic audio encoding selector according to the first embodiment. For example, audio output and audio input are from an upper layer audio service.

The PacketDetection unit 5 is a processing unit having a separating function and a discriminating function for a received packet, and separates the header part of the packet and sends it to the Loss unit 3 and the Delay unit 4. In addition, it separates the payload portion and sends it to the Decoder 8 together with the result of determining the audio coding method of the transmitted audio signal.

The Decoder unit 8 is a processing unit which performs audio decoding based on the audio encoding system of the introduced audio signal by a decoding system corresponding to the audio encoding system, and together with an Encoder unit 1 described later, a DSP (Digital Signal Processor). ).

The Loss section 3 is a processing section having a packet loss calculating function for a received packet, and calculates a packet loss rate in units of a predetermined period from the contents of the header section relating to packet loss. The Delay unit 4 is a processing unit also having a delay calculation function, and calculates a packet delay time from the time stamp information in the header. Then, the packet loss rate calculated by the loss unit 3 and the delay time calculated by the delay unit 4 are sent to the management table unit 6.

The management table section 6 is a processing section having a management table in which a plurality of speech coding rules are arranged.
The arrangement of the management table is such that a method capable of providing the highest quality in the communication network can be selected. As a result, the management table is searched using the introduced packet loss rate and delay time as parameters, and the corresponding speech coding rule is selected as the optimum one for the transmission path.

Also, when the transmission characteristics of the communication network are relatively good (when the packet loss is low or the packet has low delay), a speech coding rule with a high bit rate is combined. Thereby, the deterioration of the voice quality can be suppressed. However, when the transmission characteristics are deteriorated (when the packet loss is high or the packet has a high delay), an increase in traffic is estimated, so that a speech coding rule with a low bit rate may be used. In this case, although the packet delay remains, the frequency band in the transmission path can be effectively used.

The EncTyp section 7 is a processing section for monitoring the switching timing of the speech coding rule, and sets the switching timing at which the speech coding rule can be switched without hindering the speech decoding of the encoder 1. Then, when the speech coding rule selected in the management table unit 6 is different from the speech coding rule currently used in the Encoder unit 1, at the set switching timing.
The audio decoding of the encoder 1 is controlled.

The Encoder unit 1 is a processing unit having a function of compressing an audio signal. The Encoder unit 1 performs audio decoding of the audio input according to the selected audio coding rule, compresses the audio signal, and then compresses the audio signal. Send to 2. PacketGenera
The communication unit 2 is a processing unit that organizes a compressed audio signal into packets, adds header information such as a header portion necessary for transmission to the audio signal body, combines the packets into packets, and connects each packet to a packet communication network. Via the other device (another automatic speech coding selector).

An example of the automatic speech coding selection method according to the first embodiment will be described below. In the flowchart of the automatic speech coding selection method shown in FIG. 2, two speech coding automatic selectors having the same function as described above are used,
The respective stages of the automatic selection of the audio coding of the transmission-side processing GW # 0 by one of them and the reception-side processing GW # 1 by the other are shown in time series.

In the automatic speech coding selector of the GW # 0 on the transmitting side, when an input packet arrives at the speech coding automatic selector, it is received by the PacketDetection unit 5 and its header and payload are converted from the packet. Separate (ST0
1). Further, a coding rule check is performed to determine the selected speech coding method and notify the Decoder unit 8 (ST02).

For the header section, the loss section 3 performs a packet loss calculation process to calculate the frame period and count the number of audio signal packets that have arrived. Then, the number of packets to be arrived from the frame period is obtained, and the number of arrived packets is counted to calculate the packet loss rate due to the transmission path at this time for the packets transmitted from the opposite device (ST0).
3).

Further, since the time stamp information is added to the header of the packet, it is possible to detect the transmission time at which the packet was transmitted from the opposite device. Therefore, delay processing is performed by the delay unit 4, the transmission time is subtracted from the packet arrival time, and the delay time due to the communication network is calculated (ST04).

Based on the packet loss rate and the delay time,
The management table section 6 performs management table search processing. The management table stores in advance the speech coding rules that can provide the caller with the highest speech quality from the plurality of speech coding rules described above. When a call by a caller is set up via the communication network, the optimum speech coding rule for the communication network is searched from the management table based on the calculated packet loss rate and delay time (ST05).

The Encoder 1 performs a speech coding rule selection process for speech input. If the searched speech coding rule is different from the currently selected speech coding rule,
It detects whether the current voice of the voice input is voiced or silent, and switches the voice coding rule at the moment when the voice becomes silent (ST06).

In this case, if the voice coding rule is switched when the voice signal to the opposite device has sound, this may cause the talker to feel uncomfortable. Is reduced as much as possible.

As described above, in the automatic speech coding selection method of the transmission-side processing GW # 0, a packet (performing the reception-side processing GW # 1) is repeatedly transmitted to the opposing device via the packet communication network (ST07). ).

On the other hand, in the automatic speech coding selection method of the receiving-side processing GW # 1, the packet is transmitted by using the automatic speech coding selecting device for performing the transmitting-side processing GW # 0 as the opposing device (ST11). As described above, the speech encoding rule of the speech signal is switched in the received packet. Therefore, when packet arrival is detected, this switching is detected from the contents of the header part (ST12), and the transmission side processing G
Control of speech decoding is performed in the same manner as W # 0 (ST13-S
T17).

As described above, by replacing the processing GW # 0 on the transmission side and the processing GW # 1 on the reception side with each other, the switching of the speech coding rule by the two-way communication between the two opposing devices is performed. At this time, it is possible to select an optimal speech coding rule without hindering normal speech traffic.

Next, an example of the second embodiment of the present invention will be described. The automatic speech coding selector according to the second embodiment provides a service level SLA () for setting a grade of service to be provided to a caller for each caller as a parameter of a management table for switching a speech coding rule. Servi
ce Level Agreement).

For example, the content relating to the service level SLA is a service type TOS (Type of Servi
ce) A process that can be separated from the field is provided in the PacketDetection section, and the service level SLA
Is provided in parallel with the SLA unit having the process of calculating. further,
The management table unit may have another management table search process for searching the management table using the service level SLA as a parameter.

With this configuration, when the state of the communication network deteriorates, a call (low quality, high delay) of a low service level SLA is selected from a plurality of calls for which a call is being set up via the communication network. , Or the call with the lower service level SLA. Then, the speech coding rule for the selected call is changed to a speech coding rule with a low bit rate, and the above-described speech decoding is controlled.

Thus, in the voice communication service,
It is possible to define a service level SLA associated with a communication network, and a service level SL
It is possible to provide the caller with the voice quality of the service according to A.

Further, it is possible to allow a margin in an allowable frequency band by the communication network, and as a result, even if the traffic of the communication network deteriorates, a call of a high service level SLA (high quality, low delay) For (2), it is possible to maintain a high-quality call while suppressing the deterioration of the voice quality, and to provide voice quality equivalent to that of an existing circuit-switched network.

[0041]

As described above in detail, the effect of the present invention is to define a parameter that can affect voice quality, select a coding method based on the transmission characteristics of a received packet in real time, , And is transmitted to the packet communication network. For this reason, audio information can be relayed with audio quality that follows the routing of the communication network, and the optimal audio encoding can be transmitted on the transmission path.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an example of an automatic speech coding selector according to a first embodiment and a second embodiment of the present invention.

FIG. 2 is a flowchart of each processing step by the automatic speech encoder selector shown in FIG. 1;

[Explanation of symbols]

 1 Encoder section 2 PacketGeneration section 3 Loss section 4 Delay section 5 PacketDetection section 6 Management table 7 EncTyp section 8 Decoder section

Claims (6)

[Claims] An audio restoration means for receiving a packet of audio information via a packet communication network, decoding an audio signal from the received packet to output an audio signal, and setting a parameter in a transmission route of the packet to a header of the packet. Transmission parameter calculating means for calculating based on the content of the information; coding method selecting means for selecting one of a plurality of coding methods based on the parameters obtained by the calculation; and Voice transmitting means for encoding the voice input according to the coding scheme selected by the coding scheme selecting means and transmitting the voice input through the communication network, for the voice input corresponding to the voice output by the voice decoding stage. Automatic speech coding selector. 2. The automatic speech coding selector according to claim 1, wherein the packet communication network is a communication network conforming to the Internet protocol. 3. The automatic speech coding selector according to claim 1, wherein said transmission parameter calculating means is a means for calculating, as a parameter, a packet loss originating from a packet transmission path or a delay of the packet. . 4. The coding method selection means searches for one coding rule from one or a plurality of parameters by using a table in which a plurality of coding rules are arranged in combination with values of unique parameters. 2. The automatic speech coding selector according to claim 1, wherein said selecting means is a means for selecting by selecting a speech code. 5. The transmission parameter calculation means calculates a service level, which is a priority when a voice communication service is provided, as a parameter.
The automatic speech coding selector according to claim 1. 6. The coding method selection means always selects a high-priority coding rule for a packet to which a high-priority service level is set, and gives a priority to a packet to which a low-priority service level is set. 6. The automatic speech coding selector according to claim 5, wherein said means is a means for selecting a coding rule having a low degree.