JPH07212320A - Voice band signal packeting device - Google Patents

Abstract

PURPOSE:To realize sufficient transmission quality regardless of kinds of voice band signals. CONSTITUTION:After the kind of an inputted voice band signal is detected 106 and the voice band signal inputted by the optimum encoding system according to the kind is encoded 102 to 104, a packet is assembled and it is transmitted 105, on the side of a transmission. On the side of a reception, a received packet is decomposed, it is converted 111 into an encoding voice band signal, the kind of the voice band signal is discriminated 112 at the time, the signal is decoded in accordance with the encoding system according to the discriminated kind and the voice band signal is reproduced 108 to 110. Thus, the encoding system according to the kind of the inputted voice band signal can be applied and sufficient transmission quality can be attained regardless of the kind of the voice band signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention converts a voice band signal given from an analog voice terminal into a packet, or converts a packet given from a packet switching network into a voice band signal and gives it to an analog voice terminal. The present invention relates to a voice band signal packetizer.

[0002]

[Prior art]

Reference: "1987 Institute of Electronics, Information and Communication Engineers Information Network Study Group," Study on Discarded Packet Compensation in Voice Packet Communication Using High Efficiency Coding Method ", NTT Transmission Systems Laboratories, Junji Suzuki, Masahiro Takataka" In the voice band signal packetizing device as a circuit, as described in the above-mentioned document, a voice signal from a telephone or a voice band data signal other than voice from a facsimile terminal or a modem device (hereinafter referred to as VBD (Voice B
and data) signal), and an analog input voice band signal (voice signal and VBD signal are collectively referred to as such) is transmitted by a predetermined encoding method (for example, CC
ITT Recommendation G. 727 32 kb / s embedded AD
After assembling the packet after the encoding according to the PCM encoding system) or disassembling the received packet, the packet is decoded according to the predetermined encoding system and output to the telephone, the facsimile terminal or the modem device.

[0003]

However, in the conventional voice band signal packetizing apparatus, whether the analog voice terminal for transmitting and receiving the voice band signal via the analog line is a telephone set, a facsimile terminal or a modem unit. In other words, regardless of whether the voice band signal passing through the analog line is a voice signal or a VBD signal, since the encoding or decoding is performed according to the same encoding method determined in consideration of the packet switching network to be applied, Some voice band signals have insufficient transmission quality. Recently, a voice band signal (modem signal) from a facsimile terminal or a modem device tends to have a higher speed, and in particular, it has been difficult to provide sufficient communication guarantee for these signals.

When the coding bit rate is increased in order to solve such inconvenience, the amount of packets flowing through the packet switching network increases, and the fear of congestion increases.

[0005]

In order to solve such a problem, as a transmitting side configuration, a voice band signal type detecting means for detecting which kind of input voice band signal is two or more kinds, and a voice band A plurality of encoding means prepared for each type of signal, a transmitting side selecting means for selecting one of the encoding means according to the detection output of the voice band signal type detecting means, and the selected encoding means are input. Packetized means for assembling the coded voice band signal obtained by coding the voice band signal into a packet and sending it to the packet switching network.

Further, as a receiving side configuration, depacketizing means for decomposing a packet from the packet switching network and converting it into an encoded voice band signal, and a voice band signal for identifying the type of the voice band signal related to the received packet Identification means, a plurality of decoding means prepared for each type of voice band signal,
Based on the discrimination result of the voice band signal type discriminating means, a decoding means is selected, and a voice side signal obtained by the decoding means decoding the encoded voice band signal is outputted to the analog voice terminal. And means.

[0007]

In the transmitting side, the type of the input voice band signal is detected, the input voice band signal is encoded by the optimum encoding method according to the type, and then the packet is assembled and transmitted. On the receiving side, the received packet is decomposed and converted into a coded voice band signal, at the same time, the type of voice band signal is identified, and the voice band signal is decoded by decoding according to the coding method according to the identified type. Reproduce.

In this way, it is possible to apply a coding method corresponding to the type of the input voice band signal, and it is possible to achieve sufficient transmission quality regardless of the type of the voice band signal.

[0009]

【Example】

(A) First Embodiment A first embodiment in which the present invention is applied to a voice band signal packetizer for an ATM (asynchronous transfer mode) network will be described in detail below with reference to the drawings. Here, FIG. 1 shows the voice band signal packetizing device (voice band signal cellizing device) 100 of the first embodiment from the viewpoint of processing voice band signals. In this embodiment, the packet and the ATM network are taken into consideration to call the cell.

A voice band signal cellizing device 100 functioning as a line circuit accommodates an analog voice terminal 1 for outputting a voice band signal, such as a telephone, a facsimile terminal, a modem device, etc., via an analog line, and an analog line. A voice band signal given via the cell is converted into a cell and transmitted to the ATM network 2, and conversely, a cell given from the ATM network 2 is received, converted into a voice band signal and output to the analog voice terminal 1. To do.

(A-1) Transmission Configuration of First Embodiment First, the transmission configuration and operation of the first embodiment will be described.

In FIG. 1, a voice band signal (voice signal or VBD signal) transmitted from the analog voice terminal 1 through an analog line is converted by a hybrid circuit 101 into a 2-line conversion and a A-D conversion ( It is given to the analog / digital conversion unit 102 and the voice / VBD detection unit 106.

The A / D conversion unit 102 here performs analog / digital conversion and supplies a digital signal (coded voice band signal) to the changeover switch 103. The change-over switch 103 has a one-input, two-output configuration, one output terminal is connected to the cell conversion unit 105, and the other output terminal is connected to the high efficiency encoding unit 104. The changeover switch 103 selects an output terminal according to the detection signal of the audio / VBD detection unit 106.

A voice / VBD detection section (voice band signal type detection means) 106 detects whether the voice band signal given through the hybrid circuit 101 is a voice signal or a VBD signal, and switches the detection signal. Switch 103,
Celling unit 104 and CAC (Connection Admission Cont)
rol: Connection admission control) output to the unit 107. It should be noted that in a state where communication is not performed (default state) and the input voice band signal is a voice signal, the voice / VBD detection unit 106 switches the changeover switch 103.
Is connected to the high-efficiency encoding unit 104, while if the input voice band signal is a VBD signal, the voice / V
The BD detection unit 106 switches the changeover switch 103 to the cell conversion unit 1.
It is designed to be connected to 05.

The high-efficiency encoding unit 104 further encodes the input encoded voice band signal, that is, the encoded voice band signal obtained by converting the voice signal by the A / D conversion unit 102, and further encodes it into more bits. Reduce the rate.

As described above, when the analog voice terminal 1 is outputting a voice signal, a low bit rate encoded voice signal is transferred from the high efficiency encoding unit 104 to the cell conversion unit 10.
5 and the analog voice terminal 1 is outputting a VBD signal, a coded voice signal having a high bit rate is input to the cell conversion section 105.

Therefore, in the case of the first embodiment, for the audio signal, the A / D converter 102 and the high efficiency encoder 1 are used.
04 constitutes the encoding means, and the A / D converter 102 constitutes the encoding means for the VBD signal.

The cell conversion unit 105 is a voice / VBD detection unit 1
Based on the detection signal from 06, it is possible to recognize which type of voice band signal the encoded voice band signal is input to. When the detection signal indicates a voice signal, that is, when a coded voice signal having a low bit rate is input, the cell conversion unit 105 encodes in a buffer included in the cell based on a clock signal corresponding to the speed. The encoded voice band signal is taken in to assemble cells according to the format defined for the ATM network 2 and send it to the ATM network 2. On the other hand, when the detection signal indicates the VBD signal, that is, when a coded voice signal having a high bit rate is input, the cell conversion unit 105 incorporates a buffer based on the clock signal corresponding to the speed. Then, the encoded voice band signal is taken in, the cells according to the same format as the above are assembled and transmitted to the ATM network 2.

At the time of assembling the cell, the cell assembling unit 105, as shown in FIG. 2, sends a voice band signal for transmission to any position in the adaptation area of the cell (in the control area in FIG. 2). A voice / VBD identifier 200 for identifying whether it is a voice signal or a VBD signal is inserted. Although not shown in the figure, the header information including the destination information and the like is input to the cell assembling unit 105 in a system different from the voice band signal processing system. Is included.

The CAC unit 107 is commonly used by a plurality of voice band signal cell conversion devices 100, for example. The CAC unit 107 includes a voice band signal cell conversion device 1
At the time of call connection of 00, the cell flow rate per predetermined time is reported to the ATM network 2. When the detection signal from the voice / VBD detection unit 106 indicates a voice signal, the A / D
When the cell flow rate is declared according to the bit rate of the encoded voice band signal from the conversion unit 102 and the detection signal indicates the VBD signal, the bit rate of the encoded voice band signal from the high efficiency encoding unit 104. Declare the cell flow rate according to. In terms of this function, the CAC unit 107 constitutes a packet flow rate reporting means.

As described above, the voice band signal input by the different encoding method according to the type of the voice band signal is encoded, and the encoded voice band signal is assembled into a cell of a predetermined format.
It can be output to the TM network 2.

Various PCM encoders can be applied to the A / D converter 102. For example, CCITT recommendation G.G. 64 kb / s P according to 711
A CM encoder can be applied.

FIG. 3 shows CCITT Recommendation G.264. 7 shows an example of a 64 kb / s PCM encoder according to 711, which is a single channel successive approximation type PCM encoder.

In FIG. 3, the band of the input voice band signal (analog signal) is sequentially limited to the voice band through the low-pass filter 300 and the high-pass filter 301, and in the high-pass filter 301, a clock generator (not shown) is provided. Sampling and holding is performed based on the sampling clock from and the PAM signal is converted.

Such a PAM signal is transmitted to the capacitive array D /
A converter 302a and resistor string D / A converter 30
2b to the CR companding D / A converter 302. To the CR companding D / A converter 302, the digital signal that is sequentially updated is input from the successive approximation register 303. In the CR companding D / A converter 302, the capacitance array D / A converter 302a takes charge of the quantization step of the segment (broken line) in the conversion curve, and the quantization step in the segment is performed by the resistor string D / A converter. 302
The difference between the input PAM signal and the D / A converted analog signal is given to the comparator 304, and the comparator 30
4 changes the digital signal in the successive approximation register 303 according to the difference information. By repeating such a process, the digital signal is determined from the upper side, and the finally obtained digital signal is serially output from the output register 305. The comparator 304 gives an offset signal according to the difference information to the offset compensation circuit 306,
The offset compensation circuit 306 changes the offset of the PAM signal, so that the resistor string D / A converter 3
The small quantization step carried by 02b can be obtained accurately.

As described above, CCITT Recommendation G. 7
The μ-law 64 kb / s PCM signal (encoded voice band signal) defined by 11 is obtained and output.

As the high-efficiency encoding unit 104, those conforming to various encoding systems can be applied. For example, CCITT recommendation G. 32kb / according to 727
An embedded ADPCM encoder can be applied.

FIG. 4 shows CCITT Recommendation G.264. 7 shows an example of a 32 kb / s embedded ADPCM encoder according to 727.

In FIG. 4, the A / D converter (CCITT
Recommendation G. A 64 μb / s PCM encoder according to 711) outputs a signal, and inputs the μ-law 64 kb / s PCM signal via the change-over switch 103.
By the CM conversion unit 310, P having a uniform quantization step is used.
It is converted into a CM signal (linear PCM signal). Subtractor 31
1, the linear PCM signal and a prediction signal for the linear PCM signal formed as described later are input, and the difference signals thereof are obtained and input to the adaptive quantizer 312, and adaptively selected. It is quantized by the generated quantization step and the like, converted into a 32 kb / s ADPCM signal (coded voice band signal), and output to the cell conversion section 105.

The ADPCM signal is transferred to the bit mask section 313.
Is also applied to replace the low-order few bits (enhanced bits) with "0", and then the adaptive inverse quantizer 314 dequantizes the data and converts it into a quantized differential signal for local reproduction. This quantized difference signal is given to the adaptive predictor 315 and the adder 316. A prediction signal is also given to the adder 316, and the prediction signal is added to the quantized difference signal to obtain a local reproduction signal, which is given to the adaptive predictor 315.
The adaptive predictor 315 adaptively forms a predicted signal at the next timing based on the quantized difference signal and the local reproduction signal, and supplies it to the subtractor 311 and the adder 316.

The μ-law 64k input as described above
b / s PCM signal is CCITT Recommendation G.264. It is converted into a 32 kb / s ADPCM signal (encoded voice band signal) specified by 727 and output.

As the voice / VBD unit 106, any one of a neural network system, a band filter system and a prediction filter system may be applied. 5 and 6
Each shows an example of the voice / VBD unit 106. Note that both of the voice / VBD detection units 106
The VBD signal is premised on that the frequency and power are stable as compared with the voice signal, and the determination is made after the VBD signal is once converted into a parameter for detecting such a feature.

First, the voice / VBD detector 1 shown in FIG.
The detailed configuration of 06 will be described. In FIG. 5, the input voice band signal is input to the autocorrelation calculation unit 320, and a plurality of autocorrelation coefficients (which are normalized by power) having different separation times in the correlation calculation are obtained. It is input to the neural network unit 321. The neural network unit 321 previously detects the voice signal and the VBD.
The weighting factors such as synapse weighting are learned so that the signals are input and the outputs take values that can identify them.
Therefore, in the detection operation, the neural network unit outputs an output signal that processes the autocorrelation coefficient group related to the currently input voice band signal and can identify whether the voice band signal is a voice signal or a VBD signal. 321 is output to the determination unit 322. For example, when the number of neurons at the final stage of the neural network unit 321 is one, the determination unit 322 compares the value of the output signal with a threshold value, for example, and detects whether the voice band signal is a voice signal or a VBD signal. Output a signal.

The voice detection unit 323 detects whether or not a voice band signal is input. The output of the voice detection unit 323 is given to the determination unit 322. The determination unit 322 makes a valid determination in the input state. Do. Further, in the case of this embodiment, when there is no input, the determination unit 322 gives the changeover switch 103 a detection signal instructing a voice signal.

Next, the voice / VBD detector 10 shown in FIG.
The detailed configuration of 6 will be described. In FIG. 6, the input voice band signal has a zero-crossing calculation section 330 and a power calculation section 33.
1 and the voice detection unit 332. Zero-crossing calculator 3
The determination unit 3 obtains the number of zero crossings in a predetermined time.
33, and the power calculation unit 331 calculates the average power in a predetermined time and supplies the average power to the determination unit 333. The determination unit 333
The minimum and maximum number of zero-crossings are compared with a threshold value, the difference between the minimum and maximum number of zero-crossings is compared with a threshold value, and the average power fluctuation is compared with a threshold value. The result is comprehensively determined to determine whether it is a voice signal or a VBD signal. The detection signal thus determined is given to the hangover addition unit 334. The hangover addition unit 334 may make an incorrect determination when switching between VBD signals, so when the detection signal changes from the VBD signal instructed state to the voice signal instructed state, a predetermined time elapses from the change point. Also indicates the voice signal starting with the detection signal output when the voice signal is in the instruction state.

The voice detection section 332 detects whether a voice band signal is input (voiced) or not (voiceless), and the detected signal is used as the zero-crossing calculation section 33.
0, the power calculation unit 331, the determination unit 333, and the hangover addition unit 334 to effectively operate these units when being input. Note that the hangover unit 334 gives a detection signal indicating a voice signal to the changeover switch 103 when there is no input.

(A-2) Reception Configuration of First Embodiment Next, the reception configuration and operation of the first embodiment will be described with reference to FIG.

In FIG. 1, the cells transmitted from the ATM network 2 are given to the decellizing unit 111, and the decellizing unit 111 sequentially stores the received cells in a buffer having a built-in cell. A voice / VBD identifying unit 112 is provided in association with the decelling unit 111, and the voice / VBD identifying unit 112 is provided.
Is a voice / VBD inserted in a predetermined position of the receiving cell
The identifier 200 (see FIG. 2) is taken out to identify whether the receiving cell is associated with a voice signal or a VBD signal, and gives a voice / VBD identification signal to the decellizing unit 111 and the changeover switch 110.

The decellizing unit 111 disassembles the cells according to the control information inserted in the control area of the reception cell, and makes the encoded voice band signal inserted in the user information area of each reception cell continuous. Output to the changeover switch 110. When disassembling the received cells and assembling and outputting the encoded voice band signal, the decellizing unit 111 encodes at a speed according to the voice signal when the voice / VBD identification signal indicates the voice signal. The voice band signal is output, and when the voice / VBD identification signal indicates the VBD signal, the encoded voice band signal is output at a speed according to the VBD signal.

The changeover switch 110 has a one-input / two-output configuration, one output terminal is connected to the D / A conversion unit 108, and the other output terminal is connected to the high-efficiency decoding unit 109. The changeover switch 110 supplies the coded voice band signal from the decelling unit 111 to the high-efficiency decoding unit 109 in a default state where no transmission is performed and when the voice / VBD identification signal indicates a voice signal. When the voice / VBD identification signal indicates the VBD signal, the encoded voice band signal from the decelling unit 111 is given to the D / A conversion unit 108.

The high-efficiency decoding unit 109 has an inverse configuration to the high-efficiency encoding unit 104 in the transmission configuration, decodes the input encoded voice band signal,
The bit rate is increased and given to the D / A conversion unit 108. D
The A / D conversion unit 108 is the A / D conversion unit 1 in the transmission configuration.
The high-efficiency decoding unit 10
9 is applied to the input coded voice band signal regardless of whether it is applied from the changeover switch 110.
Converted (decoded) and converted back to analog voice band signal,
It outputs to the external analog voice terminal 1 via the hybrid circuit 101.

As described above, even if a cell in which a coded voice band signal according to a different coding method is inserted is given from the ATM network 2, it is appropriately decoded according to the type of voice band signal. It is possible to output the obtained voice band signal to the analog voice terminal 1.

As described above, the high-efficiency coding unit 104 can be adapted to various coding systems. Therefore, the high-efficiency decoding unit 109 having the inverse structure can also be adapted to various coding systems. it can.

FIG. 7 shows an example of the high-efficiency decoding unit 109, and the corresponding high-efficiency encoding unit 104 is CCIT.
Recommendation G. 32 kb / s embedded AD according to 727
This is the case of the PCM encoder (see FIG. 4).

In FIG. 7, the 32 kb / s ADPCM signal is input to a feedforward adaptive dequantizer (corresponding to the adaptive quantizer 312 described above) 400, dequantized and converted into a quantized difference signal, and an adder. It is given to 401. The adder 401 is also supplied with a prediction signal formed as described later, and a reproduction signal (linear PCM signal) is obtained by adding the prediction signal to the quantized difference signal. This reproduced signal is converted into a μ-law 64 kb / s PCM signal through the PCM conversion unit 402 and given to the synchronous encoding correction unit 403. The ADPCM signal and the prediction signal are given to the synchronous encoding correction unit 403, and the μ-law 6 from the PCM conversion unit 402 is set so that the encoding distortion due to the continuous processing of the PCM encoding and the ADPCM encoding is not accumulated.
The 4 kb / s PCM signal is corrected and output to the D / A conversion unit 108.

The input 32 kb / s ADPCM
The signal is also given to the bit mask unit 404, and after the lower few bits (enhancement bit) are replaced with “0”,
The feedback adaptive dequantizer 405 dequantizes and converts it into a quantized difference signal. This quantized difference signal is given to the adaptive predictor 407 and the adder 406. A prediction signal is also given to the adder 406, and a reproduction signal is obtained by adding the prediction signal to the quantized difference signal and given to the adaptive predictor 407. The adaptive predictor 407 is
The predicted signal at the next timing is adaptively formed based on the quantized difference signal and the reproduced signal, and the adder 401 and the synchronous encoding correction unit 403 of the feedforward path described above are adaptively formed.
And to the adder 406 of the feedback path.

CCITT input as described above
Recommendation G. 32 kb / s ADPCM defined by 727
Signal (coded voice band signal) is μ-law 64 kb / s PCM
It is converted into a signal and output.

Further, as described above, the A / D converter 102
To the D / A converter 108 having various configurations can be applied to the D / A conversion unit 108 having the reverse configuration.

FIG. 8 shows an example of the D / A converter 108, and the corresponding A / D converter 102 is CCITT Recommendation G.264. This is the case of a 64 kb / s PCM encoder according to 711 (see FIG. 3).

In FIG. 8, the μ-law 64 kb / sP supplied from the changeover switch 110 or the high-efficiency decoding unit 109.
The CM signal is a CR companding D / A converter composed of a capacitor array D / A converter 411a and a resistor string D / A converter 411b.
It is given to the A converter 411. In the CR companding D / A converter 411, the quantization step of the segment (polygonal line) in the conversion curve is calculated by the capacitance array D / A converter 411.
a and the resistance string D / A converter 411b carries out the quantization step in the segment, and the input μ-law 64
Convert the kb / s PCM signal to a voltage signal. Such a voltage signal is sample-held by the sample-hold circuit 412 and converted into a PAM signal having a 100% duty ratio, and this is converted into an analog signal (voice band signal) by the low-pass filter 413 and output.

As described above, μ-law 64 kb / sPC
The M signal (encoded voice band signal) is converted into an analog signal and output.

(A-3) Effect of the First Embodiment According to the first embodiment, when the input voice band signal is the VBD signal, it is encoded into a coded voice band signal having a high bit rate. If the input voice band signal is a voice signal, it is encoded into a coded voice band signal having a low bit rate, the coded voice signal is assembled into cells and transmitted to the ATM network 2, and the ATM network 2 is also used. When the cell is received from 2, the decoding process is switched back to the voice band signal by switching the decoding process according to whether the coded voice band signal inserted in the cell is the voice signal or the VBD signal. Sufficient transmission quality can be realized irrespective of the type of voice band signal to be used.

For example, if the audio signal is 32 kb / s ADPC
Encode M signal and convert VBD signal to μ-law 64 kb / s PC
When encoding the M signal and assembling the cells for transmission, sufficient transmission quality can be realized for both the voice signal and the VBD signal.

In practice, CCITT is recommended G7 as an encoding method for assembling cells for the ATM network 2.
In 27, a 32 kb / s embedded ADPCM coding method is recommended. However, when a VBD signal is coded according to this coding method, the transmission quality is deteriorated due to its high speed for a VBD signal of 9600 b / s or more. . Therefore,
As in the embodiment, the VBD signal can be encoded at a higher bit rate than the voice signal, assembled into cells, and transmitted to achieve sufficient transmission quality.

Incidentally, in order to secure the transmission quality of both the VBD signal and the voice signal, it is possible to apply the same coding method and convert them into a coded voice band signal of a high bit rate together to form cells. Conceivable. However, in this case, the cell flow rate to the ATM network 2 increases and the burden on the ATM network 2 increases. Therefore, the transmission quality of VBD signals is unavoidable,
For a voice signal that can secure a sufficient transmission quality with a coded signal of a low bit rate, such a coding method may be adopted as in the embodiment.

It is conceivable to declare the cell flow rate to the ATM network 2 in the case of a voice signal as well as a value in consideration of the VBD signal. However, for proper operation of the ATM network 2, as in the embodiment. It is preferable to declare an appropriate value for each type of voice band signal.

Further, according to the first embodiment, the default connection destinations of the changeover switches 103 and 110 are the audio signal processing circuits (the high efficiency encoding unit 104 and the high efficiency decoding unit 109). That is, since the audio signal having a statistically large amount of traffic is taken into consideration, the switching is often unnecessary at the start of communication, and the influence of switching noise or the like can be reduced.

Furthermore, according to the first embodiment, the effect that a sufficient transmission quality can be realized regardless of the type of voice band signal used for transmission can be obtained with a simple configuration. That is, for example, even if an audio signal and a VBD signal are encoded without distinction, the A / D conversion unit 10
2 and the high-efficiency coding unit 104 are required, and in the above-described embodiment, it is only necessary to add the changeover switch 103 and the voice / VBD detection unit 106 to this, so that a simple configuration can be realized. Also, regardless of the type of voice band signal,
Since the cell conversion unit 105 and the decellization unit 111 are common, the configuration can be simplified from this point as well.

(B) Other Embodiments FIG. 9 shows a second embodiment of the present invention, in which the same or corresponding parts as in FIG. 1 are designated by the same reference numerals. The second embodiment has the same receiving configuration as that of the first embodiment,
This is different from the first embodiment in that the insertion position of the changeover switch 103 in the transmission configuration is provided on the output side of the high efficiency encoding unit 104. However, the operation is similar to that of the first embodiment, and the same effect as that of the first embodiment can be obtained.

In addition to the above-described first and second embodiments, the present invention allows various modified embodiments. Some examples are as follows.

(1) In the first and second embodiments, the type of the voice band signal for switching the coding method is shown as the voice signal and the VBD signal, but other types may be used. . For example, the audio signal and the VBD signal of less than 9600 b / s and the VBD signal of 9600 b / s or more may be divided into two. In the configuration shown in FIG. 5, the type can be detected by learning the neural network unit 321 according to such type classification,
Further, in the configuration shown in FIG. 6, each type can be detected by appropriately selecting the threshold value of the determination unit 333.

(2) In each of the first and second embodiments, the voice band signal is classified into two types, but it may be classified into three or more types and the coding system may be switched. For example, audio signal and VBD less than 9600b / s
The signal and the VBD signal of 9600 b / s or more may be classified into three. In this case, for example,
VBD of less than 9600 b / s by encoding a voice signal into a 16 kb / s encoded signal (for example, applying the CELP method)
Encode the signal into a 32 kb / s encoded signal,
A cell may be assembled by encoding into a VBD signal of b / s or more and an encoded signal of 64 kb / s.

(3) In the first and second embodiments, two types of encoding methods are used in some circuits (A / D converter 102 and D / D converter).
Although the A conversion unit 108) is shared, a separate encoding unit and decoding unit may be used for each type. At present, an IC chip for each encoding system already exists, and when using an existing IC chip, the above-mentioned configuration may be adopted.

FIG. 10 shows a conceptual transmission structure to which the modified embodiments of (2) and (3) are applied.
The same or corresponding parts as in FIG. 1 are designated by the same reference numerals. In FIG. 10, a voice band signal type detection unit 500
Detects which of the three types the input voice band signal is, gives a detection signal to the changeover switch 501, and the encoding units 502a and 502 corresponding to the input voice band signal.
The coded voice band signal from b or 502c is selected and given to the cell assembling unit 105, made into cells and outputted to the ATM network 2.

(4) In the first and second embodiments, the packet switching network is the ATM network 2. That is, the voice band signal packet (cell) for the ATM network 2 is shown.
Although the packetizer 100 is shown, the present invention can be applied to a voice band signal packetizer for another packet switching network, for example, a variable length packet switching network. Depending on the target packet switching network, it is not necessary to declare the packet transmission amount to the network.

(5) In the first and second embodiments, CA
Although the C unit 107 is shown as being commonly used by a plurality of voice band signal cellizing devices 100, the CAC unit 107 may be provided for each voice band signal celling device.

(6) In the first and second embodiments, the type of the voice band signal is transmitted from the transmitting side to the receiving side by inserting the identifier 200 into all the packets, but other types are shown. The information may be transmitted by a method. For example, the information may be inserted only in a few packets immediately after the start of transmission or immediately after the type is changed, or a dedicated packet for transmitting only the type of voice band signal may be provided. Depending on the packet switching network, the channel other than the channel for transmitting the voice band signal may be used for transmission.

(7) The plurality of different coding methods depending on the type of voice band signal are not only those which form coded signals of different bit rates but also those which form coded signals of the same bit rate. May be. For example, the quantization conversion curves may have different nonlinearities.

[0069]

As described above, according to the present invention, the coding method is switched according to the type of voice band signal used for packet transmission, so that a sufficient transmission quality is obtained regardless of the type of voice band signal. Can be realized.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of a first embodiment.

FIG. 2 is an explanatory diagram showing a cell format of the first embodiment.

FIG. 3 is a block diagram illustrating a detailed configuration example of an A / D conversion unit 102 according to the first embodiment.

FIG. 4 is a block diagram illustrating a detailed configuration example of a high efficiency coding unit 104 according to the first embodiment.

FIG. 5 is a block diagram showing a detailed configuration example (No. 1) of the voice / VBD detection unit 106 of the first embodiment.

FIG. 6 is a block diagram showing a detailed configuration example (No. 2) of the voice / VBD detection unit 106 of the first embodiment.

FIG. 7 is a block diagram illustrating a detailed configuration example of a high efficiency decoding unit 109 according to the first embodiment.

FIG. 8 is a block diagram illustrating a detailed configuration example of a D / A conversion unit 108 according to the first embodiment.

FIG. 9 is a block diagram showing an overall configuration of a second embodiment.

FIG. 10 is a block diagram showing a transmission configuration of another embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Analog voice terminal, 2 ... ATM network (packet switching network), 100 ... Voice band signal cell-ized device (voice band signal packetized device), 102 ... A / D conversion part, 103 ... Changeover switch (transmission side selection means) ), 104 ... High efficiency coding unit, 105 ... Celling unit (packetizing unit), 106 ... Voice / VBD detecting unit (voice band signal type detecting unit), 10
7 ... CAC section, 108 ... D / A conversion section, 109 ... High efficiency decoding section, 110 ... Changeover switch (reception side selection means),
111 ... Decellization unit (depacketization means), 112
... voice / VBD discriminating unit (voice band signal type discriminating means).

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Masanori Nozaki 1-7-12 Toranomon, Minato-ku, Tokyo Oki Electric Industry Co., Ltd.

Claims (10)

[Claims] 1. A voice band signal packetizer for converting a voice band signal given from an analog voice terminal into a packet and transmitting the packet to a packet switching network. Voice band signal type detecting means for detecting whether or not, a plurality of encoding means prepared for each type of voice band signal, and any one of the encoding means according to the detection output of the voice band signal type detecting means. A transmitting side selecting means for selecting; and a packetizing means for assembling a coded voice band signal obtained by coding the input voice band signal by the selected coding means into a packet and transmitting the packet to the packet switching network. A voice band signal packetizing device characterized by being provided. 2. The method according to claim 1, wherein the packet switching network is provided.
In the voice band signal packetizer for returning the packet sent by the voice band signal packetizer described in 1) to the voice band signal and outputting it to the analog voice terminal, the packet from the packet switching network is decomposed into the encoded voice band. Depacketizing means for converting into a signal, voice band signal identifying means for identifying the type of voice band signal related to the received packet, a plurality of decoding means prepared for each type of voice band signal, and the above voice band Based on the identification result of the signal type identification means,
A voice band, comprising: a receiving side selecting unit for selecting the decoding unit and outputting the voice band signal obtained by the decoding unit decoding the encoded voice band signal to the analog voice terminal. Signal packetizer. 3. A voice band signal given from an analog voice terminal is converted into a packet and transmitted to a packet switching network,
A voice band signal packetizer for converting a packet given from a packet switching network into a voice band signal and outputting the voice band signal to the analog voice terminal. A voice band signal packetizing device having the configuration described in 1. 4. The voice band signal packetizing device according to claim 1 or 3, wherein the packetizing means adds a voice band signal corresponding to a detection output of the voice band signal type detecting means to each assembled packet. A voice band signal packetizer characterized by inserting type information. 5. The voice band signal packetizer according to claim 2 or 3, wherein the voice band signal identifying means determines the voice band signal based on the type information of the voice band signal inserted in the received packet. A voice band signal packetizer characterized by identifying a type. 6. The voice band signal packetizer according to claim 1, wherein each of the coded voice band signals according to the type of voice band signal has a different bit rate. Voice band signal packetizer. 7. The voice band signal packetizer according to claim 1, wherein the voice band signals are classified into voice signals and voice band data signals. Signal packetizer. 8. The voice band signal packetizing device according to claim 7, wherein the encoding method followed by the encoding means and / or the decoding means for the audio signal is μ-law 64 kb / s PCM encoding method and 32 kb / s. s embedded ADPCM encoding method, and the encoding method followed by the encoding means and / or the decoding means for a voice band data signal is μ-law 64 kb / sP
A voice band signal packetizing device characterized by a CM coding system. 9. The voice band signal packetizing device according to claim 1, wherein the packet switching network targeted by the voice band signal packetizing device is an ATM network, packet assembly or packet. A voice band signal packetizing device characterized in that the disassembly is performed according to an ATM network. 10. The voice band signal packetizing device according to claim 9, which has a transmission configuration, and which declares a packet flow rate to the ATM network according to a detection output of the voice band signal type detecting means. A voice band signal packetizing device comprising: