JPH0799505A - Voice conference circuit of loop type and its establishing system - Google Patents

Abstract

PURPOSE:To maintain excellent conference speech quality by introducing an attenuation constant in a loop of a loop transmission line so as to eliminate noise in the loop. CONSTITUTION:A reception input signal Rin(t) from a reception channel received from a reception input terminal 30 of a conference circuit, a transmission output signal Sout(t) to a transmission channel outputted from a transmission output terminal 40, a transmission input signal Sin(t) from a concerned terminal equipment received from a transmission input terminal 10, and a concerned terminal equipment reception output signal Rout(t) outputted from a reception output terminal 20 have the relation of satisfying equations I-IV (alpha is an attenuation constant smaller than 1, N is the number of terminal equipments, T is a transmission delay time of an operation state for loop circulation, and (t) is a time variable). An overall loop gain is alphaN when the attenuation constant alpha is introduced in the loop, the loss attenuates noise in coding/decoding repeated in each node to suppress a noise component in the loop.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement of a voice addition method when a plurality of terminals are connected in a loop to hold a multipoint voice conference.

[0002]

2. Description of the Related Art A technique is known in which a plurality of voice terminals (telephones) are connected via a communication network to hold a voice conference. In particular, a technique is also known in which a terminal uses two communication paths (channels) and connects the respective communication paths to different grounds to form a loop (loop) transmission path as a whole to conduct a voice conference.

For example, there is a "conference connection establishing method and apparatus" disclosed in Japanese Patent Laid-Open No. 3-289850. In the above-mentioned publicly known example, a plurality of terminals conducting a conference call are incorporated in a network as a ring node, and voice drop / insert is performed at each node. More specifically, the voice inserted by the own terminal in the past is subtracted from the voice received from the loop to make it the received voice of the own terminal, and the new voice of the own terminal is further added as the transmission signal to the loop. The so-called N-1 addition scheme has been described.

[0004]

The above-described technique has the following practical problems.

First, actually, the transmission / reception signal for the loop transmission line is a high-efficiency coded signal for band compression.
For example, it is a PCM (pulse code modulation) μ-Law code or ADPCM (adaptive differential pulse code modulation). Since such a code cannot be directly added / subtracted, it is necessary to convert it to a linear code before performing the addition / subtraction operation (N-1 addition). At this time, so-called quantization noise is unavoidable. This quantization noise can occur at each node in the loop. Then it goes around the loop.
However, since the loop itself is considered to form a perfect integrator in signal processing, there is a serious problem in that the noise power increases and the original call quality deteriorates.

Second, there is a problem of echo (echo) at each terminal. In general, in each terminal, a part of the reception component may become an echo and be mixed again in the transmission input. This reverberant component circulates in the loop together with the main signal, but when the number N of connected terminals becomes large, this effect cannot be ignored and there is a problem that the call quality deteriorates.

An object of the present invention is to provide a voice conference circuit capable of removing the above-mentioned noise in the loop and further reducing the reverberant component to maintain good conference call quality. Another object is to provide means for establishing such a voice conference circuit configuration.

[0008]

The loop type transmission line having the above-mentioned conventional terminal as a node is itself a perfect integrator of noise. Therefore, the attenuation constant α (α <1) was introduced into the loop to attenuate the noise component. Also, because of the voice conference circuit configuration, voice is input from each node toward the loop, but the loop has a transfer characteristic like an integrating circuit. In order to inversely correct this characteristic, the compensation circuit is provided at the input part to the loop or the output part from the loop. When extracting the sound from the loop, the circuit structure was designed so that it does not subtract the self-voice after one round, but immediately.

Now, the received input signal from the transmission path is Rin.
(T), Sout (t) is a transmission output signal to the transmission path, Sin (t) is a transmission input signal of the own terminal, and Rout (t) is a reception output signal of the own terminal. Is expressed by the following equation. First, if you put the compensation circuit in the input section

[0010]

[Equation 3]

Here, t is a time variable, T is a loop loop transmission line delay time in an operating state, N is the total number of points, and α is an attenuation constant.

Alternatively, a compensation circuit is provided at the output section.

[0013]

[Equation 4]

Further, in order to cancel as much as possible the reverberant components that try to circulate through the loop, the polarities of the Sin (t) signals in (Equation 1) to (Equation 8), for example, even-numbered terminals have opposite phase inputs (- Sin (t) input) and odd-numbered terminals are Sin (t) input.

On the other hand, such a voice conference circuit requires two parameters for establishing the circuit. The delay time T for one loop and the total number N of participants in the conference. In deciding this parameter prior to the transition to the call operation, the device acting as the chairperson has a specific pattern A and another specific pattern B directed to the member devices incorporated in the loop and the order of joining the loop. The specified counter value is sent out.

The initial value of the counter value is zero. Each member device provides a unit delay time between Rin and Sout when incorporating a loop, and passes the above-mentioned specific pattern. When the counter value is received, +1 is incremented and stored as the participation number of itself, and the counter value is sent to the next member. The chairperson apparatus captures the change time point of the pattern A and the pattern B that has made a round, and measures the difference between the change time point at the time of transmission and T, and increments the received count value by +1 to obtain the total number of participants N. Continue to deliver (T, N) values to all members, where each member (T, N)
N) Receive the value to establish the conference circuit, and send the input signal Sin by the even / odd of the participation number stored previously.
The polarity of (t) is controlled. After confirming a round of parameter values, the chairperson apparatus itself is configured to establish a conference circuit in the same manner as the member apparatus and shift to call operation.

[0017]

In the present invention, as shown in (Equation 1) to (Equation 4) or (Equation 5) to (Equation 8), an attenuation multiplier of α is introduced in each terminal. That is, the transmission output is Sout (t) = α · Rin (t) for the reception input component Rin (t). When N terminals are incorporated into the loop, the total loop gain becomes Equation A, and this loss attenuates the repeated coding / decoding noise at each node. Further, the transfer characteristic of one round of the loop is expressed by the formula B. Here, Z to ¹ represents a unit delay, and T is a total delay amount of the loop included in the terminal / transmission / switching measured by the unit delay. The input / output transfer characteristic H (Z) of the adder circuit for externally adding voice to the loop at each node has the following integral type.

[0018]

[Equation 5]

[0019]

[Equation 6]

Therefore, it is necessary to correct the inverse characteristic of the above equation at any place. When the equation C is provided in the transmission input section of each terminal, the circuit (input differential type) shown in (Equation 1) to (Equation 4) is provided, and when it is provided in the reception output section of each terminal, (Equation 5) to (Equation 8) is obtained. It will be the circuit shown (output differential type).

[0021]

[Equation 7]

In any circuit, it is not necessary to return its own input component to itself, so the Sin (t) component is subtracted in the calculation of the Rout (t) component.
In the input differential type circuit, first, a difference from the past input signal is created in advance, and the difference is added to the past component (for the loop) to be the input at that time. When this component goes around the loop, it becomes the old component, and the difference is added again to become the new component. Due to the superposition principle, the voice components of N terminals are synthesized on the loop. In each terminal, only the component of its own terminal is subtracted to obtain the Rout (t) component. On the other hand, in the output differential type circuit, the integrated voice circulates on the loop (for N persons). The output section differentiates it and subtracts its own voice.

Next, the canceling action of the echo component will be described. Now, in a certain terminal, the reflection coefficient from the reception output Rout (t) to the transmission input Sin (t) is r ₁ . That is, the unwanted bond Sin (t) = r ₁ · Rout
Suppose there is (t). At this time, if (Equation 1) to (Equation 4) are analyzed, the transmission output Sout from the reception input Rin (t)
The following equation is obtained as the transfer function to (t).

[0024]

[Equation 8]

When these are connected in two stages, that is, when the two points are combined, r ₂ is the reflection coefficient of the second point.

[0026]

[Equation 9]

If r ₁ r ₂ << 1, r ₁ + r ₂ = 0, then T ₁ (Z) T ₂ (Z) ≈α ² . That is, if the phase relationship between the reception output amplifiers or the transmission input amplifiers is set to be opposite in the terminals adjacent to each other, the reverberant components of the same reflection environment are canceled and the loop circulation can be prevented. As can be seen from the above description, the original voice signal other than noise is also attenuated by α, α ² , α ³ , α, ... Each time it passes through the node due to the attenuation constant α. However, since α is generally set to a constant close to 1 and smaller than 1, the effect can be ignored in practice.

The call status circuit described above is established by the initial procedure as described in the above [Means], and shifts to its operation mode.

[0029]

Embodiments of the present invention will be described with reference to the drawings. First, FIG. 3 illustrates a loop-shaped transmission line. , N are the communication channels CHi and CH, respectively.
They are connected in a ring shape via j. An insertion signal Sin (t) is added to each of the conference circuits from the outside, and the reception signal from the loop is dropped out to obtain Rout (t).
In this way, N persons (N terminals) can talk freely (conference)
Can be executed.

FIG. 1 illustrates details of the conference circuit in each terminal. Reference numerals 10, 20, 30, 40 denote signal Sin (t) insertion terminals from terminals, output terminals for dropping signals (Rout (t)) to terminals, terminals for receiving signals Rin (t) from channels, and channels. It is a signal Sout (t) transmission terminal to. Reference numeral 11 denotes a circuit for converting the insertion signal Sin (t) into a linear code. Specifically, when Sin (t) is the output signal of the μ-Law encoder, it is a μ-Law → linear conversion circuit or the like. is there. 21
Is its inverse conversion circuit. Reference numeral 41 is an encoder for highly efficient encoding of a linear code, and may be, for example, a μ-law encoder or an ADPCM encoder. Reference numeral 31 is a circuit for converting a received signal from the loop transmission line into a linear code, and is specifically a μ-Law decoder, an ADPCM decoder or the like. Reference numeral 100 is a subtractor for subtracting the signal Sin (t−T) sent by the terminal in the past from the Sin (t) signal. Reference numeral 200 denotes a multiplier for adjusting the amplitude of Sin (t−T), and the multiplication coefficient is given by Expression D according to the total number N of terminals arranged on the loop. An adder 400 superimposes the differential speech of the terminal itself on the loop circulation signal. 5
Reference numeral 00 denotes a multiplier newly introduced to attenuate cyclic quantization noise and the like, and the multiplication coefficient α is a value smaller than 1 which is close to 1. 300 indicates Rout (t) from the loop signal
It is a subtractor for canceling the inserted signal when extracting the signal.

[0031]

[Equation 10]

Reference numeral 600 is a delay circuit in which a value equal to the signal group delay time T of one loop is set.

The configuration of FIG. 1 has the above equation (Equation 1),
It is obvious that (Equation 2), (Equation 3) and (Equation 4) are realized. Quantization noise, etc., which is introduced at the time of mutual conversion between a linear code and a transmission path code, is added moment by moment, but is attenuated each time a node passes, so that it does not accumulate in the loop.
The original audio signal component is also 20Logα (d
It suffers from the attenuation of b), but its amount is small, and there is no problem in communication. In this way, good call quality is secured in the loop type transmission line.

Next, as another embodiment, (Equation 5) to (Equation 8)
FIG. 2 shows a mode in which the above is realized. Although it may not require much explanation, this configuration is a method of differentiating the integrated signal on the loop at the output section. Descriptions of reference numerals and the like are the same as those in FIG.

Next, referring again to FIG. 3, an embodiment in which the reverberant components are canceled out will be described. In the embodiment of FIG. 3, the transmission input formula E applied at each terminal is input through the inverting amplifier at the even-numbered terminals. As a result, it is possible to prevent the echo signal from looping around as described in the section of the action.

[0036]

[Equation 11]

This inverting amplifier can actually be easily realized by the sign inverting process in the input sign conversion unit 11.

Next, the operation of establishing the conference circuit will be described. Figure 1
Alternatively, in order to establish the call conference circuit of FIG. 2, it is necessary to measure the transmission delay time T around the loop and the number N of participants in the conference. The values of the parameters (T, N) shall be measured by the chairman apparatus as the organizer at the start of the conference and delivered to the participating member devices. FIG. 4 is a time chart for explaining this operation. Currently, the network that forms the loop is ISDN (Integrated Service Digital Network).
rk) and each terminal has two communication paths (B channel 2
It is connected to the ring via a book. Transmission synchronization (bit synchronization and octet synchronization) is established, and the audio conference device can transmit and receive octet data. After the loop is established, the transmission unit (SND) of the chairperson apparatus starts transmitting a certain octet pattern A. As the pattern A, for example, a callless pattern of a transmission line code can be adopted. After sending this for a while, it is switched to another pattern B and a timer is started to measure the delay time. The pattern B has the largest code distance with respect to the pattern A. This is to maximize the resistance to transmission errors. After transmitting the pattern B for several words, the code of the value 0 is continuously transmitted as the conference participation order counter for several words, and then returned to the pattern A. The above data is transmitted via the transmission line to the first participating member M1.
Reach The member device M1 sends the received octet pattern to the next member with a unit delay time (Z ~ ¹ = 125 μs). In addition, the time when the pattern A and the pattern B change is searched for while protecting the data error of the pattern A and the pattern B. When the change time point is detected, counting from there, a few words later are recognized as the sequential counter value. The counter value is incremented by +1 and stored in its own memory as its participation order and transferred to the next member. Each member repeats this operation. After one round of the loop, the data is returned to the receiving unit (REC) of the chairperson apparatus M0. R
In the EC, similar to Mi (i = 1 to N-1), the change point is searched while protecting the reception errors of the pattern A and the pattern B. When the change point is detected, the timer is stopped and the value of T is obtained. The value after several consecutive words is recognized as the value of the participation order counter, and is incremented by +1 in the same manner as each member to obtain the total number N of participants. At this point, the transmitter stops sending the pattern A, which has been continuously output, sends the pattern B for several words, sends the measured values T and N a plurality of times, and returns them to the pattern A again. This data is delivered to each terminal. Each terminal detects the change point to the pattern B as before, recognizes the data a few words after that point as the (T, N) value, and after error-protecting this, the T value, the expression F value,
Set to and switch to call mode.

[0039]

[Equation 12]

Of course, the polarity control state of the transmission input signal is also set. Each terminal shifts to the call mode one after another. The above data makes one round in the loop and finally reaches the receiver of the chairperson M0. The chairperson apparatus confirms that the (T, N) value that has made one round matches the transmitted (T, N) value and shifts to the call mode. The loop conference call will start from here. The time until the above circuit is established depends on the transmission line delay time, and is determined by the time of about 2T. It can be said that this is a sufficiently short time for practical use. For example, assume that the average transmission time between terminals is 20 milliseconds. If a ring is formed at N = 5 points, 2T = 2 × 5 × 20 = 20
The conference circuit is established in 0 milliseconds. In general, this order value can be ignored compared to the time of switching connection, that is, the time until the loop is established. In the above, the value of the measured value T of the delay time of the transmission line is used as it is, and the delay circuit (600) type G is used.
And the T value. This is the encoder ENC (4
This is a case where the processing delay of 1) and the decoding device DEC (31) is unit time Z ~ ¹ . Needless to say, when the group delay time of the signal associated with the encoding / decoding is large, this group delay time needs to be further corrected with respect to the measured value T.

[0041]

[Equation 13]

The circuits of the present invention implemented as described above are specifically realized by using a signal processor (DSP). FIG. 5 shows the hardware configuration of the conference device configured using the DSP. Transmitter / receiver (microphone / speaker)
The analog voice signal from the PCM codec is converted into a digital signal by Sin (t), Rout.
It is supplied to the DSP as a (t) signal. The DSP has a memory (ROM, RAM) outside thereof, and executes the above-mentioned circuit establishing operation and call circuit operation under program control. The program is stored in the ROM. The transmission / reception signals Sout (t) and Rin (t) for the communication path are connected to the network via the transmission interface circuit. In this embodiment, the network is a so-called ISDN and is connected in a ring shape via two B channels. I
The SDN interface circuit and the DSP are entirely controlled by the host controller MPU. Since the communication path is normally capable of bidirectional transmission, it is possible to obtain another ring path that circulates in the opposite direction to the voice. Other media such as audio signals or image signals can be passed through this path.

[0043]

As described above, according to the present invention, it is possible to remove the noise component which is introduced into the loop and which circulates in the loop by introducing the loop damping constant. The loop noise specifically includes quantization noise introduced by the high-efficiency encoder, equivalent noise caused by a transmission path error, overflow noise caused by intra-node addition, and the like. Noise circulating in the loop can also destabilize the loop and cause loop oscillations. Since this can be suppressed to enable a good N-party conference, the effect of the present invention is great.

Further, the loop circulation of the echo component due to the external space reverberation is also eliminated, and a good loop type N-party conference can be realized.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a conference circuit connected to a loop transmission line according to the present invention.

FIG. 2 is a diagram showing another embodiment that replaces FIG.

FIG. 3 is an explanatory diagram of an overall situation in which a conference circuit is connected in a loop and an N-party conference is held.

FIG. 4 is an explanatory diagram showing a situation in which a loop parameter is measured when a conference circuit is established.

FIG. 5 is a configuration diagram of specific hardware of the conference circuit device according to the present invention.

[Explanation of symbols]

1, 2, ..., N ... Audio conference circuit, 10 ... Transmission input terminal,
20 ... Reception output terminal, 30 ... Reception input terminal, 40 ... Transmission output terminal 11,21 ... Circuit for mutually converting input / output signals into linear code, 31,41 ... Circuit for mutual conversion between linear code and high efficiency code ( Transcoder), 100, 300, 4
00 ... Adder / subtractor, 200, 500 ... Multiplier, 600 ... Delay circuit.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Tsutomu Irishima 1-1-6 Uchisaiwaicho, Chiyoda-ku, Tokyo Nippon Telegraph and Telephone Corporation

Claims (4)

[Claims] 1. A plurality of (N) connected to a loop type transmission line.
1 is a circuit configuration of a voice-type terminal that allows free conversation between individual terminals, and a reception input signal Ri from a reception channel
n (t), output signal Sout sent to the transmission channel
Between (t), the transmission input signal Sin (t) from the own terminal, and the reception output signal Rout (t) from the own terminal, Here, α is an attenuation constant smaller than 1, N is the number of terminals included in the loop type transmission line, T is an operation state transmission delay time of one round of the loop, and t is a time variable. Characteristic loop type audio conference circuit. 2. A plurality of (N) connected to a loop type transmission line.
1 is a circuit configuration of a voice-type terminal that allows free conversation between individual terminals, and a reception input signal Ri from a reception channel
n (t), output signal Sout sent to the transmission channel
Between (t), the transmission input signal Sin (t) from the own terminal, and the reception output signal Rout (t) from the own terminal, Here, α is an attenuation constant smaller than 1, N is the number of terminals included in the loop type transmission line, T is an operation state transmission delay time of one round of the loop, and t is a time variable. Characteristic loop type audio conference circuit. 3. When inputting a transmission input signal Sin (t) from its own terminal, half (N / N) of a plurality (N) of terminals are input.
The loop type audio conference circuit according to claim 1 or 2, wherein the 2) terminals are configured to input the inverted signal -Sin (t) and the remaining terminals input Sin (t) as they are. 4. When measuring the values of the total delay time T and the number of conference participants N, which are the parameters of the loop, prior to the call operation, the chairperson device as the advocate of the conference is first determined after the loop connection is completed. The pattern A is sent for a predetermined time, and then another pattern B is sent for a predetermined time. On the other hand, each of the member devices of the conference assembled in the loop is set a unit delay time, and a unit delay time is set. It is assumed that the data is transferred to the transmission channel, and the above-mentioned chairman analyzes the received data that has gone through the loop to find the change time point between pattern A and pattern B, and calculates the difference between the change time point at the time of transmission and the total delay of the loop. The time T is measured, and N = 0 is sent as the value of the number of participants counter. On the other hand, when each member device receives the counter value, it increments +1 and participates in the loop. When the chairman device receives the counter value which has been stored as a sequence value and is sent to the next member device and which has gone through a loop, the chairman device increments it by 1 to obtain the value of the number N of conference participants. The value (T, N) is delivered to all members in the loop using the same channel, and when each member receives the (T, N) value, this value is set in the circuit unit according to claim 1 or 2. Further, in accordance with the stored even number / odd number of the sequence value, shifts to the talk mode by controlling the positive / negative phase of the transmission amplifier according to claim 3, and the chairperson device itself also becomes a member according to the (T, N) value. Similarly, a method of establishing a loop type audio conference circuit characterized by setting a circuit parameter section and shifting to a call mode.