JPS6041907B2 - side sound protection circuit - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a side sound protection circuit that is advantageous when applied to a public address telephone.

Amplified telephones are one type of telephone terminal service, but they have a stronger degree of acoustic coupling than regular telephones.
More margin is required in sidetone characteristics.

For this reason, as is well known in the art, a voice switch circuit is added to the loudspeaker telephone to perform gain control in accordance with the magnitude relationship of the transmitted and received signal levels. However, voice switch circuits inherently have the drawback that they tend to be cut off at the beginning or end of words.
On the other hand, in telephone lines, 2-wire to 4-wire conversion is normally performed between one terminal station, and a hybrid circuit is used as this conversion circuit.

Hybrid circuits do not have any problems when they are matched, but they often become unbalanced due to line fluctuations, branch changes, etc., causing unnecessary signals to wrap around, or so-called echo signals. In order to solve this problem, various echo prevention devices (echo cancellers) have been developed that create a pseudo-echo signal using the received human input signal and the transmitted output signal, and use this to cancel the echo signal due to looping from the received signal. ing. If the principle of this echo canceller is applied to side sound protection circuits for loudspeaker telephones, it is thought that problems such as those of voice switch circuits will be solved.

However, current echo cancellers are designed for long-distance transmission applications, and therefore require high-speed convergence control, complex correction control, or signal processing to make them adaptive, making the hardware configuration simpler. Hard to get. For example, when trying to configure an adaptive echo canceller, due to the unique characteristics of audio signals, the signal must be decorrelated (linear predictive coding, etc.) in order to perform a training operation that is appropriate and does not damage the beginning of words, and the training operation must be In order to achieve high-speed convergence, it becomes very complicated, requiring calculations and memory proportional to the square of the taps of the transversal filter for identifying pseudo-echo characteristics. On the other hand,
In the side sound protection circuit of a loudspeaker telephone, it is sufficient to perform correction to ensure a margin for acoustic coupling, so performance such as that of an echo canceller is not required, and in particular, the convergence method can be simplified. That is, the convergence method may not be an adaptive type in which training operations are performed all the time, but a preset type in which training operations are performed only when a line is connected or changed.
By the way, in this case, since the communication is two-way, there is a possibility that the training operations are performed simultaneously.

In that case, training signals for training operations will be mixed into each other's signals at the same time, resulting in the inability of both parties to perform appropriate training operations. Similarly, when a training signal has arrived from the other party's side, even if you try to have your own station perform a training operation, the appropriate operation will not be performed. This invention was made in view of the above points, and its purpose is to provide a side sound prevention circuit that can always be expected to operate accurately without affecting the audio signal and has a relatively simple configuration. It's about doing.

Hereinafter, this invention will be explained in detail with reference to Examples.

FIG. 1 shows an outline of an embodiment of the present invention.

10 is a public address telephone; 11 is a selection circuit that selectively sends out a transmission output signal from this telephone 10 and a training signal from a training signal source 12; the output of the selection circuit 11 is a hybrid for 2-wire to 4-wire conversion; The signal is input to the circuit 13 and also to the sidetone cancellation circuit 14 .

15 is when connecting the line. A detection circuit 16 that detects a training command signal generated when a line changes, etc. and a training signal from the other party's telephone, and a discrimination circuit 17 that determines whether the telephone 10, which is the local telephone, is the calling party or the called party. This circuit controls the selection circuit 11 by its output.

When the transmission output signal from the telephone 10 is transmitted via the selection circuit 11 and the hybrid circuit 13, the sidetone canceling circuit 14 eliminates a signal (sidetone signal) that gets mixed into the receiving human signal due to the imbalance of the hybrid circuit 13. ), an example of which is shown in FIG.

In FIG. 2, 20 is an A/D converting the training signal inputted through the selection circuit 11 into a digital signal.
A converter, the output of which is stored in an input register 21.
The output of this input register 21 and the output of a register 22 that stores estimated sidetone characteristics as tap coefficients are integrated by a convolution integration circuit 23, and the integration result is converted into an analog signal by a D/A converter 24. As a result, a pseudo sidetone signal is created. Therefore, by subtracting the pseudo sidetone signal by the subtraction circuit 25 from the received human power signal inputted through the hybrid circuit 13, the sidetone signal is canceled. Then, the sidetone signal component remaining in the output of the subtraction circuit 25, that is, the error signal, is converted into a digital signal by the A/D converter 26 and input to the tap coefficient correction circuit 27. A modification of the tap coefficient proportional to the product of the signal from the input register 21 and the signal from the input register 21 is performed for each tap. The above is the training motion. Of course, during the period when the training signal is not input, the input register 21,
The contents of the tap coefficient register 22 are held, and the sidetone signal is erased. This type of sidetone cancellation circuit does not have the problems of cutting off the beginning or end of words that occur with voice switch circuits, and there is also less risk of howling.
It can be said to be ideal for public address telephones.

By the way, when the sidetone characteristics change, the sidetone cancellation circuit 14
If the pseudo sidetone signal created within remains constant, new wraparound or howling will occur. In such a case, it is necessary to modify the contents of the tap coefficient register 22 again, that is, to perform retraining. Here, since the sidetone characteristics change only at specific times such as when a line is connected or when the line changes, there is no need to constantly perform training operations like an echo canceller. It is only necessary to perform training when there is a fluctuation. The respective training sequences for the calling side and the called side in this invention will be explained with reference to FIGS. 3 and 4 in the case where both sides are loudspeaker telephones.

First, when the line is connected as shown in FIG. 3a, the calling party performs training preferentially as shown in FIG. 3b. After the training of the calling side is completed, the called side performs training as shown in c of the same figure. Next, when a line change occurs as shown in Figure 4 a, priority is given to the one that started retraining first, but if retraining starts at the same time as shown in Figure 4 B and c, the Priority is given to retraining of the calling side, and the called side immediately stops retraining. After the retraining of the calling side is completed, the called side performs a new retraining as shown in c of the same figure. On the other hand, if only one side, that is, only the calling side or only the called side, is a public address telephone, the sequence is the same as when both sides are public address telephones, except that training is performed after confirming the incoming call or sending out the incoming call signal. Do some training. The training sequence described above is achieved by the control circuit 15. The control circuit 15 will be explained with reference to FIG. When the training command signal 50 shown in FIG. 6a is input when the line is connected, if there is no training signal from the other party's telephone and the output 51 of FIG. 6b of the detection circuit 16 is 4°0, the inverter 52 The output becomes "F', and the training pulse generation circuit 54 is operated by the output of the AND gate 53. The training pulse thus generated is supplied to the selection circuit 11 through the 0R gate path 55 and the M-width gate 56. As a result, the selection circuit 11 selects and transmits the training signal from the training signal source 12. If a training signal is occasionally input from the other party's telephone during this process, the AND gate 56 is closed during this process. The transmission of the training signal is interrupted by the training command signal a.The transmission state of the training signal in this case is shown in FIG. 6c.Furthermore, as shown in FIG. When the training signal from the other party's telephone enters and the output 51 of the detection circuit 16 becomes "1", the AND gate 56 similarly closes during that time, so that after the training signal from the other party's telephone disappears, the output 51 of the detection circuit 16 becomes "1". Training is carried out as follows.

On the other hand, if training is started on both the calling and receiving sides when the line is changed, a training signal from the other party's telephone side is detected immediately after the training command signal 50 is input.

In such case, as shown in FIG. The AND gate 59 detects the simultaneity with the output d of the edge trigger circuit 58, and the training gate pulse generating circuit 60 is operated by the output of the AND gate 59. The training gate pulse e generated by this is applied to the N1 gate 62 controlled by the output 61 of the discrimination circuit 17, and only when the own telephone is the calling party, the output 61 of the discrimination circuit 17 is By becoming 1'', it passes through the AND gate 62 and is applied to the AND gate 55 via the 0R gate 55. In this case, a similar operation is performed on the other party's telephone, but if the local telephone is the calling party, the equivalent of the training gate pulse generation circuit 60 will not operate on the other party's telephone, so the sending of the training signal will be stopped immediately. Stop. Therefore,
On the local telephone side, the training gate pulse e that has passed through the 0R gate 55 becomes AN after the output c of the detection circuit 16 disappears.
It passes through the D gate 56 and is applied to the selection circuit 11.
A training signal is sent out as shown in Figure f. In this way, training is prevented from being performed simultaneously on the calling and called sides, and appropriate training operations are always performed.

In the above description, when both the calling side and the called side start training at the same time, the calling side is given priority, but the called side may be given priority.

In that case, the output 61 of the discrimination circuit 17 may be inverted and input to the control circuit 15. Although a random signal such as an M-sequence pattern or white noise is used as the training signal, it is necessary for the detection circuit 16 to detect this separately from the audio signal.

In this case, it may be possible to use voice recognition, etc.
The configuration becomes complicated and expensive. Detection circuit 16 is shown in FIG.
Here is a simple example. In Fig. 8, the received human power signal 8
1 is applied to a low pass filter 82 and a monostable multivibrator 83.

The output of the low-pass filter 82 is input to a level determination circuit 84, where it is sliced at a constant level and sent to an edge circuit 8.
5, an edge is detected. The output of edge trigger circuit 85 is input to counter 86 . The counter 86 is reset by the output of the AND gate 87. That is, if the output of the level determination circuit 84 does not change during the output pulse period of the monostable multivibrator 83, the counter 86 continues counting the output pulses of the edge trigger circuit 85 without being reset. When the count value reaches a certain value, it is determined that the training signal has been input as the received human power signal 81, and the flip-flop 88 is inverted. Thereby, a detection output can be obtained as the output of the flip-flop 88. As explained in detail above, the sidetone prevention circuit of the present invention estimates the sidetone characteristics using the sidetone signal from the training signal, and cancels the sidetone signal using the pseudo sidetone signal created based on the sidetone characteristics. Therefore, there are no problems with the beginning or ending of words being cut off. Further, since it is a preset type in which the training operation for correcting the pseudo sidetone signal is performed only when the line is connected or changed, the configuration is relatively simple. Furthermore, in this invention, especially when the line changes, when one of the calling and called sides is training while the other starts training, or when both start training, Either one of the training operations can be performed preferentially. Therefore, erroneous operation due to the mixing of both training signals is prevented, and a reliable sidetone cancellation effect can be achieved through correct training operation.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an outline of a side sound prevention circuit according to an embodiment of the present invention, FIG. 2 is a diagram showing a specific example of a side sound canceling circuit, and FIG.
4 and 4 are time charts showing the basic operation of the same embodiment, FIG. 5 is a diagram showing a specific example of the control circuit, FIGS. 6 and 7 are time charts showing the operation, and FIG. 8 is a detection FIG. 3 is a diagram showing a specific example of a circuit. 10... Loudspeaker telephone, 11... Selection circuit, 12...
- Training signal source, 13... hybrid circuit,
14... Sidetone cancellation circuit, 15... Control circuit, 16.
. . . Training signal detection circuit, 17 . . . Outgoing/incoming call discrimination circuit.

Claims (1)

[Claims] 1. A selection circuit that selectively sends a training signal and a transmission output signal from a local telephone to a hybrid circuit for 2-wire to 4-wire conversion, and a training signal and its sidetone signal sent from this selection circuit. A circuit that estimates the sidetone characteristics from the source to create a pseudo sidetone signal and subtracts this signal from the received input signal to the local telephone to eliminate the sidetone signal, and a detection circuit that detects the training signal from the other party's telephone. and a discrimination circuit for discriminating whether the local telephone is a calling party or a called party, and a control circuit for controlling the selection circuit according to a training command signal and the outputs of the detection circuit and discrimination circuit, If a training command signal is input while a training signal from the other party's telephone is not detected, the control circuit causes the selection circuit to send out a training signal, and the control circuit causes the training signal to be sent from the selection circuit within a certain minute time after the training command signal is input. The prevention device is characterized in that, when a training signal from the telephone side is detected, the selection circuit is configured to transmit the training signal only when the local telephone is either the calling side or the called side. Sidetone circuit.