JPS58131827A - Echo suppressing device - Google Patents

Abstract

PURPOSE:To reduce the deterioration of an echo return loss improving amount when a nearest call is generated, by inhibiting the operation of a weight value correction circuit when a function of the absolute value of a weight value correction output exceeds a prescribed threshold value. CONSTITUTION:When an echo suppressor enters a stable operating state, a weight value is not almost corrected and the weight value equal to the value just before is given from a load value correction circuit 15. In this state, if the nearest call takes place, an adaptive control circuit 13 is operated to cancel it temporarily to generate a weight value correction output with a larger absolute value. A control circuit 17 integrates and sums the absolute value of the weight correction output for a prescribed sample number and when this integrated and summed value exceeds the preset threshold value, a switching circuit 14 is opened, and the correction of a new weight by the circuit 15 is inhibited.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of the Invention] The present invention relates to an improvement in an echo suppression device for compensating for echo phenomena occurring in long-distance communication lines. In particular, the present invention relates to a device using a learning identification method suitable for digital communication signals (POM signals), which can maintain a large echo return loss improvement amount (ERI, K) when a near-end call is present.

[Description of prior art]

FIG. 1 shows a block diagram of a device using a conventional learning identification method. In FIG. 1, reference numeral 1 denotes an echo suppression device, which includes transmitting human power Sin, transmitting output 5 out, receiving human power R1n,
and reception output Rout; Reference numeral 2 designates a transmission line, for example, a domestic transmission line, and the subscriber terminal 4 is connected to the far end of the transmission line via a two-wire/four-wire converter 3'51:2. Terminals 5O1lt and Rln are connected to extremely long distance communication methods such as satellite communication or submarine cables. Here, the above four terminals Fin, 3out,
No. 16 passing through Rln, Rou, and t is a digital communication signal, for example, a PCV signal having a sampling period of 125 μs and an 8-pit configuration.

The signal at terminal R1n is taken into delay circuit 10. This delay circuit] is a well-known tapped delay circuit that connects the terminal I (in
It consists of registers that sequentially store signal bits and shift them sequentially. The readout output of this delay circuit 10 is input to a convolution calculation circuit 11. This circuit 11
is configured to multiply the output x (k) of the delay circuit 10 by a weight value hn(k)'ii=to be described in detail later, and add the result over a predetermined number of samples N.

The output of this convolution calculation circuit 11 is input to a subtraction circuit 12, and the difference with the transmission input signal (Sin) is taken.
It is sent out as a transmission output signal (Bout).

The output y(k) of the convolution calculation circuit 11 is connected to the terminal Rou
If the signal passing through t is exactly equal to the echo y(k) generated by reflection at the two-wire/four-wire converter 3 and arriving at the terminal Sin, the echo y(k) is canceled at the terminal 5out.

In order to control the load value, the output of the subtraction circuit 12 and the output of the delay circuit 10 are taken into the adaptive control circuit 13,
Here, a load value correction output Δh n (k) f is generated.

This load value correction output Δh n (k) is
is applied to the load value correction circuit 15 via. here,
This load value correction output is added to the immediately previous load value 7 to obtain a new load value h n (k), which is applied to the convolution calculation circuit 11 .

In other words, the adaptive control circuit 13
When positive or negative load/value correction outputs Δhn (k) f occur one after another, the operation of this system gradually converges and the echo at the terminal Bout is maintained at its smallest state.

In this state, there is a call being sent from subscriber terminal 4, which is connected to terminal Sin.
, it passes through the subtraction circuit 12 and is sent to the terminal 5o.
Appears in ut. Since the signal at this terminal 5out at this time is not an echo, this signal must not be controlled to be small. The near-end call detection circuit 16 detects the presence of a near-end call by detecting the signal levels of the terminals Sin and 5out, and when there is a detection output, opens the opening/closing circuit 14 and corrects the load value of the adaptive control circuit 13. The output is prevented from being applied to the load value correction circuit 15. As a result, the load value correction circuit 15 is prohibited from correcting the load value only while near-end communication exists, and continues to send out the load value before IU.

FIG. 2 shows changes in the echo retardance improvement amount (ERLK) of such an echo suppression device. When this device is reset at time 0 in FIG. 2, load value correction outputs are sent out one after another as described above, and as ERLE increases, they reach the saturation point E. When a near-end call starts at time T1, a new load value correction output is generated until it is detected by the near-end call detection circuit 16 at time T2, so that ERLF!
gradually deteriorates. At time T2, the switching circuit 14 opens,
Since there is no correction to the load value from this time T2, BRLB
becomes constant at the value E2. When the near-end call disappears at time T5, the load value is corrected again and increases toward the FiRLK threshold E1.

When a near-end call occurs in this way, there is a phenomenon in which the ERLK deteriorates until it is detected. This is because the near-end call detection circuit 16 determines that there is a near-end call based on the signal levels of the terminals Rout and Sin, for example, when the level difference therebetween exceeds a threshold value. In other words, the method of detecting the presence or absence of a near-end call based on this level is an excellent method with accurate operation, but in real call audio, a level that suddenly exceeds the threshold is reached from the start of the call. Since there are few numbers, the detection will inevitably be delayed, and the ERLB will be degraded by that amount.

Regarding this conventional device, see Takaran, Itamoto, and Watanabe, "General Purpose Echo Canceller," Electronics and Communication Technology Committee Study Group [Communication Methods Study Group] sponsored by the Institute of Electronics and Communication Engineers, December 22, 1980, document number O8-81. -139 has a detailed description.

[Object of the present invention]

The present invention aims to improve this problem and to provide an echo suppression device that reduces the deterioration of the echo return loss improvement t (ERLE) when a near-end call occurs.

[Key points of the invention]

The present invention is configured to prohibit the correction of the load value based on the load value correction output obtained at the output of the adaptive control circuit, assuming that a near-end call has started when this change is rapid. In particular, the present invention is characterized by comprising means for prohibiting the load value correction operation of the load value correction circuit when the function of the absolute value of the load value correction output exceeds a predetermined threshold value.

[Explanation based on examples]

FIG. 3 is a block diagram of an apparatus according to an embodiment of the present invention. Compared to the conventional device shown in FIG. 17 is passed through an OR circuit 18, together with the output of the near-end call detection circuit 16, and is characterized in that it can be applied manually to the control i11 of the opening/closing circuit 14. The configuration of other parts is similar to the conventional device shown in FIG.

In this example, the control circuit 17 cumulatively adds the absolute value of the load value correction output for a predetermined number of samples (for example, 8 samples),
It is detected whether this cumulatively added value exceeds a preset threshold. When this value is exceeded, a signal is sent to the output, and in this case, the switching circuit 14 is opened and the load value correction circuit 15 is prohibited from making a new correction of the load value.

That is, once the echo suppression device enters a stable operating state,
Almost no correction is made to the load value, and a load value that is equal to the immediately previous load value is sent out from the load value correction circuit 15. When a near-end call occurs in such a state, the adaptive control circuit 13 operates to temporarily cancel the call and generates a load value correction output with a large absolute value. The control circuit 17 monitors this, quickly detects that this is due to the occurrence of a near-end call, and prohibits the load value correction operation.

Returning to FIG. 2 to explain this, when a near-end call occurs at time T, this is detected at time T21, and modification of the load value is prohibited from this point on. As a result, the echo return loss improvement amount (ERLE '1' is maintained at the value E2' (dotted chain line in FIG. 2).

The disappearance of the near-end call is detected at time T3', and recovery begins from this point. Therefore, the deterioration of ERLFi is smaller than in the case of the conventional device, resulting in good characteristics as a whole.

The operation of the control circuit 17 is to calculate N (here, a predetermined number of samples, for example 8) and detect whether or not this exceeds a predetermined threshold. Alternatively, the number of samples Ni may be made smaller or larger. It is easiest to extract absolute values from hardware.

In addition, it is possible to detect changes over time (time differentiation) in the absolute value of the load value correction output, etc., using functions based on the absolute value.
This sudden change in the load value correction output can be detected using various calculation forms.

FIG. 4 is a diagram showing an example of the configuration of the control circuit 17.

This input contains the absolute value of the load value correction output.
△ △ 1Δho(kl l lΔh, (k)l, lΔh2(k
)I 1...--lΔRin1 (k) N pieces are given in chronological order like 1. This is cumulatively added every N by an accumulative adder 21, and is applied to one input of a comparator 22. A certain threshold value is given to this other input. Further, this comparator 22 is given a set signal C that controls its comparison operation. The output d of this comparator 22 is
It is sent to the output terminal via the no-longover time setting circuit 23.

FIG. 5 is an operational waveform diagram of each part of the control circuit 17.

FIG. 5 a to θ show waveforms at points a to θ in FIG. 4. That is, the line connection response signal is received at time 1=0, and the load (+i
When the l correction circuit 15 is cleared, the echo suppression device operates from the time point t1 has elapsed. At this time, the output of the cumulative adder 21 is zero. In subsequent times, the load value is modified significantly one after another and gradually becomes stable. When the time t2 elapses, the correction meter becomes smaller and stabilized.

After stabilization, a set signal C is applied to the comparator 22 to start a comparison operation. After this stable state continues, the output a of the cumulative adder 21 increases, and when it exceeds the threshold value S, the output d is sent to the comparator 22. This is sent out at the output e and L7 after a slight delay of time t5 in the circuit 23, and the modification of the load value is prohibited.

The output of the accumulative adder 21 becomes small again, and when the output falls below the threshold value, the so-called)-overtime t41
4. Additionally, the output signal θ is stopped.

This hangover time is chosen to correspond to a syllable break in a call, and is provided to avoid unnecessary repeated activations and stops in the middle of a call. The time period during which the modification of the load value is prohibited from 1 to 1 in the operation shown in FIG. 5 is t5.

In the above embodiment, the sampling synchronization T (T in FIG. 5) is 125 μs.

The accumulative adder 21 has a configuration in which a delay circuit whose delay time is equal to the period T is connected to the output of the adder 7, and the output of this delay circuit is feedback-connected to the input of the adder. This delay circuit is reset at a predetermined cumulative addition number.

The embodiment device described in FIG. 3 includes a near-end call detection circuit 16 using level detection, which is also included in the conventional device. This circuit 16 is effective when the near-end call is a voice that suddenly rises at a high level, and its operation is stable. Therefore, as in the above embodiment, this circuit 6 is used in combination with the control circuit I7. This is desirable. However, the present invention can also be implemented by omitting this near-end call detection circuit 16ffi and configuring only the control circuit 7 to prohibit the load value correction operation.

The above explanation assumes that the call signal is a digital signal, but
Although the operation of the echo suppression device is digital, it can also be adapted to the case where the call signal is an analog signal by inserting an AD converter or a DA converter into each of the transmitted and received signals.

[Explanation of effects]

As explained above, according to the device of the present invention, the echo return loss improvement t (KRLE
) can be obtained. In particular, the present invention has an excellent effect of accurately detecting the occurrence of near-end calls whose initial level is low.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a conventional device. FIG. 2 is a diagram showing an example of a temporal change in echo return loss improvement t (BRLE), where the solid line is for the conventional device and the dashed-dotted line is for the embodiment of the present invention. FIG. 3 is a block diagram of an apparatus according to an embodiment of the present invention. FIG. 4 is a block block diagram of the control circuit. FIG. 5 is an explanatory diagram of the operation. ■...Echo suppression device, 2...Transmission line (for example, domestic transmission line), 3...Two-wire/four-wire converter, 4...Subscriber terminal. Patent Applicant Nippon Telegraph's Kamika Public Corporation Agent Patent Attorney Nao Ide Takazume 2 Figures and Figures J￥13

Claims (3)

[Claims] (1) Delay circuit (10
'), a convolution calculation circuit (l]) that multiplies the output of this delay circuit by a weight value and adds the result of the multiplication over a predetermined number of samples; a subtractor circuit (12) for generating a transmission output signal; an adaptive control circuit (13) for generating a load value correction output as a function of the output of the delay circuit so as to reduce the echo appearing in the output of the subtractor circuit; In an echo suppression device equipped with a load value correction circuit (15) that adds a load value correction output to the immediately preceding load value to obtain a new load value, the load value correction output given to the load value correction circuit becomes smaller. The present invention is characterized by comprising means for prohibiting the load value correction operation of the load value correction circuit when the function of the absolute value of the load value correction output exceeds a predetermined threshold after reaching a stable state. Echo suppression device. (2) The echo suppression device according to claim (1), wherein the function of the absolute value of the load value correction output is a value obtained by cumulatively adding this absolute value within a range of a predetermined number of samples. (3) The echo suppression device according to claim (1), wherein the function of the absolute value of the load value correction output is a value obtained by cumulatively adding the square value of the load value correction output within a predetermined number of samples. .