JPS6051820B2 - Digital audio input method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention utilizes a digital audio insertion method (DigitalSp).
echInterpolation System (abbreviated as DSI).

With the DSI method, there are 5 audio pauses in telephone conversations.
It is adopted to improve the efficiency of the line by taking advantage of the fact that it occupies more than 0%.

In other words, a certain DSI terminal station monitors the status of lines connected to a large number of subscribers (hereinafter referred to as subscriber channels), selects only subscriber channels with audio, and sends them to the other party's DSI terminal station. On the other hand, transmission is performed using DSI channels whose number is half the number of subscriber channels. The other party's DSI terminal station correctly connects the DSI channel to the subscriber channel of the corresponding subscriber, and the call is then completed. Therefore, in the DSI system, lines in use are disconnected during calls in order to selectively connect the subscriber channel and the DSI channel.

Note that the above line may not necessarily be disconnected. In other words, lines with signals are successively connected, and when there is insufficient signal, the lines without signals are cut off and new lines are connected. In this way, the line is often left uninterrupted, so that disconnections at the beginning of speech can be reduced in the case of rhyme communication.
Since the above-described operation is normally performed in DSI, the line may be connected even when there is no signal on the transmitting side. As a result, the subscriber may hear room noise or line noise while the other party is talking, but sometimes there is no sound at all, which makes the subscriber feel very unnatural. For this reason, conventionally, when a DSI channel and a subscriber channel are separated, the receiver side inserts low-level noise into the separated subscriber channel to eliminate the ゜゜silence state. The difference between the background noise and the inserted noise makes the call unnatural. Further, in a DSI system, a voice detector is required that has good detection sensitivity for the voice (transmitted signal) of a near-end speaker and has few malfunctions due to noise.

By the way, general telephone lines include a mixture of four-wire lines and two-wire lines, and a hybrid coil is used at the connection point of both lines, and impedance mismatch occurs due to variations in the characteristic impedance of the lines, causing echoes. cause
If such a reverberant signal is input to the transmitting section of the DSI, the voice detector will respond to it and determine if the period during which voice is present in the subscriber channel on the transmitting side (voice activity day) exceeds 50%. This will interfere with the operation of the DSI system. Particularly on long-distance lines such as international lines, the presence of reaction signals causes communication interference, so the necessary echo suppression equipment is used in advance, so even if the DSL system is newly introduced, such problems will not occur. Although this does not occur, when the DSI's system is introduced to an international line, etc., which has not previously been equipped with an echo prevention device, an echo prevention function becomes necessary.
In this case, there is a problem 4 in that installing an echo prevention device independently of the DSI increases cost and equipment installation space, and also complicates maintenance.

An object of the present invention is to provide a DSI system that eliminates the unnatural "silence" caused by the operation of the echo suppression function.

Another object of the present invention is to provide a DSI with a function of adapting the level of artificial noise inserted when a DSI channel is disconnected and when an echo rejection operation is being performed to adapt to the background noise during a call. The goal is to provide a method.

According to the present invention, only subscriber channels with audio are selected from subscriber channels connected to a large number of subscribers and transmitted through a digital audio insertion channel whose number of channels is smaller than the number of subscriber channels. On the receiving side, in a digital audio insertion method that correctly provides the signal sent from the digital audio insertion channel to the corresponding subscriber channel, the level of the signal of each channel received through the digital audio insertion channel is observed. When a subscriber channel connected to the digital audio insertion channel is disconnected from the digital audio insertion channel, noise whose level is controlled according to the immediately previous signal level of the subscriber channel is applied to the subscriber channel. A digital audio insertion method is obtained, which is characterized in that the audio is inserted.

Further, according to the present invention, only subscriber channels in which audio exists are selected from subscriber channels connected to a large number of subscribers, and transmitted through a digital audio insertion channel whose number is smaller than the number of subscriber channels. However, on the receiving side, in a digital audio insertion method that correctly provides the signal sent from the digital audio insertion channel to the corresponding subscriber channel, the signal level of a certain subscriber channel on the transmitting side is determined by the signal level of the corresponding subscriber channel on the receiving side. When the signal level is greater than the level obtained by subtracting a certain appropriate value, the allocation of the digital audio insertion channel is requested, and the signal level for transmission is greater than the corresponding signal level on the receiving side subtracted by a certain appropriate value. When the signal level on the receiving side is small and exceeds a certain predetermined threshold, the estimated value of the noise level of the subscriber channel on the transmitting side, which has been obtained so far, is used instead of the signal of the subscriber channel on the transmitting side. A voice insertion method is obtained, which is characterized in that low-level noise is inserted in response to the noise. Next, the present invention will be explained in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a conventional DSI system that does not include an echo prevention function. i-th (
Near-end subscriber transmit signals (1=1 to N) are applied to input terminal 1 of PCM encoder 110 through subscriber channels. The PCM encoder 110 converts the transmission signals from the N terminals 1 to N into a time-division multiplexed digital signal S, and supplies the digital signal S to the speech detector 130 and the digital delay circuit 120. The voice detector 130 receives the time-division multiplexed digital signal S from the PCM encoder 110, detects a line with a voice signal, and transmits a channel allocation request signal to the channel allocation circuit 150. On the other hand, the digital delay circuit 120 delays the signals of each line by a certain period of time and provides them to the speech memory 140. Channel assignment circuit 15
0 determines the combination of the subscriber channel and the DSI channel in accordance with the channel assignment request signal from the voice detection circuit 130, and selects the one that has been assigned the DSI channel among the input signals given to the speech memory 140. Write to the memory location corresponding to the DSI channel. Message and speech memory 140 representing the allocation status determined by channel allocation circuit 150
The contents are time-division multiplexed in the output circuit 160 and sent to the DSI channel through the terminal 165. 1st
A partner station having the same configuration as shown in the figure receives a signal from the partner station sent through the DSI channel at a terminal 165'', and an input circuit 170 separates the assignment status message from the speech information, and the assignment status message is sent to the receiving channel. To the allocation circuit 190, speech information is provided to the receive speech memory 140.

The reception channel allocation circuit 190 controls the write address or read address of the reception speech memory 180 in accordance with the allocation status message sent from the transmitter. The gate circuit 200 includes a receiving speech memory 1.
80 and a gate signal provided from terminal 195 allows only information corresponding to the subscriber channel currently connected to the DSI channel to pass through. The noise generator 2F0 generates low-level noise and supplies it to the gate circuit 220. Gate circuit 220 sends noise to coupling circuit 230 only when gate circuit 200 is off. As a result, a time-division multiplexed signal R for N subscriber channels appears at the output of the combining circuit 230, with low-level noise inserted during voice pause periods. The PCM decoder 240 decodes and demultiplexes this time-division multiplexed signal R to each subscriber terminal 1'' to N connected to the subscriber channel.
The method for realizing the SI method is described in the following patent publications published in 1962: (1) 1972-31805 rTASI control method, (2) 1972 33409rTASI control method, (3) 1977-33410rTASI control method, and (4) 1982-33411 "TASI control method," so the following explanation will be limited to only those directly related to the present invention. We will proceed with this.

FIG. 2 is a block diagram showing an embodiment of the present invention, in which the digital delay circuit 120 and the audio detector 1 in FIG.
30, the noise generator 210, the gate circuit 220, and the coupling circuit 230 are modified.

In Fig. 2, the parts marked with reference numbers in the 100-man range indicate newly added parts, and the components indicated with reference numbers in the 100-20 man range have the same meaning as the same reference numerals in Fig. 1. belongs to. Reference numbers in the 100 range in Figure 2, 1310 and 1420 are level detectors, 1320 and 1460 are amplitude limiters, 1330 and 147
0 is a data hold circuit, 1360 and 1430 are comparators, 1370 and 1440 are hangover circuits, 1
380 is a forbidden gate, 1340, 1410 and 148
0 indicates an attenuator, and 1350 indicates an echo rejection switch. Also, reference numeral 1450 is a fixed threshold input line. j The level detector 1420 is connected to the terminal 2 via the attenuator 1410.
05 is detected and its output is provided to comparators 1360 and 1430 and amplitude limiter 1460.

The level detector 1310 detects the level of the 7th transmission signal S applied to the terminal 115 and sends the output signal to the comparator 1.
360 amplitude limiter 1320. If the loss of the echo path (path from j'' to i in Figure 1) is considered to be greater than B (the smaller the loss, the greater the echo),
A comparator 1360 compares the received signal level with a value that is 6 cIB smaller than the transmitted signal level. Furthermore, if the loss of the echo path is known, that value is subtracted from the transmitted signal level and compared with the received signal level. At this time, if the value obtained by subtracting an appropriate value from the transmitted signal level is higher, the signal is transmitted to the inhibit gate 1380, the attenuator 1410, and the terminal The voice detection symbol is output to 135. The voice detection signal applied to the terminal 135 is used as a channel allocation request signal as described above. Period prohibition gate 13 when voice detection signal exists
The output of 80 is always 0, and the attenuator 1410 operates to produce a constant predetermined attenuation between the input and output signals during the period when there is a transmitted signal and even when there is a received signal. The comparator 1430 compares the received signal level with a certain threshold 1450, and if the received signal level is higher than the threshold 1450, provides an echo prevention signal to the inhibit gate 1380 via the hangover circuit 1440.At this time, If the received signal level is higher than the transmitted signal level, the inhibit input voltage of the inhibit gate 1380 becomes 0, so the output of the inhibit gate 1380 becomes "1". In this state, it means that there is sound on the receiving circuit side, so the echo prevention switch 1350 connects the output of the digital delay circuit 120 to the terminal 1.
25 and the low level noise provided from the noise generator 210 is routed through the attenuator 1340 to the terminal 12.
Output to 5. Therefore, the echo signal is prevented from being sent to the receiving section through the terminal 125. When the transmitted signal becomes larger than the received signal, the output of the inhibit gate 1380 is ゜゛0.
Therefore, the echo prevention switch 1350 is immediately connected to the transmitting signal side. At the same time, a channel assignment request signal was applied to the channel assignment circuit 150 (FIG. 1) through the terminal 135. If the subscriber channel is not connected to a DSI channel, after some time the speech signal on terminal 125 is connected to said DSI channel. The digital delay circuit 120 functions to compensate for cutting off the beginning of speech due to delay in voice detection. Like this 3 S
The integrated structure of the I function and the echo prevention function has the effect of reducing unnecessary cutting at the beginning of speech. That is, if the echo prevention device is installed independently of the DSI device, the beginning of the speech is cut off by the echo prevention device, and as a result, the voice is detected again by the DSI device, resulting in an additional delay in voice detection. On the other hand, in such an integrated configuration, the cutoff at the beginning of a speech is caused by only the echo suppression switch (when channel assignment has already been made), or by a delay in DSI channel assignment (when new channel assignment is made; in this case, (the echo rejection switch is always connected first), but not both. Note that when no channel is assigned, the receiving side
As shown, low-level noise is inserted into the output to eliminate unnatural silence.

This function is also achieved in FIG. 2 by inserting the output of the noise generator 210 provided through the attenuator 1480 when no channel assignment is made by the gate circuit 220 and the coupling circuit 230. When the echo prevention switch 1350 on the transmitting side is activated with channel assignment done, noise is not inserted on the receiving side, but in this case, noise is inserted on the transmitting section, so naturalness is maintained in this case as well. is not damaged. Note that the state in which channels are assigned while the echo prevention switch 1350 is operating is when the number of subscriber channels with audio on the transmitting side is small, and when the channels are once connected and kept as they are. This retention is performed to reduce unnecessary line disconnections and cut-offs at the beginning of speech.Furthermore, the noise insertion level is controlled by adapting to the background noise of the speech. This makes it possible to reduce the difference in noise level caused by turning on and off the echo rejection switch 1350 and increase the naturalness of the noise level. This operation is performed by sampling the value at the rising edge of the output inhibit gate 1380 of the transmitted signal level detector 1310 through the transmitter 1320 as described above. , data hold circuit 1330
This is done by writing in . It is only after the sound on the transmitting side becomes low that the sound appears on the receiving side and the echo rejection switch is activated, so that the background noise level is captured by the data hold circuit 1330. The data hold circuit 1 is controlled by the mitter 1320 so as not to generate noise exceeding an appropriate level as background noise.
The input to 330 is clamped at a certain level. The transmitter 1460 and the data hold circuit 1470 are similarly connected to the output of the received signal level detector, and when the channel connection state signal applied through the terminal 195 changes from the connected state to the disconnected state. The received signal level of is written to data hold circuit 1470.

Attenuator 14 depending on the contents of data hold circuit 1470
80 attenuation amounts can be switched. The channel is disconnected after the transmitter no longer detects audio, and this configuration makes it possible to insert noise in accordance with the background noise level. The limiter 1460, like 1320, defines the upper limit of the insertion noise level. Level detectors 1310, 1420, data hold circuits 1330, 1470 and hangover circuit 137
0.1440 is implemented using separate memories for each subscriber channel.

Further, level detectors 1310 and 1420 can be realized by first-order digital filters. Comparator 1360
can be easily realized using a commercially available digital comparator IC. Data old circuits 1330 and 147
0 is also a commercially available random access memory (RandOm
accessmemOry) and a shift register, and is not technically difficult. The attenuators 1340, 1410, and 1480 can be realized by writing values corresponding to several attenuation amounts in advance in a read-only memory (ReadOnlym-EmOry) and switching memory areas. As explained above, the present invention provides a DSI system with a simple circuit in which there is less unnaturalness in calls caused by intermittent connections between the DSI channel and the subscriber channel, and a simple circuit that prevents echoes. It is a DSI method that has functions and makes calls less unnatural.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a conventional digital voice insertion system, and FIG. 2 is a block diagram showing an embodiment of the present invention. In FIG. 2, reference numeral 120 is a digital delay circuit, reference numeral 140 is a speech memory, reference numeral 150 is a channel assignment circuit, reference numeral 160 is an output circuit, reference numeral 170 is an input circuit, reference numeral 180 is a received speech memo). 8 Reference number 200 is a gate circuit, reference number 210 is a noise generator, reference number 220 is a gate circuit, reference number 2
30 is a coupling circuit, reference numbers 115, 165, 165'',
235 is a terminal, reference numbers 1310 and 1420 are level detectors, reference numbers 1320 and 146) 0 are amplitude limiters, reference numbers 1330 and 1470 are data hold circuits, reference numbers 1360 and 1430 are comparators, reference numbers 1370 and 1440 is the hangover circuit, reference numeral 1380 is the inhibit gate, reference numeral 1340,
1410, 5 and 1480 are attenuators, reference number 1350
is an echo rejection switch and reference numeral 1450 is a fixed threshold input line.

Claims (1)

[Claims] 1. Select only subscriber channels with audio from among subscriber channels connected to a large number of subscribers and transmit them through a digital audio insertion channel whose number is smaller than the number of subscriber channels. On the receiving side, in a digital audio insertion method that correctly provides the signal sent from the digital audio insertion channel to the corresponding subscriber channel, the level of the signal of each channel received through the digital audio insertion channel is observed, and the When a subscriber channel connected to a digital audio insertion channel is disconnected from the digital audio insertion channel, noise whose level is controlled according to the immediately previous signal level of the subscriber channel is inserted into the subscriber channel. A digital audio insertion method characterized by the following: 2. Select only subscriber channels with audio from among subscriber channels connected to a large number of subscribers, transmit them through a digital audio insertion channel whose number is smaller than the number of subscriber channels, and on the receiving side In a digital audio insertion method that correctly provides a signal sent from an audio insertion channel to a corresponding subscriber channel, the signal level of a subscriber channel on the transmitting side is set to a certain appropriate level that the signal level of a subscriber channel on the corresponding receiving side. If the level is greater than the level obtained by subtracting a certain value from the corresponding signal level on the receiving side, the signal level for transmission is smaller than the level obtained by subtracting a certain appropriate value from the corresponding signal level on the receiving side, and the signal level on the receiving side is requested to be allocated. is larger than a certain predetermined threshold, the signal of the subscriber channel on the transmitting side is replaced with a low-level noise corresponding to the previously determined noise level estimate of the subscriber channel on the transmitting side. A voice insertion method characterized by inserting.