Brief Description of the Figures
These and other objects of the instant invention will become apparent when reading the accompanying description and drawings in which:
FIG. 1 shows a block diagram of an embodiment of the invention;
FIG. 2 comprised of 2A-2D shows a signal level diagram for the explanation of the embodiment shown in FIG. 1,
FIG. 3 shows a schematic diagram of a maximum value extraction circuit employed in the embodiment shown in FIG. 1;
FIG. 4 is a block diagram showing a modification for a part of the embodiment shown in FIG. 1; and
FIG. 5 is a detailed block diagram showing an embodiment of a part of the embodiment shown in FIG. 1.
Detailed Description of the Invention
In the echo suppressor (the circuit is adapted completely to digital information signals) of the invention shown in FIG. 1, the calling subscriber's signal, i.e., digitized voice signal, is transmitted through means such as a long-distance communication satellite or submarine cable 100 and is supplied to a four-wire input terminal 101 after being demodulated by suitable demodulator means forming part of the terminal equipment (not shown). The demodulated signal is then led to a two-wire line terminal 106 through a variable attenuator 181, a two-wire output terminal 102, and a hybrid circuit 105, and then sent to a called subscriber set 108 through transmission lines and telephone exchanges designated by the dotted line 107.
In the reverse direction the called subscriber's replying signal (i.e. voice message) is transmitted to the calling subscriber through the lines and exchanges 107, terminal 106, the hybrid circuit 105, a two-wire input terminal 103, a switching means 182 which is normally in the closed state, a four-wire output terminal 104, and the long distance network 100. If the impedance balance at the hybrid circuit 105 is perfect, the calling subscriber's signal supplied at the input terminal 101 can be effectively transmitted to the called subscriber set 108 without any echo. As aforesaid, however, since the impedance unbalance at the hybrid circuit 105 is inherent, a portion of the incoming calling subscriber's signal appears on the two-wire input terminal 103 through the hybrid circuit 105, and is transmitted to the calling subscriber through the four-wire output terminal 104 and the long distance network 100, to be recognized as talker's echo.
To prevent this, the echo suppressor of the invention comprises, in addition to the four-wire terminals 101 and 104 and two-wire terminals 102 and 103: a pulse source 110 for generating timing pulses having a preset repetition period T approximately equal to the delay time in the echo path from the terminal 102 to the terminal 103; the variable attenuator 181 connected to the four-wire terminal 101 for selectively inserting in response to a control signal a prefixed value of attenuation between the four-wire input terminal 101 and the two-wire output terminal 102; terminals 109 for receiving the timing pulses; first and second maximum value extraction circuits 111 and 112 connected respectively to the two-wire output and input terminals 102 and 103 and the terminal 110 for generating maximum values of the calling and called subscriber's signals falling in the period beginning with each one of the timing pulses; an echo path attenuation factor measuring circuit 120 supplied with the output signals from the maximum value extraction circuits 111 and 112 for generating an attenuation-factor-representing signal in the echo path lying between the terminal 102 and the terminal 103 (the attenuation factor being the ratio of the signal level at the terminal 103 to that of the terminal 102) and for providing a delayed and an undelayed incoming signal level; a leakage level estimation circuit 130 for generating an estimated-leakage-level-representing signal by multiplying the echo path attenuation factor by the larger one of the delayed and undelayed incoming signal level; a called subscriber's signal detector 150 connected to the two-wire input terminal 103 and supplied with the estimated-leakage-level-representing signal for generating an output indicating the presence of the called subscriber's signal at the terminal 103; a first hangover circuit 70 connected to the called subscriber's signal detector 150 for allowing the output from the circuit 150 to pass therethrough and for holding it for a prefixed period (for example, 300 milliseconds) so as to generate the output from the called subscriber's signal detector 150 as the control signal; a calling subscriber's signal detector 140 connected to the two-wire output terminal 102 for generating an output indicating the presence of the called subscriber's signal at the terminal 102; a second hangover circuit 160 connected to the calling subscriber's signal detector 140 for allowing the output from the circuit 140 to pass therethrough and for holding it for another prefixed period (for example, 50 milliseconds) so as to transfer the output from the calling subscriber's signal detector 140 as a switch-open signal; an inhibitor 180 supplied with the control signal from the first hangover circuit 70 at an inhibit input terminal 180a and the switch-open signal from the second hangover circuit 160 at another input terminal for generating a switch-deriving signal upon reception of the switch-open signal exclusively in the presence of the control signal; and a switching means 182 interposed between the terminals 103 and 104 for disconnecting the outgoing signal circuit in response to the switch-deriving signal so as to suppress the leakage component.
The echo path attenuation factor measuring circuit 120 is comprised of: a first register 121 connected to the first maximum value extraction circuit 111 and to the terminal 109 for storing the output from the circuit 111 in response to the timing pulses, thereby to generate a signal representative of the calling subscriber's signal level with a delay of the repetition period T of the timing pulses; a second register 122 connected to the second maximum value extraction circuit 112 and the terminal 109 for storing the output from the circuit 112 in response to the timing pulses thereby to generate a signal representative of the called subscriber's signal level with a delay of the repetition period T of the timing pulses as an output thereof; a first selector 123 for selectively passing therethrough the larger signal as between the output of the first maximum value extraction circuit 112 and that of the second register 122, thereby to generate a signal as an output representing the outgoing signal level at the terminal 103 in the time period twice as large as the repetion period (2T) of the timing pulses; a division circuit 124 wherein the output from the selector 123 is divided by the output from the first register 121 for generating the echo-path-attenuation-factor-representing signal whose maximum value is unity and whose minimum is zero; and a third register 125 coupled to the division circuit 124 and the terminal 109 for holding the output of the division circuit 124 for the repetition period T of the timing pulses to the echo-path-attenuation-factor-representing signal from the division circuit 124, so as to facilitate the detection of the initial portion of the called subscriber's signal.
The leakage level estimation circuit 130 is comprised of; a second selector 131 for selectively passing therethrough the larger one as between the output from the first maximum value extraction circuit 111 and the output from the first register 121, thereby to generate a signal representative of the incoming signal level at the terminal 102 in the time period corresponding to the output of the first selector 123; and a multiplier 132 wherein the echo-path-attenuation-factor-representing signal from the register 125 is multiplied with the output from the third selector 131, thereby to generate the estimated leakage level supplied to the called subscriber's signal detector 150.
Referring to the FIG. 2 showing waveforms observed at various points in the embodiment, the operation of the embodiment will be described hereunder.
Under the state where the calling subscriber's signal is present at the terminal 101 and the called subscriber's signal is not present at the terminal 103, the control signal is not provided to the variable attenuator 181 (to be described later) so that the variable attenuator 181 acts as a signal path without any attenuation, whereby the calling subscriber's signal received at the terminal 101 appears at the terminal 102 as it is. Under this state, the calling subscriber's signal having a signal level shown in FIG. 2, waveform diagram (B) in a solid line Fo appearing at the terminal 102 is converted by the maximum value extraction circuit 111 and the first register 121 to the level representing signal being a locus of the maximum values extracted during each of the first time windows of the interval T (the repetition period of the timing pulses as shown in FIG. 2, waveform diagram (A), and then is applied to the division circuit 124 as the divisor data. The first time windows (designated by codes I.sub.1, I.sub.2, . . . , I.sub.n. . . ) are in synchronism with the timing pulses. The leakage (echo) component appearing at the terminal 103, whose level is shown in FIG. 2, waveform diagram (C) in a solid line G, is processed in a manner similar to the incoming signal by the second maximum extracting circuit 112 and the second register 122, and then applied to the first selector 123, whereby the larger one of the two leakage-level-representing signals is selectively supplied to the division circuit 124 as the dividend data. The output from the selector 123 represents therefore a locus of maximum values sampled during each of the second time windows of the interval 2T (twice as large as the repetition period of the timing pulses). The second time windows are in synchronism with the alternate timing pulses, and are designated J.sub.1, J.sub.2, . . . , J.sub.n, . . . as shown in FIG. 2, waveform diagram (C), each corresponding to the first time window I.sub.1, I.sub.2, I.sub.3 . . . , I.sub.n, . . . . Inasmuch as the time interval T is selected approximately equal to the delay time in the echo path as stated above, a part of the calling subscriber's signal present in one of the first time windows (for instance I.sub.1) unavoidably appears as a leakage component in the corresponding second time window (in this case, J.sub.1). Thus, the division circuit 124 generates a substantially real echo-path-attenuation-factor-representing signal which is then given the delay T (the repetition period of the timing pulses) through the register 125 which output is shown in FIG. 2, waveform diagram (D). The echo-path-attenuation-factor-representing signal tends to indicate a higher value under the effect of the difference of the intervals in the first and second time windows, especially at the initial part and trailing part of the calling subscriber's signal. The detection of the called subscriber's signal, however, is not substantially affected by the above-mentioned tendency because the called subscriber's signal rarely begins in these leading or trailing parts of the calling subscriber's signal and the highest attenuation factor is limited to unity, insuring at least the detection capability of the conventional echo suppressor associated with the hybrid having the poorest attenuation factor. When both the calling and the called subscriber's signal levels are zero, the attenuation factor is also set to unity.
The second selector 131 operates in a similar manner to the first selector 123, providing the called subscriber's signal is as shown by F.sub.1 in FIG. 2, waveform diagram (C). The called subscriber's signal is sent to the multiplier 132. The echo-path-attenuation-factor-representing signal from the third register 125 is applied to the multiplier 132, where the level of the called subscriber's signal is multiplied by the echo path attenuation factor, so as to generate the estimated-leakage-level-representing signal as shown by E in FIG. 2, waveform diagram (C).
Now the estimated leakage level is supplied to the called subscriber's detector 150, where the called subscriber's signal level is compared therewith. Under this state, the detector 150 does not provide any output signal because there is no called subscriber's signal. As a result, no control signal is supplied from the first hangover circuit 70 to short-circuit the variable attenuator 181.
On the other hand, the calling subscriber's signal is detected by the detector 140, and the output therefrom is supplied as a switch open signal to one of the input terminals of the inhibition gate circuit 180 through the second hangover circuit 160. Since there is no control signal at the inhibit input terminal 180 a under this state, the switchopen signal is applied to the switching means 182 through the inhibitor 180. The switching means 182 is opened accordingly, so that the leakage component appearing at the terminal 103 is suppressed and prevented from being transmitted to the calling (remote end) subscriber as talker's echo. The hangover circuit 160 holds the output from the calling subscriber's signal detector 140 for a prefixed period od time, so as to prevent the chattering of the switching means 182.
Now the operation of the embodiment assuming the presence of the called (near end) subscriber's signal will be detailed hereunder. The called subscriber's signal at the terminal 103 is compared at detector 150 with the estimated leakage level supplied from the multiplier 132 as shown by curve E in FIG. 2, waveform diagram (C). Inasmuch as the estimated leakage level represents the closely estimated real leakage (echo) level in the period of the steady state of the calling subscriber's signal, the called subscriber's signal (H in FIG. 2, waveform diagram (C)) superimposed on the leakage level G is detected immediately at time point P. While the estimated-leakage-level-representing signal starts to increase after a preset delay time in accordance with the increase of the called subscriber's signal level, the initial part of the called subscriber's signal having relatively low level is quickly detected. The initial clipping is thus estimated (The initial clipping is inherently inevitable for a conventional echo suppressor wherein the called subscriber's signal is detected at the time point Q through comparison with the threshold level, as shown in curve K in FIG. 2, waveform diagram (C), which is set, to 6 dB (half in amplitude) below the calling subscriber's signal level F1. As shown, the initial clipping extends from time point P to Q.) The preset delay time for the estimated-leakage-level-representing signal is supplied to the third register 125, wherein the echo-path-attenuation-factor-representing signal from the division circuit 124 is stored in synchronism with the timing pulses and read out after one repetition period T. The output of the called subscriber's signal detector 150 is fed as the control signal to the inhibit input terminal 108a of the inhibitor 180 through the first hangover circuit 70, so that the switch-open signal from the second hangover circuit 160 is inhibited. Hence the switching means 182 is closed to complete the outgoing signal path for the called subscriber's signal. The control signal is also led to the variable attenuator 181, which gives a preset attenuation (for instance, 6 dB) to the incoming signal path. As the result the calling subscriber's signal is attenuated, and the reference level (the estimated leakage level) for the detection of the calling subscriber's signal is lowered accordingly. It will now be apparent that with the present circuit structure the initial clipping of the called subscriber's signal can be avoided even for a low-level voice signal.
Description will now be given referring to FIG. 3 which shows an embodiment of the maximum value extraction circuit 111 (identical to the circuit 112) employed in FIG. 1. At the comparator 303, an absolute value (obtained by just eliminating a sign bit) of the digitized voice signal supplied through the terminal 102 and the output of a register 306 are compared. Depending on the result of the comparison, a switch 304 is driven to select the larger signal in absolute value, and then stores the selected signal in the register 306. Accordingly, the register 306 stores maximum absolute values of the input signal falling in the period beginning with each one of the timing pulses supplied at the terminal 109 from the timing source 110 (see FIG. 1). The maximum value extraction circuit is described in detail in U.S. patent application Ser. No. 530,620 filed on Dec. 9, 1974 by the inventors of the present application. Therefore, any further description will not be given here.
Reference is next made to FIG. 4 showing in block form a control circuit to be substituted for the third register 125. While the third register 125 serves as a delay means, it is not capable of preventing the increase of the echo-path-attenuation factor owing to the break-in of the called subscriber's signal or a pause of the calling subscriber's signal, resulting in the interruption of the calling subscribers signal. To prevent this interruption, the attenuation factor increasing rate control circuit 300 has a third selector 401, a fourth register 402, an amplifier 403 and a terminal 109, where the timing pulses are applied to enable a synchronous operation of the circuit. The output of the fourth register 402 is fed back to the third selector 401 through the amplifier 403 which has a fixed amplification factor. The third selector 401 selectively passes therethrough to the fourth register 402 the smaller one of the output from the division circuit 124 and the output from the amplifier 403. Thus, the echo-path-attenuation-factor-representing signal is prevented from increasing very quickly.
Referring to FIG. 5, further details of the division circuit 124 and an echo path attenuation factor holding circuit 510 coupled thereto are shown in block form. The calling subscriber's signal level F from the first register 121 is applied to the first priority encoder 501, which converts the level F to a signal F' (integer) representing the exponent part of the logarithm of F to the base 2. In other words, the priority encoder 501 generates the highest place of a 1 's bit of binary -- expressed input signal, and it can be purchased in the commercial market. In like manner, another second priority encoder 502 converts the called-subscriber's-signal-level-representing signal H to a signal H' (integer) representing the exponent part of the logarithm of H to the base 2. The subtractor 230 executes the subraction of H' from F' to generate the difference thereof. The difference represents the ratio of the called subscriber's signal level to the calling subscriber's signal level, because the F' and H' both indicates the logarithm of the F and H to the common base 2 respectively.
The signal representative of .alpha. is applied to the echo path attenuation factor holding circuit 510 which prevents the same from decreasing towards zero quickly, eliminating the chattering of the switch means 182.
For this purpose, the comparator 511 compares the .alpha.- representing signal with the output from a register 512. When the result of the comparison shows that the .alpha.-representing signal is larger than the output from the register 512, the comparator 511 supplies a signal SET 1 to the register 512 where the .alpha.-representing signal is stored in synchronism therewith, and also supplies a signal CL1 to a counter 513 which is cleared thereby. When the .alpha.-representing signal is smaller than the output of the register 512, the signals SET 1 and CL1 are not generated to hold the contents of the register 512. Instead, a signal SET 2 is supplied to the counter 513 which counts up occurrences of the signal SET 2 by unity. When the counted number of the counter 513 reaches a preset number, it clears itself and generates a siganl CL2 which clears the register 512.
Since the comparison is synchronized with the timing pulses applied at a terminal 109, the preset number corresponds to a certain time period defined by an integral multiple of a unit time interval T of the timing pulses. Therefore, the echo-path-attenuation-factor-representing signal obtained at the output of the register 512 is held for the certain time period and prevented from decreasing quickly to eliminate the chattering.
In the description heretofore, the calling and called subscriber's signals are assumed to be digitized signals. Therefore, to adapt the embodiment shown in FIG. 1 to the anologue type communication network, two conventional anologue-to-digital converters are needed. One of the converters may be disposed between the terminal 102 and the first maximum value extraction circuit 111 and the detector 140, and the other between the terminal 103 and the second maximum value extraction circuit 112 and the detector 140.
Although electro-magnetic switches have been shown in FIGS. 1 and 3 as the switching means 182 and 304, any other types of switches may be substituted therefore such as, for example, electronic switches.
While the invention has been described in connection with the specific embodiment, it is to be clearly understood that this description is made by way of example and not as a limitation to the scope of the invention as set forth in the accompanying claims for patent.