JPS62163424A - Acoustic echo canceller - Google Patents

Abstract

PURPOSE:To improve the echo cancelling capability at a high frequency by adding a pre-emphasis circuit to an echo canceller of the band split form so as to apply the emphasis to a signal component at high frequencies by giving a higher gain to that at low frequencies. CONSTITUTION:A reception signal given to a terminal 14 is fed to an echo path 8 and to a pre-emphasis circuit 3A. The signal through the circuit 3A is fed to an LPF 4A, an HPF 5A and echo cancellers 1, 2. The echo cancellers 1, 2 apply echo cancellation and correct the tap coefficient of adaptive filters 10, 20. The LPFs 4A, 4B apply the decimation processing in response to the degree of band reduction from the original transmission band to reduce the signal rate. An LPF 4C applies the interpolation processing to restore the signal to the original rate.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an acoustic echo canceler, and particularly to an acoustic echo canceler suitable for canceling room echoes in conference communication systems that transmit broadband audio signals.

[Conventional technology]

Echo cancellers are used in conference communication systems to cancel room echoes that occur during calls due to acoustic coupling between indoor speakers and microphones. This acoustic echo canceller is usually equipped with an adaptive filter, which is a transversal filter that automatically corrects tap coefficients, and generates a pseudo echo signal by passing the received audio signal from the other station through the adaptive filter. The pseudo echo signal is subtracted from the transmitted audio signal generated by the microphone to cancel out and eliminate the indoor echo component. However, if the frequency bandwidth of the audio signal to be transmitted is wide, the number of taps in the adaptive filter must be increased in proportion to the bandwidth, and the signal processing time in the adaptive filter must be increased in inverse proportion to the bandwidth. must be shortened, making it difficult to realize the circuit. In order to avoid such difficulties, a band-splitting echo canceller that divides the transmission band into two parts, low and high, and performs echo cancellation for each of the low and high valley bands, has been developed. Some proposals have been made (for example, the echo canceller described in the National Conference of the Institute of Electronics and Communication Engineers, Lecture No. 1994). It is easy to change the number of band divisions to three or more, and generally, by using a band division format in which the transmission band is divided into a plurality of bands, the above-mentioned difficulty in realizing the circuit can be solved.

[Problem that the invention seeks to solve]

However, with conventional band-splitting acoustic echo cancellers, the level of the high-frequency components in the audio signal is considerably lower than that of the low-frequency components, so the problem is that the desired echo canceling ability is often not achieved in the high-frequency range. There is a point. In other words, the echo canceller generally indicates the echo canceling ability.
The return loss enhancer 1- (ERLE) increases as the level of the input received signal increases, and conversely decreases as the level of the input received signal decreases. On the other hand, in the spectrum distribution of an audio signal, the level is usually high in the low range, and the level decreases as the frequency increases. Therefore, in the band-splitting type echo canceller, the echo canceling ability in the high frequency band is often insufficient.

SUMMARY OF THE INVENTION An object of the present invention is to provide an echo canceller for phonetic writing that solves all of the above-mentioned problems and has sufficient echo canceling ability even in high-frequency divided bands in addition to a wide transmission bandwidth.

[Means for solving problems]

The echo canceller of the present invention includes a first filter group that divides a first received signal into a plurality of band components, and a second filter group that includes an echo generated by the first received signal via an echo path.
a second filter group that divides the received signal of the filter into a plurality of band components; and generates a pseudo echo signal for canceling the echo in response to a transmission signal of one filter of the first filter group, respectively; a plurality of echo cancellers that subtract from and transmit signals from one filter of the second filter group corresponding to the second filter group, and a third filter group for band-synthesizing the transmit signals of the plurality of echo cancellers; an adder for adding and transmitting signals from each filter of the third filter group; and an adder for giving a predetermined gain to each of the band components divided by the first and second filter groups. for giving the output signal of the adder a gain set such that the sum of the two pre-emphasis circuits and the gain given by the pre-emphasis circuit has a substantially flat frequency characteristic over the entire transmission band; and at least one de-emphasis circuit.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

FIG. 1 is a block diagram showing a first embodiment of the present invention. The echo canceller of this embodiment has a pre-emphasis circuit (PE) 3 in addition to the conventional band division type echo canceller.
It has a configuration in which A and 3B are additionally connected to a de-emphasis circuit (DE) 7, and transmits and receives audio signals to and from the outside via terminals 12 to 15. For example, when used in a conference communication system, the echo path 8 consists of a speaker [4< bus for transmitting the received signal from the other station, and a path for guiding the transmitted signal emitted from the microphone of the own station. This transmission signal is
2 and after passing through PB3B, it is passed through a low-pass filter (
The signal is divided into both low and high band components by an LPF (LPF) 4B and a high pass filter (HPF) 5B, respectively. Both band signals each have an echo canceller (EC) 1 for low frequency.
After being subjected to echo cancellation through echo canceller (EC) 2 for high frequencies, LPF4C and HPF5C
are further added to each other in an adder 6. The output signal of the adder 6 passes through a de-emphasis circuit (DB) 7, and then is sent out from a terminal 13 as a transmission signal to the other station.

On the other hand, the terminal 14 is given a received signal from the other station. This received signal is sent as it is to the echo path 8 via the terminal 15, and is also sent to the PE 3A. PE3A
The received signal that has passed through the LPF 4A and HPF'5A is divided into low and high band components, respectively, and then sent to the ECI and EC2. EC1 (or E
C2) is the adaptive filter 10 (or adaptive filter 2).
0) generates a pseudo echo signal that spreads to the low frequency component (or high frequency component) of the received signal, and subtracter 11 (or subtracter 21) subtracts the pseudo echo signal from the transmitted signal to generate a low frequency component ( In addition, the tap coefficients of the adaptive filter 10 (or the adaptive filter 20) are automatically corrected according to the value of the output signal of the subtracter 11 (or the subtracter 21).

PE3A and PE3B have the same frequency characteristics, for example, have frequency characteristics that exhibit a slope characteristic in which the gain in the high range is higher than in the low range, and the high range has a relatively lower level than the low range. This is to emphasize the signal components of the signal.

LPF4A and 4B (or) (PE5A and 5
In B), the low frequency component (or high frequency component) is separated and decimation processing is performed according to the degree of bandwidth reduction from the original transmission band to reduce the signal rate.

Conversely, LPF4C (or HPF5C) performs interbolation processing to return to the original signal rate. Further, DB7 has a frequency characteristic that exhibits an inverse characteristic to PE3A and 3B, that is, a frequency characteristic that restores a flat characteristic when combined with PE3A and 3B.

FIG. 2 is a characteristic diagram illustrating the frequency characteristics of the audio signal and the reverberation time of the echo path 8 to be subjected to echo cancellation according to this embodiment. Frequencies ft and fh indicate the lower limit and upper limit frequencies of the transmission band, respectively, and frequency fm is a frequency set to divide the transmission band into two. As shown by the solid line, the frequency characteristics of the audio signal are considerably lower in the high range, ie, the frequency band f m -f h, than in the low range, ie, the frequency band ft - fm. Furthermore, the reverberation time-frequency characteristics of room echoes are also much shorter in the high range than in the low range, as shown by the broken line.

When echo cancellation is performed using a conventional frequency division type echo canceler for an audio signal with such characteristics,
Echo cancellation ability in high frequencies is insufficient. In this embodiment, as mentioned above, the high frequency signal components are enhanced by giving higher gain than the low frequency through KPE3A and 3B.
Sufficient echo cancellation ability can be obtained even at high frequencies. Of course, like the conventional frequency division type echo canceller, the signal rate after band division can be lower than the original signal rate, and the difficulty in realizing a circuit for wideband transmission can be solved. In addition, as shown in Figure 2, the reverberation time of room echoes in high frequencies is much shorter than in low frequencies, so the number of taps of the adaptive filter 20 of EC2 for high frequencies is set to be smaller than that of ECI for low frequencies. It is possible to reduce the circuit size considerably.

FIG. 3 is a block diagram showing a second embodiment of the invention. In this example, PE3A and 3B in the first example,
The change was made so that DB7 was connected to the circuit path after band division. In other words, PE3 after LPF4A
0A, PE31A after HPF5A, PE30B and PE31B after LPF4B and HPF5B, and DB70 and DB71.
is connected in front of LPF4C and LPF5C. PE30A and 31A (or PE30B
and 31B) are PE3A and 3 in the first example.
As with B, it is clear that it is sufficient to provide frequency characteristics such that the gain in the high range is greater than that in the low range.

Note that both the first and second embodiments are cases in which the transmission band is divided into two, but it is easy to change the number of band divisions to three or more, and the same effect as in the above embodiment can be obtained. It is clear that

〔Effect of the invention〕

As described above, the present invention has the effect of realizing an acoustic echo canceller that has sufficient echo canceling ability even in high frequency subbands when the transmission bandwidth is wide.

[Brief explanation of drawings]

FIGS. 1 and 3 are block diagrams showing embodiments of the present invention, and FIG. 2 is a characteristic diagram illustrating the frequency characteristics of an echo path to be subjected to echo cancellation according to the embodiment of the present invention. 1.2...Echo cancellation (EC), 10°20...Adaptive filter, 11.22...
・Subtractor, 3A, 3B, 30A, 30B, 31A, 31
B...Pre-emphasis circuit (PE), 4A~
4C...Low pass filter (LPF), 5A~5C
...High-pass filter (HPF), 6...
Adder, 7, 71.72...De-emphasis times M (DB), 8...Echo path, 12$ 1
Hyper

Claims (1)

[Claims] a first filter group that divides a first received signal into a plurality of band components; and a second filter group that divides a second reception signal including an echo generated by the first reception signal via an echo path into a plurality of band components. a second filter group that generates a pseudo echo signal for canceling the echo in response to a transmission signal of one filter of the first filter group; A plurality of echo cancellers that subtract from and send out signals from one filter, a third filter group for band-synthesizing the outgoing signals of the plurality of echo cancellers, and sending out each filter of the third filter group. an adder for adding and transmitting signals; and at least two adders for giving predetermined gains to each of the band components divided by the first and second filter groups, respectively.
at least one pre-emphasis circuit for giving the output signal of the adder a gain set such that the sum of the gain given by the pre-emphasis circuit has a substantially flat frequency characteristic over the entire transmission band. An acoustic echo canceller characterized by comprising one de-emphasis circuit.