Zero-delay structure for sub-band echo cancellation

Details Zero-delay structure for sub-band echo cancellation Zero-delay structure for sub-band echo cancellation

1999-10-07

2003-12-09

Tieu, Binh (Department: 2643)

Telephonic communications

Echo cancellation or suppression

Using digital signal processing

C379S406090, C379S406130, C370S291000

Reexamination Certificate

active

06661895

ABSTRACT:

FIELD OF THE INVENTION
This invention relates in general to echo cancellation in telecommunications applications, and more particularly to a zero-delay topology for sub-band echo cancellation. A zero-delay structure, created from this topology, can be incorporated with any sub-band or wavelet echo canceller to ensure the removal of all delays incurred during daily telephone conversations (or telecommunication activities).
BACKGROUND OF THE INVENTION
An echo is the phenomenon in which a delayed and distorted version of an original signal is reflected back to the source. In general, echoes are caused by the reflection of signals from the boundaries of discontinuities in the transmission medium. In the transmission of telephone signals, an echo occurs whenever the transmitted signal encounters an impedance mismatch in the circuit. Such echoes cause impairment on the fidelity of the speech signals and are often detrimental to users.
Network echo cancellers are devices designed to remove these undesirable echoes in telephone transmission. The basic concept of an echo canceller is to synthesize a replica of the echo and to subtract it from the total received signal. An efficient algorithm to obtain the estimated echo path is the Least Mean Square (LMS) adaptive algorithm.
When the echo duration is long (400-500 ms), as in long distance phone calls or in hands- free telephones in a conference room, the number of taps in the adaptive algorithm increases proportionately, and the convergence rate of the adaptive filter slows down significantly. Even more seriously, it may not be able to remove the echoes sufficiently for a good quality speech signal. To overcome this difficulty, the sub-band filtering scheme, and, more recently, the wavelet decomposition scheme, have been proposed. Decomposition of the signal can reduce the computational complexity and can increase the adaptation speed by using the subsampled signals. With M sub-band decomposition, the adaptive filter length is reduced by a factor of M in each sub-band level. Sub-band decomposition is the process of down-sampling, the method of reducing or decimating a sample rate, and band splitting of a signal into several frequency bands. Also, because of the down-sampling procedure, the adaptive algorithm can be implemented at a rate which is M times slower than the original rate. As a result, the computational complexity is reduced, and the convergence rate is improved.
The basic structure for the sub-band echo canceller is shown in
FIG. 1
, where the signal decomposition and reconstruction could be implemented with either filter bank, any block transform, or wavelet decomposition. No matter what methods are used, the signal decomposition and reconstruction can always be equivalently expressed with a set of FIR (Finite Impulse Response) filter banks which satisfy the PR (Perfect Reconstruction) conditions. Both the echo signal and the reference signal will pass through a separate set of decomposition filter banks and then be down-sampled. The adaptive algorithm is operated at the sub-sample level with a reduced computational rate. Then the echo residual signals at different sub-bands will go through the reconstruction filter back to recover an echo-reduced far-end speech signal.
The problem with the sub-band echo canceller is that when the far-end signal goes trough the decomposition and the reconstruction filter banks, a delay is introduced. This delay can be minimized by the careful design of the PR filter banks, but cannot be totally eliminated. Also, in the filter bank design, when the delay is taken into consideration, the performance of the echo cancellation has to be sacrificed, because the minimum delay PR filter bank may not have the best performance for the echo cancellation.
The existence of delay in sub-band echo cancellation limits it applications. With speech transmission, it is always beneficial to the users if the delay introduced in the echo cancellation is zero, thus minimizing the end-to-end transmission delay of the speech.
It is an object of the present invention to provide a structure which will remove all delays from sub-band echo cancellation activities.
SUMMARY OF THE INVENTION
According to the present invention, a zero-delay structure has been developed such that the delay introduced in sub-band decomposition and reconstruction is eliminated without sacrificing echo cancellation performance. This structure can be used in combination with any sub-band-based echo canceller or wavelet echo canceller.


REFERENCES:
patent: 5329587 (1994-07-01), Morgan et al.
patent: 5610919 (1997-03-01), Willard et al.
patent: 5774561 (1998-06-01), Nakagawa et al.
patent: 0739102 (1996-10-01), None
INSPEC abstract of non-patent literature: “Delayless subband active noise control” by Thi et al in p. 181-4 vol. 1, IEEE, New York, NY, USA 1993, 5 volumes.
INSPEC abstract of Journal article: “Minimum mean-square error echo cancellation and equalization for digital subscriber line transmission. I. Theory and computation” by Lin, D.W. in IEEE Transactions on Communications vol. 38, No. 1 p. 31-8, Jan. 1990.

Affiliated with

Also associated with

Rating

Zero-delay structure for sub-band echo cancellation does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Zero-delay structure for sub-band echo cancellation, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Zero-delay structure for sub-band echo cancellation will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3173369