Multichannel Acoustic Echo Cancellation

The methods for acoustic echo cancellation remove certain unwanted signal components (the echos) from desired signals. The echos come from an electro-acoustic coupling with directly accessible sources. Since these disturbing sources act as reference signals, echo cancellation allows a high interference rejection if suitable methods are employed. For that purpose, the echo path(s) is/are modeled using an adaptive system. By this system the echo(s) is/are compensated. As the echo paths vary over time, the system parameters are continuously adapted.

Multichannel acoustic echo cancellation (Figure 1) is crucial for hands-free and full-duplex communication systems with multichannel sound reproduction (e.g., stereo or 5.1 channel - surround systems). Applications can be found, e.g., in home entertainment, virtual reality (e.g. games, simulations, training), or advanced teleconferencing and teleteaching. However, due to the often very ill-conditioned nature of the underlying least-squares problem in the multichannel case, even for only two input channels (i.e., stereo), most adaptive algorithms are very complex and/or show poor convergence behaviour. Only few algorithms provide acceptable properties in a multi-loudspeaker environment for real-world conditions.

Our Contributions

During our studies on multichannel acoustic echo cancellation, it turned out that multichannel algorithms in the frequency domain are especially well suited for input signals which are not only auto-correlated but also highly cross-correlated among the channels [B03-1,C01-3,C02-1].

However, most frequency-domain algorithms, as they are well known from the single-channel case, are derived from existing time-domain algorithms and are based on different heuristic strategies which cannot easily be generalized to the multichannel case.

In [B03-1,J04-1] we present a new rigorous derivation of a whole class of multichannel adaptive filtering algorithms in the frequency domain based on a recursive least-squares criterion. An analysis of this multichannel algorithm shows that the mean-squared error convergence is independent of the input signal statistics. A useful approximation provides interesting links between both, some well-known algorithms for the single-channel case, and new and extended single- and multichannel algorithms. We also give design rules for important parameters to optimize the performance in practice.

Due to the rigorous approach, the proposed framework inherently takes the coherence between all input signal channels into account, while the computational complexity is kept low by introducing several new techniques, such as a robust recursive Kalman gain computation in the frequency domain and efficient fast Fourier transform (FFT) computation tailored to overlapping data blocks. Both, simulation results and real-time performance for multichannel acoustic echo cancellation show the high efficiency of the approach.

The adaptation performance was further improved, especially during double talk, by explicitly taking into account all fundamental statistical signal properties of the local near-end signals. This was achieved using the generic TRINICON framework, as shown in [C08-2],[C08-6].

In addition to the development of efficient multichannel adaptation algorithms, suitable inaudible preprocessing of the reproduction signals for partial decorrelation is known as another key approach. In [C07-2] we presented a novel efficient perceptually motivated convergence enhancement which is suitable for surround sound reproduction. Experimental evaluation by listening tests showed superior perceptual quality without sacrificing convergence speed.

We have implemented a fully scalable multi-threaded software which allows, e.g., stereo or 5.1 channel acoustic echo cancellation in real-time on a regular PC platform [C01-1,B03-1,C01-3,C02-1]. To handle the double-talk case we also developed improved double-talk detection schemes. Moreover, for all proposed algorithms exist extended versions derived using robust statistics (e.g., [C03-9,J05-2]).

The generalization of the frequency-domain algorithms also leads to a novel sub-class of efficient and very fast converging adaptive filters, which we call extended multidelay filter (EMDF) [C03-4,C03-9,J05-2] and additionally takes into account inter-partition correlations. It is tailored to very long adaptive filters and highly autocorrelated input signals as they arise, e.g., in high-quality full-duplex audio applications. The approach exhibits good tracking capabilities of the signal statistics and very low delay. Moreover, it is shown that the low order of computational complexity of the conventional frequency-domain adaptive algorithms can be maintained thanks to efficient realizations. The algorithm allows a tradeoff between the well-known multidelay filter (MDF) and the recursive least-squares (RLS) algorithm. It is also well suited for an efficient generalization to the multichannel case.

Advanced wave field synthesis-based systems (which eliminate the 'sweet spot') using both, a microphone array, and also a loudspeaker array, present one of the ultimate challenges in multichannel research.
Hence, in recent years we have developed promising novel and highly efficient approaches to multichannel acoustic echo cancellation for general wave field synthesis-based systems, such as wave-domain adaptive filtering and source-domain adaptive filtering. These new adaptive filtering frameworks can be also used for several other large-scale adaptive MIMO systems with a high number of channels.
More details on Wave-Domain Adaptive Filtering (WDAF) and related publications can be found here.

For more information contact Herbert Buchner.

Related Publications

[C11-2]	K. Helwani, S. Spors, and H. Buchner, "Spatio-temporal signal preprocessing for multichannel acoustic echo cancellation," Proc. IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP), Prague, Czech Rep, May 2011.

[C10-5]	K. Helwani, H. Buchner, and S. Spors, "On the robust and efficient computation of the Kalman gain for multichannel adaptive filtering with application to acoustic echo cancellation," Proc. Asilomar Conf. on Signals, Systems, and Computers, Pacific Grove, CA, USA, Nov. 2010.

[C10-1]	K. Helwani, H. Buchner, and S. Spors, "Source-domain adaptive filtering for MIMO systems with application to acoustic echo cancellation," Proc. IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP), Dallas, TX, USA, Mar. 2010.

[C09-2]	S. Spors, H. Buchner, and K. Helwani, "Block-based multichannel transform-domain adaptive filtering," Proc. European Signal Processing Conference (EUSIPCO), Glasgow, Scotland, Aug. 2009.

[C08-8]	H. Buchner and S. Spors, "A general derivation of wave-domain adaptive filtering and application to acoustic echo cancellation," Proc. Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, CA, USA, Oct. 2008.

[C08-6]	H. Buchner, "Acoustic echo cancellation for multiple reproduction channels: From first principles to real-time solutions," Proc. ITG Conf. on Speech Communication, Aachen, Germany, Oct. 2008. Best Paper Award.

[C08-4]	S. Spors and H. Buchner, "Multichannel transform domain adaptive filtering: A two stage approach and illustration for acoustic echo cancellation," Conf. Rec. Intl. Workshop on Acoustic Echo and Noise Control (IWAENC), Seattle, WA, USA, Sept. 2008.

[C08-2]	H. Buchner and W. Kellermann, "A fundamental relation between blind and supervised adaptive filtering illustrated for blind source separation and acoustic echo cancellation," Conf. Rec. Joint Workshop on Hands-Free Speech Communication and Microphone Arrays (HSCMA), Trento, Italy, May 2008. Invited presentation.

[C07-2]	J. Herre, H. Buchner, and W. Kellermann, "Acoustic echo cancellation for surround sound using perceptually motivated convergence enhancement," Proc. IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP), Honolulu, HI, USA, April 2007.

[J05-2]	H. Buchner, J. Benesty, T. Gaensler, and W. Kellermann, "Robust Extended Multidelay Filter and Double-talk Detector for Acoustic Echo Cancellation," IEEE Transactions on Audio, Speech, and Language Processing, vol. 14, no. 5, pp. 1633-1644, Sept. 2006.

[C04-3]	H. Buchner, S. Spors, and W. Kellermann, "Full-Duplex Systems for Sound Field Recording and Auralization Based on Wave Field Synthesis," Conf. Rec. AES 116th Convention, Berlin, Germany, May 2004.

[J04-1]	H. Buchner, J. Benesty, and W. Kellermann, "Generalized Multichannel Frequency-Domain Adaptive Filtering: Efficient Realization and Application to Hands-Free Speech Communication," Signal Processing, vol. 85, no. 3, pp. 549-570, March 2005.

[C03-9]	H. Buchner, J. Benesty, T. Gaensler, and W. Kellermann, "An Outlier-Robust Extended Multidelay Filter with Application to Acoustic Echo Cancellation," Conf. Rec. IEEE Intl. Workshop on Acoustic Echo and Noise Control (IWAENC), Kyoto, Japan, pp. 19-22, September 2003.

[B03-1]	H. Buchner, J. Benesty, and W. Kellermann, "Multichannel frequency-domain adaptive filtering with application to acoustic echo cancellation," In J.Benesty and Y.Huang (eds.), Adaptive signal processing: Application to real-world problems, Springer-Verlag, Berlin/Heidelberg, pp. 95-128, Feb 2003.

[C03-4]	H. Buchner, J. Benesty, and W. Kellermann, "An Extended Multidelay Filter: Fast Low-Delay Algorithms for Very High-Order Adaptive Systems," Proc. IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP), Hong Kong, China, vol. 5, pp. 385-388, April 2003.

[C02-2]	H. Buchner, S. Spors, W. Kellermann, and R. Rabenstein, "Full-Duplex Communication Systems with Loudspeaker Arrays and Microphone Arrays," Proc. IEEE Int. Conference on Multimedia and Expo (ICME), Special session on coding and transmission formats for 3D audio, Lausanne, Switzerland, August 2002.

[C02-1]	H. Buchner and W. Kellermann, "Improved Kalman Gain Computation for Multichannel Frequency-Domain Adaptive Filtering and Application to Acoustic Echo Cancellation," Proc. IEEE Int. Conference on Acoustics, Speech, and Signal Processing (ICASSP), Orlando, FL, USA, pp. 1909-1912, May 2002.

[C01-4]	H. Buchner and W. Kellermann, "An Acoustic Human-Machine Interface with Multi-Channel Sound Reproduction," Conf. Rec. IEEE Intl. Workshop on Multimedia Signal Processing (MMSP), Cannes, France, pp. 359-364, October 2001.

[C01-3]	H. Buchner and W. Kellermann, "Acoustic Echo Cancellation for Two and More Reproduction Channels," Conf. Rec. IEEE Intl. Workshop on Acoustic Echo and Noise Control (IWAENC), Darmstadt, Germany, pp. 99-102, September 2001. Best Student Paper Award.

[C01-1]	H. Buchner, W. Herbordt, and W. Kellermann, "An Efficient Combination of Multi-Channel Acoustic Echo Cancellation With a Beamforming Microphone Array," Conf. Rec. 1st IEEE Intl. Workshop on Hands-Free Speech Communication (HSC), Kyoto, Japan, pp. 55-58, April 2001.

[R00-1]	H. Buchner, "Hands-Free Man-Machine Audio Interfaces With Automatic Speech Recognition Systems And Multi-Channel Sound Reproduction," Diploma thesis XE689, University of Erlangen-Nuremberg, Erlangen, Germany, 174 pages, June 2000.

Latest update: Jun 2011