AES E-Library

AES E-Library

Velocity-Controlled Kalman Filter for an Improved Echo Cancellation with Continuously Moving Microphones

Document Thumbnail

Modern telepresence systems incorporating spatial audio can be realized using multichannel loudspeaker reproduction in combination with close-up microphones attached to the sound sources. With the microphones being tracked in space by external sensors, this setup provides an excellent basis for creating interactive virtual acoustic environments. However, the induced acoustical echo loop has to be handled by a suitable acoustic echo cancellation (AEC) system. A popular state-of-the-art adaptive filter with desirable properties for AEC in multichannel systems is the frequency-domain adaptive Kalman filter (FDKF). Combined with previously proposed enhancements, it shows good performance for minor or abrupt echo path changes but has shortcomings with massive and continuous echo path changes, as caused by moving microphones. This article proposes a velocity-controlled FDKF (VC-FDKF) exploiting the knowledge of the microphone motion for a twofold velocity-dependent contribution to the update step-size. The method has been evaluated in simulations with nonsynthetic recorded measurement data considering different trajectories, velocity profiles, signal types, and loudspeaker setups. Common existing approaches, as the shadow filtering technique, are outperformed by the proposed VC-FDKF in our experiments. Furthermore, two extensions of the proposed technique, namely, a position-dependent gain-and-delay compensation and alternative velocity definitions, are briefly studied.

Authors:
Affiliation:
JAES Volume 72 Issue 1/2 pp. 33-43; January 2024
Publication Date:
Permalink: http://www.aes.org/e-lib/browse.cfm?elib=22376

Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!

This paper costs $33 for non-members and is free for AES members and E-Library subscribers.

Learn more about the AES E-Library

E-Library Location:

DOI:

Start a discussion about this paper!


AES - Audio Engineering Society