Efficient Block Frequency Domain Filtered-x Applied to Phantom Sound Source Generation
A phantom sound source is a virtual sound image which can be utilized in many applications such as stereo base widening, multimedia, and virtual reality engines. Generation of such a virtual sound image using adaptive filters that employ the filtered-x NLMS algorithm is discussed. Such filters are long and complex for the virtual source to cover the whole audio frequency range. Due to this complexity, real-time implementation on a sample-by-sample basis is not practical and block processing techniques, such as block frequency domain adaptive filters, have to be used. Several versions of the filtered-x algorithm in the frequency domain are derived and the different implementations are compared in terms of convergence, behavior and acoustical performance.
This paper costs $20 for non-members, $5 for AES members and is free for E-Library subscribers.