Reducing the coherence of loudspeaker signals prior to reproduction solves the non-uniqueness problem of stereophonic acoustic echo cancellation. However, there is a trade-off between the effectiveness of coherence reduction methods and the subjective audio quality of the pre-processed signals. Sub-band-domain coherence reduction methods present the advantage that this tradeoff can be controlled for each sub-band. Among these, the linear periodically time-varying (LPTV) phase modulation method provides a good trade-off, while the frequency shift method, which is also LPTV, achieves the maximum coherence reduction. Here, the performance of these sub-band-domain LPTV methods is evaluated and compared, to provide further insight into their suitability as pre-processors for stereophonic acoustic echo cancellation.
展开▼
