Re: Precedence effect in real systems?

[Christopher Hummersone <c.hummersone@xxxxxxxxxxxx>]

Hi Dick, Pierre,

On 2 Sep 2014, at 01:11, Richard F. Lyon <dicklyon@xxxxxxx> wrote:

> They had another the same year, which reached slightly different conclusions about whether an interaural coherence measure is needed:
> Hummersone, C., Brookes, T., and Mason, R. (2010). “A comparison of computational precedence models
> for source separation in reverberant environments,” in Audio Engineering Society Convention 128.

Perhaps slightly off-topic, but this is an interesting point. The IC formulation that I used, proposed by Faller & Merimaa [1], requires a relatively long time frame (in sound source separation terms) in order to produce a reliable measure. So in practice, this kind of cue selection only helps in the long term. Conversely, Park & Stern [2] note that “early reflections are especially problematical for sound source segregation because they affect the cross-correlation within the same frame as the direct sound wave”. Hence they used an inverse filter to deal with such reflections.

> I wonder why nobody applied these ideas in the subsequent challenges.

It’s on my to-do list!

Best regards,

Chris

[1] Faller, C. & Merimaa, J. (2004), Source localization in complex listening situations: Selection of binaural cues based on interaural coherence, The Journal of the Acoustical Society of America 116, 5, 3075–3089.

[2] Park, H.M. & Stern, R. (2007), Missing feature speech recognition using dereverberation and echo suppression in reverberant environments, in Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), volume 4, 381–384.

P.S. the above-mentioned paper was re-published as a journal paper, after a slight re-work, and is available for free from the publisher: http://www.aes.org/e-lib/browse.cfm?elib=16867.