[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Rationale for Critical Bands

I checked some books, and come back to my original statement that the "critical bandwidth" is a simplification of the concept of a psychophysical "auditory filter". The former typically ignores level dependence (or assumes level independence) and ignores shape, while the latter includes shape and the dependence of shape, and hence of bandwidth, on level.

I found a citation to this paper to support a statement that the CB is known to be level dependent:
title={{Formulae describing frequency selectivity as a function of frequency and level, and their use in calculating excitation patterns}},
  author={Moore, B.C.J. and Glasberg, B.R.},
  journal={Hearing Research},

I don't have the paper; the abstract talks about level dependence of the filter, not of the bandwidth per se. Probably that's in the paper, as these authors do commonly characterize their filter shapes with an equivalent rectangular bandwidth.

The support cited for CB being level independent, e.g. in some of the papers that Martin linked, was Ehret 1985, a bit earlier. So probably this was the timeframe when the level dependence of a psychophysical CB was being recognized, via careful study of auditory filters. In his 1997 book, Ehret says critical ratios are "less intensity-dependent in many, mainly higher-order neurons of the CN as compared to in the AN", citing his 1983 and 1984 papers.
He probably says more later in the book about IC, but I can't see those pages. Perhaps it's still the case that in IC the CB is level independent? It would be very interesting if so. It's not clear to me how measurement in IC (or CN or AN for that matter) relate to psychophysical measures; maybe someone can say?