[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: formant filtering
I referred to the level of auditory nerve were even a pure tone is
transmitted over a certain tonotopic range corresponding to nearly one
millimeter of cochlear parition. Otherwise we would hear a pure tone less
loud or not at all.
Yes, the width of that "frequency band" (being a unisono standing traveling
wave) is not necessarily identical with the width of a formant. Even if the
latter is difficult to define, I don't exclude improvement of
intelligibility by means of restriction to the bandwidth down to its
physiologically relevant minimum.
However, I suspect a weak point of psychophysics is its hidden preference
for expected agreement with established theories. I vaguely recall several
cases of astonishment about missing effects of spectral fine-structure.
Perhaps, not a single one was published, of course. Psychophysics also
tends to suffer from lacking options for direct measurement.
What about revisiting the (auto-) correlation idea, I am aware of findings
by Christian Kaernbach and Laurel Demany to be agreed with it. Also,
neither the first (cochlear) spectral analysis nor the putative second
(neural) one is a mathematically ideal one, because of CB, refractory time,
and other peculiartities.
Nonetheless I hope Roy Patterson's intention has been met. He wrote in 1998:
For years, it has been common to hear psychoacousticans and Physiologists
argue about the sorts of model.... The psychoacousticans ... have been
happy to use models in which squaring and autocorrelation are applied
directly to stimulus waves. To physiologists is this an anathema.....Their
models, however, are usually restricted to the cochlea or a single nerve
cell in the brain stem....
> Eckard Blumschein wrote:
>> please don't take it amiss if I admit my guess: A narrower bandwidth at the
>> level of auditory nerve would not improve but on the contrary degradate
>> accuracy of hearing in general.
>Then, psychoacoustic experiments that find that narrower formant bandwidth
>improves intelligibility could help you revisit your theory.
>I think this has already been shown many times, and our colleagues
>(probably Mr. Sivaprasad himself?) could give us some references on
> Laszlo Toth
> Hungarian Academy of Sciences *
> Research Group on Artificial Intelligence * "Failure only begins
> e-mail: firstname.lastname@example.org * when you stop trying"
> http://www.inf.u-szeged.hu/~tothl *