[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
phoneme recognition and stimulus length
I wonder how human performance in phoneme recognition depends on the
stimulus length. I'm thinking of an experiment something like "name the
last phoneme you hear", and the length of the sound excerpt played would
vary between, say, 10ms and a couple of seconds (with the end always fit
to the end of a phoneme). Since I'm interested in
the importancy of long-term adaptation, nonsense worlds should be used to
exclude linguistic help. Does anyone know about such or similar
P.S.: My special interest is whether information earlier than the "magic"
250ms is important, and how much (I would like to draw some conclusions
for automatic speech recognition).
Hungarian Academy of Sciences * "Heavy Metal is the
Research Group on Artificial Intelligence * QUINTessence of music"
e-mail: firstname.lastname@example.org *
http://www.inf.u-szeged.hu/~tothl * /Anonymous/