At 15:11 +0000 30-11-05, bulbul20 wrote:
>You are totally correct about 5000 features being aprioristically
>correlated, but I do not quite understand what do you mean by putting
>them into bins?
for instance, if they are power spectral densities along a near-continuous frequency scale, one could consider using an LTAS instead. But if they are samples of a waveform, this will not be possible.
Institute of Phonetic Sciences, University of Amsterdam
Herengracht 338, 1016CG Amsterdam, The Netherlands