Probabilistic Inference of
Speech Signals from Phaseless Spectrograms
University of Toronto
In this work, we describe a generative model of time-domain speech signals
and their spectrograms, and show how an efficient optimizer can be used to
find the maximum a posteriori speech signal, given the spectrogram.
Related publication
Kannan Achan, Sam T.
Roweis, Brendan J. Frey, Probabilistic Inference of Speech Signals
from Phaseless Spectrograms, to appear in Neural Information Processing Systems
(NIPS) 16. Dec 2003.
Under construction