Speech decoloration based on the product-of-filters model

We present a single-channel speech decoloration method based on a recently proposed generative product-of-filters (PoF) model. We take a spectral approach and attempt to learn the magnitude response of the actual coloration filter, given only the degraded speech signal. Experiments on synthetic data...

Full description

Saved in:
Bibliographic Details
Main Authors: Dawen Liang, Ellis, Daniel P. W., Hoffman, Matthew D., Mysore, Gautham J.
Format: Conference Proceeding
Language:English
Subjects:
audio
Bayesian modeling
decoloration
Graphical models
Radio frequency
Reverberation
Speech
Speech enhancement
variational inference
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We present a single-channel speech decoloration method based on a recently proposed generative product-of-filters (PoF) model. We take a spectral approach and attempt to learn the magnitude response of the actual coloration filter, given only the degraded speech signal. Experiments on synthetic data demonstrate that the proposed method effectively captures both coarse and fine structure of the coloration filter. On real recordings, we find that simply subtracting the learned coloration filter from the log-spectra yields promising decoloration results.
ISSN:1520-6149
2379-190X
DOI:10.1109/ICASSP.2014.6854030