| A comparative evaluation of variance flooring techniques on HMM-based speaker verification |
Title: A comparative evaluation of variance flooring techniques on HMM-based speaker verification
Author(s): H. Melin, Johan Koolwaaij, J. Lindberg, & F. Bimbot
Reference: Proceedings of the International Conference on Spoken Language Processing (ICSLP'98), Vol. 5. pp. 1903-1906
Keywords: Speaker Recognition
There is a PostScript version (321662 bytes) available.
The problem of how to train variance parameters on scarce data
is addressed in the context of text-dependent HMM-based,
automatic speaker verification. Three variations of variance
flooring are explored as a means to prevent over-fitting.
With the best-performing one, the floor to a variance vector of a client
model is proportional to the correpsonding variance vector in
a non-client multi-speaker model. It is also found
that adapting the means and mixture weights from the non-client model
while keeping the variances constant works comparably to
variance flooring and is much simpler. Comparisons are
made on the large telephone quality corpora.
Error processing SSI file