A TIC-MRF approach to speaker clustering on i-vectors yields relative DER reductions of 43.22% on CRSS-PLTL and 29.37%/9.21% on two AMI meetings versus cosine K-means and movMF baselines.
CRSS-LDNN Long-duration naturalistic noise corpus containing multi-layer noise recordings for robust speech processing
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Toeplitz Inverse Covariance based Robust Speaker Clustering for Naturalistic Audio Streams
A TIC-MRF approach to speaker clustering on i-vectors yields relative DER reductions of 43.22% on CRSS-PLTL and 29.37%/9.21% on two AMI meetings versus cosine K-means and movMF baselines.