DMCA
Detection Of Target Speakers In Audio Databases (1999)
| Venue: | IN PROCEEDINGS OF ICASSP 99 |
| Citations: | 9 - 4 self |
BibTeX
@INPROCEEDINGS{Magrin-Chagnolleau99detectionof,
author = {Ivan Magrin-Chagnolleau and Aaron E. Rosenberg and S. Parthasarathy},
title = {Detection Of Target Speakers In Audio Databases},
booktitle = {IN PROCEEDINGS OF ICASSP 99},
year = {1999},
pages = {821--824},
publisher = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
The problem of speaker detection in audio databases is addressed in this paper. Gaussian mixture modeling is used to build target speaker and background models. A detection algorithm based on a likelihood ratio calculation is applied to estimate target speaker segments. Evaluation procedures are defined in detail for this task. Results are given for different subsets of the HUB4 broadcast news database. For one target speaker, with the data restricted to high quality speech segments, the segment miss rate is approximately 7%. For unrestricted data, the segment miss rate is approximately 27%. In both cases the segment false alarm rate is 4 or 5 per hour. For two target speakers with unrestricted data, the segment miss rate is approximately 63% with about 27 segment false alarms per hour. The decrease in performance for two target speakers is largely associated with short speech segments in the two target speaker test data which are undetectable in the current configuration of the detection algorithm.
Keyphrases
target speaker audio database segment miss rate detection algorithm unrestricted data likelihood ratio calculation background model hub4 broadcast news database gaussian mixture modeling evaluation procedure different subset current configuration speaker detection high quality speech segment segment false alarm rate segment false alarm short speech segment target speaker segment target speaker test data
