DMCA
Automatic speaker recognition using gaussian mixture speaker models (1995)
| Venue: | The Lincoln Laboratory Journal |
| Citations: | 41 - 1 self |
BibTeX
@ARTICLE{Reynolds95automaticspeaker,
author = {Douglas A. Reynolds},
title = {Automatic speaker recognition using gaussian mixture speaker models},
journal = {The Lincoln Laboratory Journal},
year = {1995},
pages = {173--192}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
• Speech conveys several levels ofinformation. On a primary level, speech conveys the words or message being spoken, but on a secondary level, speech also reveals information about the speaker. The Speech Systems Technology group at Lincoln Laboratory has developed and experimented with approaches for automatically recognizing the words being spoken, the language being spoken, and the topic ofa conversation. In this article we present an overview of our research efforts in a fourth area-automatic speaker recognition. We base our approach on a statistical speaker-modeling technique that represents the underlying characteristic sounds ofa person's voice. Using these models, we build speaker recognizers that are computationally inexpensive and capable of recognizing a speaker regardless ofwhat is being said. Performance ofthe systems is evaluated for a wide range ofspeech quality; from clean speech to telephone speech, by using several standard speech corpora. TASKS THAT ARE EASIlY PERFORMED by humans, such as face or speech recognition, prove difficult
Keyphrases
gaussian mixture speaker model automatic speaker recognition topic ofa conversation speech recognition speech conveys several level ofinformation clean speech speaker recognizers speaker regardless research effort speech system technology group performance ofthe system fourth area-automatic speaker recognition easily performed primary level underlying characteristic sound ofa person secondary level statistical speaker-modeling technique wide range ofspeech quality lincoln laboratory telephone speech several standard speech corpus
