"TalkPrinting": Improving speaker recognition by modeling stylistic features

Sachin Kajarekar, Kemal Sönmez, Luciana Ferrer, Venkata Gadde, Anand Venkataraman, Elizabeth Shriberg, Andreas Stolcke, Harry Bratt

Research output: Contribution to journalArticle

Abstract

Automatic speaker recognition is an important technology for intelligence gathering, law enforcement, and audio mining. Conventional speaker recognition systems, which are based on independent short-term spectral samples, suffer from a lack of noise robustness and are unable to model a speaker's idiosyncratic stylistic features. This paper describes "TalkPrinting", a program of research aimed at adding such stylistic features to conventional systems. Results on three preliminary systems based on stylistic features demonstrate that (1) the new features alone carry significant speaker information; (2) they also carry significant complementary information compared to the conventional features; and (3) they provide increasing improvements in performance with increasing test durations.

Original languageEnglish (US)
Pages (from-to)350-354
Number of pages5
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2665
StatePublished - Dec 1 2003
Externally publishedYes

    Fingerprint

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this