Combining missing-feature theory, speech enhancement, and speaker-dependent/-independent modeling for speech separation

Ming Ji, Timothy Hazen, James R. Glass

Research output: Contribution to journalArticle

17 Citations (Scopus)

Abstract

This paper considers the separation and recognition of overlapped speech sentences assuming single-channel observation. A system based on a combination of several different techniques is proposed. The system uses a missing-feature approach for improving crosstalk/noise robustness, a Wiener filter for speech enhancement, hidden Markov models for speech reconstruction, and speaker-dependent/-independent modeling for speaker and speech recognition. We develop the system on the Speech Separation Challenge database, involving a task of separating and recognizing two mixing sentences without assuming advanced knowledge about the identity of the speakers nor about the signal-to-noise ratio. The paper is an extended version of a previous conference paper submitted for the challenge.
Original languageEnglish
Pages (from-to)67-76
Number of pages10
JournalComputer Speech & Language
Volume24
Issue number1
DOIs
Publication statusPublished - Jan 2010
EventInterspeech 2006 - ICSLP - Pittsburgh, United States
Duration: 17 Sep 200621 Sep 2006

Fingerprint Dive into the research topics of 'Combining missing-feature theory, speech enhancement, and speaker-dependent/-independent modeling for speech separation'. Together they form a unique fingerprint.

  • Cite this