BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:CUED Speech Group Seminars
SUMMARY:Time-domain multi-channel speech separation and ex
 traction - Jisi Zhang\, University of Sheffield
DTSTART;TZID=Europe/London:20210608T120000
DTEND;TZID=Europe/London:20210608T130000
UID:TALK160780AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/160780
DESCRIPTION:*Abstract*: When multiple speakers talk at the sam
 e time\, each utterance is partially or completely
  overlapped by one or more competing utterances. T
 he overlapping speech is challenging to speech tec
 hnologies\, including automatic speech recognition
 \, speaker diarization\, and speaker verification.
  These challenges can be overcome by using speech 
 separation front-ends\, which aim to segregate ind
 ividual source speakers from a mixture signal. Des
 pite the recent progress of single-channel speech 
 separation driven by advances in deep learning\, i
 t still performs poorly in distant microphone scen
 arios where noise and reverberation are involved. 
 This talk focuses on the development of multi-chan
 nel speech separation techniques for separating mi
 xture signals in the distant microphone case. The 
 talk will be split into three parts. The first par
 t introduces an end-to-end neural architecture wit
 h time-domain multi-microphone input. Second\, the
  knowledge of speaker identity is exploited to ext
 end the multi-channel separation system to perform
  a multi-speaker extraction task. Finally\, an uns
 upervised approach is described\, which aims to ap
 plying the end-to-end separation system in situati
 ons where supervised data is hard to collect. The 
 methods are evaluated using simulated data with re
 verberation and ambient noise\, and in terms of si
 gnal enhancement metrics and as front-ends to ASR.
 \n\n*Bio*: Jisi Zhang is a final year PhD student 
 supervised by Professor Jon Barker\, in the Depart
 ment of Computer Science at the University of Shef
 field. He is interested in speech separation\, mul
 ti-channel processing\, and multi-talker speech re
 cognition.\n
LOCATION:Zoom: https://zoom.us/j/95352633552?pwd=RzJVK2UzOG
 ZyNU5mVHd1Y1VPT2tDUT09
CONTACT:Dr Kate Knill
END:VEVENT
END:VCALENDAR