Speaker Technology Group

From cslt Wiki
Jump to: navigation, search

Welcome to Speaker Recognition team in CSLT, RIIT, Tsinghua University.

Research focus:

1. Robust speaker recognition

We are interested in VPR approaches in adverse circumstances, including high noise, cross channel, truncated speech, emotional speech, aged speech.

2. Short utterance speaker recognition

People are reluctant to speak too much in VPR, but short utterances are highly challenging. We are interested in approaches particularly to speaker verification with speech signals as short as 1-5 seconds.

3. Multi-lingual speaker recognition

How to use multiple language data can be used collaboratively in SID is an open question, particularly, using the data of a dominant language to boost the performance for low-resource languages is highly interesting.



Weekly progress