AI People Wiki

Builders & open source

Yao Qian

Principal Researcher, Microsoft

Yao Qian is a Principal Researcher in the Speech and Dialog Research Group at Microsoft, with research on spoken language processing and, most recently, large-scale speech-language representation learning. Qian is a co-author of WavLM, a Microsoft model for self-supervised pre-training for full-stack speech processing, and of UniSpeech-SAT, which adds speaker-aware pre-training to universal speech representation learning.12345

At a glance

Current role
Principal Researcher
Company
Microsoft
Sector
Company
Work listed
2
Sources cited
8
Profile reviewed

Research

Qian's research is on spoken language processing, most recently large-scale speech-language representation learning, in the Speech and Dialog Research Group at Microsoft. A 2012 Microsoft Research publication, "Tip Tap Tones" (MobileHCI 2012), on mobile microtraining of Mandarin sounds, received an honorable mention award.16

In 2021 Qian co-authored a series of Microsoft speech pre-training papers: UniSpeech, which learns speech representations from labeled and unlabeled data; UniSpeech-SAT, which adds speaker-aware pre-training; SpeechT5, a unified encoder-decoder model for spoken language processing; and WavLM, for self-supervised pre-training across the full stack of speech tasks.7482

Professional service

Qian is a senior member of IEEE and a member of ISCA, has organized special sessions and chaired sessions at Interspeech, ICASSP and other conferences, and served as secretary of the ISCA SIG-CSLP from 2015 to 2018 and of SIG-MLSLP since 2020. A Speech Communication journal paper from work by an intern Qian mentored received the journal's best paper award for 2015 to 2020.1

Work and contributions

  • WavLM Published research

    Co-author of WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing; Microsoft Research's publication page for the paper links Qian's Microsoft Research profile.23

  • UniSpeech-SAT Published research

    Co-author of UniSpeech-SAT: Universal Speech Representation Learning with Speaker Aware Pre-Training (2021); Microsoft Research's publication page for the paper links Qian's profile.45

Timeline

  1. Co-authored "Tip Tap Tones - Mobile Microtraining of Mandarin Sounds", which received an honorable mention at MobileHCI 2012.6
  2. Began serving as secretary of the ISCA special interest group SIG-CSLP, through 2018.1
  3. Began serving as secretary of the ISCA special interest group SIG-MLSLP.1
  4. Co-authored UniSpeech, on unified speech representation learning with labeled and unlabeled data.7
  5. Co-authored UniSpeech-SAT, on speaker-aware speech pre-training.4
  6. Co-authored SpeechT5, a unified-modal encoder-decoder pre-training model for spoken language.8
  7. Co-authored WavLM, on large-scale self-supervised pre-training for full-stack speech processing.2

Where they work

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

8 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. Yao Qian — Microsoft Research profileMicrosoft Research · Official page · accessed Oct 21, 2026
  2. WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing (arXiv 2110.13900)arXiv · Paper · published Oct 26, 2021 · accessed Oct 21, 2026
  3. WavLM — Microsoft Research publication pageMicrosoft Research · Official page · published Jul 2022 · accessed Oct 21, 2026
  4. UniSpeech-SAT: Universal Speech Representation Learning with Speaker Aware Pre-Training (arXiv 2110.05752)arXiv · Paper · published Oct 12, 2021 · accessed Oct 21, 2026
  5. UniSpeech-SAT — Microsoft Research publication pageMicrosoft Research · Official page · accessed Oct 21, 2026
  6. Yao Qian — publications on Microsoft ResearchMicrosoft Research · Official page · accessed Oct 21, 2026
  7. UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data (arXiv 2101.07597)arXiv · Paper · published Jan 19, 2021 · accessed Oct 21, 2026
  8. SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing (arXiv 2110.07205)arXiv · Paper · published Oct 14, 2021 · accessed Oct 21, 2026