PeopleofSuperintelligence

Researcher

Lukáš Burget

Also known as Lukas Burget

Lukáš Burget is a speech researcher at Brno University of Technology whose work covers speech recognition, speaker recognition and diarization. After a PhD at the Faculty of Information Technology in 2004, Burget co-founded the speech company Phonexia in 2006 and joined the Johns Hopkins workshops of 2008 and 2009 from which i-vectors and the Kaldi toolkit emerged. In 2010, the recurrent neural network language model paper with Tomáš Mikolov and others carried Burget's name among its authors, and the VBx diarization paper followed in 2020. Interspeech 2021 in Brno had Burget as lead technical chair, and AMiner's 2021 AI 2000 analysis ranked Burget among the 100 most cited scholars in speech recognition. In 2025, Burget led one of four research teams at JSALT, the Johns Hopkins University workshop hosted in Brno, and is now an associate professor at the Faculty of Information Technology and scientific director of the BUT Speech@FIT group.

Now at
Faculty of Information Technology, Brno University of Technology
Badges

Timeline

  1. 2004

    PhD at Brno University of Technology

    Burget defends a PhD thesis in November 2004 at the Faculty of Information Technology of Brno University of Technology, under the supervision of Jan Černocký.

  2. 2006–2025

    Co-founding Phonexia

    Phonexia is founded in 2006 as a spin-off of the Faculty of Information Technology in Brno, and Burget is one of its founders. In December 2025, the company's sale to the South Korean investment fund Crescendo Equity Partners is announced, and Burget says the founders sought a buyer who would maintain ties with faculty research.

  3. 2008–2011

    Johns Hopkins workshops and Kaldi

    The 2008 Johns Hopkins workshop on robust speaker recognition over varying channels, where i-vectors originate, counts Burget on its team. In 2009, Burget joins the Johns Hopkins workshop where the Kaldi speech recognition toolkit begins, and Kaldi's authors say none of it would have happened without Burget. The Brno workshop of August 2010 builds the toolkit with Burget among the participants, and Kaldi is released in May 2011.

  4. 2010

    A recurrent neural network language model

    In 2010, Burget co-authors the Interspeech paper on a recurrent neural network based language model with Tomáš Mikolov, Martin Karafiát, Jan Černocký and Sanjeev Khudanpur. A mixture of several recurrent network language models cuts perplexity by around 50% against a backoff language model, and word error rate on the Wall Street Journal task drops by around 18%.

  5. 2020

    VBx speaker diarization

    A paper by Federico Landini, Ján Profant, Mireia Diez and Burget, submitted in December 2020, compares the VBx diarization method, a Bayesian hidden Markov model over sequences of x-vectors, with other approaches and finds superior performance on the CALLHOME, AMI and DIHARDII datasets. The authors also release recipes for training x-vector extractors and for VBx diarization.

  6. 2021

    Interspeech 2021 lead technical chair

    The Czech Republic hosts the Interspeech conference for the first time in 2021, in Brno, and Burget is lead technical chair. In 2021, AMiner's AI 2000 Most Influential Scholars analysis ranks Burget among the 100 most cited scholars in the area of speech recognition.

  7. 2024–2025

    DiCoW and JSALT 2025

    In December 2024, Burget co-authors DiCoW, which extends the pre-trained Whisper model with diarization labels for target-speaker speech recognition. In 2025, the 32nd JSALT workshop, organized by Johns Hopkins University, runs at the Faculty of Information Technology in Brno, and Burget leads one of its four research teams.

  8. As of October 2026

    Associate professor in Brno

    Burget is an associate professor in the Department of Computer Graphics and Multimedia at the Faculty of Information Technology of Brno University of Technology and scientific director of the BUT Speech@FIT group. The group, which applies machine learning to speech data mining, is led by Burget together with Jan Černocký and Hynek Heřmanský.

Career

  1. NowFaculty of Information Technology, Brno University of TechnologyAssociate professor
  2. NowBrno University of TechnologyScientific director
  3. 2006–2025PhonexiaFounder

Education

  1. 2004Brno University of TechnologyPhD

From the biography.

  1. 11% of AI doctorate holders from 2010 to 2016 founded or led a company or lab by 10 years after graduationIncludes Lukáš Burget. We estimate how often AI doctorate holders founded or led a company or lab by 10 years after graduation, how waiting times compare between founders and top jobs, and which universities trained the most.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Brno University of Technology. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Brno University of Technology. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Brno University of Technology. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Lukáš Burget yet. Send us a free-licensed photo.

  1. Thesis Details fit.vut.cz. Checked 21 October 2026.
  2. Phonexia becomes part of the portfolio of investment fund Crescendo Equity Partners fit.vut.cz. Checked 21 October 2026.
  3. Community | BUT Speech@FIT speech.fit.vut.cz. Checked 21 October 2026.
  4. Kaldi: History of the Kaldi project kaldi-asr.org. Checked 21 October 2026.
  5. ISCA Archive - Recurrent neural network based language model isca-archive.org. Checked 21 October 2026.
  6. Bayesian HMM clustering of x-vector sequences (VBx) in speaker diarization: theory, implementation and analysis on standard tasks arxiv.org. Checked 21 October 2026.
  7. The world's leaders in speech processing are heading to Brno fit.vut.cz. Checked 21 October 2026.
  8. Organising Committee | INTERSPEECH 2021 interspeech2021.org. Checked 21 October 2026.
  9. BUT Speech@FIT speech.fit.vut.cz. Checked 21 October 2026.
  10. DiCoW: Diarization-Conditioned Whisper for Target Speaker Automatic Speech Recognition arxiv.org. Checked 21 October 2026.
  11. Brno to Become a Global Hub for Speech Technologies and Large Language Models in Summer 2025. FIT BUT is Hosting the Prestigious International Research Workshop JSALT 2025 fit.vut.cz. Checked 21 October 2026.
  12. doc. Ing. Lukáš Burget, Ph.D. - Contact – BUT vut.cz. Checked 21 October 2026.
  13. The Czech Republic is a global powerhouse in AI research. The international workshop JSALT 2025 at FIT VUT proves it fit.vut.cz. Checked 21 October 2026.