Jean-Marc Valin, Ph.D.

email: jmvalin@jmvalin.ca

website: https://jmvalin.ca/

Profile

I hold a Ph.D. in Electrical Engineering with fifteen years of experience in signal processing, machine learning, and their applications. I have started and contributed to multiple open source projects, including the Speex, Opus and AV1 codecs, which are now deployed in billions of devices. I enjoy environments that provide me with challenging problems and where I can perform applied research in parallel with development.

Work Experience

Senior Staff Research Scientist — Google

Apr. 2024 –

  • Audio ML

Principal Applied Scientist — Amazon Web Services

Apr. 2019 – Mar. 2024

  • Deep packet loss concealment and deep redundancy (DRED) for the Opus codec
  • Standardization effort for Opus extension mechanism
  • Speech enhancement and echo cancellation

Senior Staff Research Scientist — Mozilla

Jul. 2011 – Apr. 2019

  • Standardizing the Opus speech and audio codec at IETF
  • Designing coding tools for the Daala and AV1 video codecs
  • Developing deep learning algorithms for speech enhancement and text proofreading

Software Lead Architect — Octasic

Jul. 2008 – Jun. 2011

  • Designing efficient DSP algorithms for line echo cancellation, packet loss concealment and speech bandwidth expansion
  • Managing the instruction set architecture (ISA) for an asynchronous DSP architecture
  • Designing a DSP's arithmetic logic unit (ALU) and instruction decoder

Post-Doctoral Fellow — CSIRO ICT Centre

Aug. 2005 – Jun. 2008

  • Designing the CELT ultra low-delay audio compression algorithm
  • Creating an efficient stereo acoustic echo cancellation algorithm
  • Writing a SIP-based high-quality audio/video communication application
  • Designing a 3D sound source localization system based on a microphone array

Acoustic Modelling Designer — Infospace Speech Solutions

Jan 2001 – Dec 2001

  • Improving the voice activity detection robustness to noise
  • Speeding up the company's speech recognition engine by a factor of three

Education

Ph.D. Electrical Engineering, University of Sherbrooke

Jan. 2002 – Aug. 2005

Thesis: "Auditory System For a Mobile Robot" (mobile robotics lab)
Internship at Kyoto University Audio Media Processing Group, Fall 2004
NSERC (ES B) and FQRNT (B2) scholarships

M.Sc.A. Electrical Engineering, University of Sherbrooke

Jan. 2000 – Dec. 2001

Project: "Bandwidth extension of narrowband speech" (speech and audio research group)
NSERC (ES A) scholarship

B.Eng. Electrical Engineering, University of Sherbrooke

Sep. 1995 – Dec. 1999

Co-op program (four internships)
Exchange program at University of Connecticut, Fall 1997

Skills

Expertise

speech and audio coding, deep learning, speech enhancement, acoustic and line echo cancellation, speech recognition, fixed-point signal processing, DSP architectures

Programming Languages

C, C++, Matlab/Octave, Python (Keras, Numpy)

CPU/DSP architectures

x86, ARM, Blackfin, Octasic Vocallo

Other Professional Activities

Free/Open-source software

  • Co-author and maintainer of the Opus speech and audio codec
  • Contributor to the AV1 free video codec
  • Co-author of the Daala video codec
  • Author of the Speex speech codec

Internet Engineering Task Force (IETF)

  • Co-author of 8 IETF RFCs, including RFC 6716 standardizing the Opus codec
  • Co-chair of a Birds of a Feather (BoF) session to form the codec working group

Scientific journals and conferences

  • Author of 10 journal papers and more than 50 conference papers
  • Reviewer for multiple conferences and journals