Daniel Povey

Creator of Kaldi,IEEE Fellow,Chief Speech Scientist of Xiaomi Group

Dr Daniel Povey graduated from the University of Cambridge, UK, and worked at IBM and Microsoft.He joined Johns Hopkins University in 2012 as an Associate Professor in the Centre for Language and Speech Processing.He joined Xiaomi in October 2019 as the Group Chief Scientist for Speech. In December 2022, Daniel Povey was selected as an IEEE Fellow for his outstanding contributions to speech recognition and acoustic modelling. Dr Daniel Povey is the author and main maintainer of Kaldi, a well-known open source speech recognition tool, and is known as the "Father of Kaldi" in the industry.The Kaldi project is highly influential in both academia and industry, and has been widely used in various research institutes and well-known domestic and international companies, which has strongly contributed to the intelligent speech Daniel Povey has made a lot of contributions to the technology of intelligent speech processing. He was the first to apply the sequence discriminative training method to speech recognition, and the LF-MMI modelling method he proposed is still the standard technology commonly used in commercial speech recognition systems. He was also a key leader in the use of deep learning for speech recognition, popularising time-delay neural networks in speech recognition, together with LF-MMI training, as the best combination for general use in academia and industry in 2015-2020. He is also a key leader in deep learning in voice recognition, proposing the X-vectors method, which is the first successful and universally applied deep learning speaker recognition method, widely used in industry. He has published more than 200 papers in top international conferences and journals, with nearly 46,000 Google Scholar citations, h-index 66, and serves as an Associate Lifetime Editor of Signal Processing Letters.