Keiichi Tokuda

2024
IEEE JAMES L. FLANAGAN SPEECH AND AUDIO PROCESSING AWARD

Sponsored by Mitsubishi Electric Research Labs (MERL)

“For pioneering contributions to statistical speech synthesis and speech signal processing.”

Keiichi Tokuda made speech synthesis the recognized science it is today. Creating the concept of statistical speech synthesis, he brought a paradigm shift to the field with a technique that underpins practically all current text-to-speech systems. His insight—that a hidden Markov model could generate speech—changed the game. Although questioned initially, it soon proved itself and replaced the then-dominant form of synthesis (concatenative), which is today looked upon as a primitive tool. Tokuda’s technique powers the neural-network approach that is now state of the art and generates synthetic speech that sounds nearly human. Indeed, almost every speech synthesis product on the market—from those of Apple and Facebook to the latest startup application—owes its existence to Tokuda’s work.
An IEEE Fellow, Tokuda is a Professor at the Nagoya Institute of Technology, Nagoya, Japan.