Recent Development of WFST-Based Speech Recognition Decoder
Paul R. Dixon, Tasuku Oonishi, Koji Iwano, Sadaoki Furui · Institutional Repositories DataBase (IRDB) · 2009
In this paper we present an overview of the Tokyo Tech Transducer-based Decoder T3 (pronounced tee-cubed). There is a high level overview of the engine's design and features which is accompanied by a more detailed description of the features that are unique to our engine. These include the ability to perform acoustic computations on a graphics card and generalized fast on-the-fly composition and optimization algorithms. We describe voice activity detection functionality recently added to the engine and finally results are presented which show the engine achieving very high recognition throughput at a high recognition accuracy.