Efficient Codebooks for Fast and Accurate Low Resource ASR Systems
Article : Articles dans des revues internationales ou nationales avec comité de lecture
Today, speech interfaces have become widely employed in mobile devices, thus recognition speed and resource consumption are becoming new metrics of Automatic Speech
Recognition (ASR) performance.
For ASR systems using continuous Hidden Markov Models (HMMs), the computa-tion
of the state likelihood is one of the most time consuming parts. In this paper, we propose
novel multi-level Gaussian selection techniques to reduce the cost of state likelihood
computation. These methods are based on original and e cient codebooks. The proposed ffi
algorithms are evaluated within the framework of a large vocabulary continuous speech
recognition task.