New Concept Service for the Mobile Era Using Speech Technologies

In this paper, we describe new concept services based on speech processing technologies for the new digital/mobile era called a ubiquitous society. First, we propose a compact and noise robust embedded speech recognition middleware implemented on microprocessors aiming for sophisticated HMIs (Human Machine Interfaces) of car information systems. The compactness is essential for embedded systems because there are strict restrictions of CPU (Central Processing Unit) power and available memory capacities. Second, we report a promising multi-language interpreter named “Mobilingual TM ” based on speech recognition. Our research activities aim to realize sophisticated and human-centered intelligent HMIs that will be absolutely necessary in the ubiquitous social environment. For the embedded speech middleware, we propose first, a noise robust and compact Spectral Subtraction(SS) method after exhausting evaluation stages using real speech data recorded at car running environments. Next, we propose very novel memory assignment of acoustic models based on the product codes or sub-vector quantization technique resulting on 1 fourth memory reduction for the 2000-word vocabulary.