A Training Method for the Speech Controlled Environmental Control System Based on Candidate Word Discriminations

Authors
Taro Shibanoki*, [email protected], Masaki Watanabe*, [email protected], Go Nakamura†, [email protected], Takaaki Chin†, [email protected], Toshio Tsuji‡, [email protected]
http://bs.cis.ibaraki.ac.jp

Available Online 30 September 2018.
DOI
https://doi.org/10.2991/jrnal.2018.5.2.14How to use a DOI?
Keywords
speech training; environment control system (ECS); speech recognition; candidate word; learning-type look-up table
Abstract
This paper proposes a concept of a training system for the speech controlled environmental control system: Bio-Remote based on candidate word discriminations. The proposed system can provide three-types of voice signal training: (1) volume, (2) tempo/timing and (3) candidate word which are important for accurate speech recognition based on false recognition results. During the training, such three kinds of features are extracted from measured voice signals and visually and auditory fed back to the user in real time. This allows the user to train speech abilities even if false recognition results are extracted because of slurred speech. The efficacy of the proposed system was demonstrated through training experiments for slurred speech conducted with healthy participants. The results showed that the proposed system was capable for the training of speech abilities.

Copyright
Copyright © 2018, the Authors. Published by ALife Robotics Corp. Ltd.
Open Access
This is an open access article under the CC BY-NC license (http://creativecommons.org/licences/by-nc/4.0/).