Analysis of factors of misrecognition and study of training methods for use of speech recognition system

2016 Fiscal Year Final Research Report

• Project/Area Number: 25730117
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Perceptual information processing
• Research Institution: Daido University
• Principal Investigator: Tsuge Satoru 大同大学, 情報学部, 准教授 (00325250)
• Project Period (FY): 2013-04-01 – 2017-03-31
• Keywords: 音声認識 / 音声インターフェース / 音声認識システム訓練方法 / 誤認識要因の解明

Outline of Final Research Achievements

The main theme of my study is the wide spread of speech recognition system. In this study, we focused on "analysis of factors of mis-recognition" and "training method for speech recognition system users". For analysis of factors of mis-recognition, I investigated the factors of mis-recognition under the noisy environment using CENSREC1 speech database. From this investigation, one of mis-recognition caused by the error of voice activation detection. In addition, I collected the speech data when the speaker uttered the speech recognition system for long term. Using these speech data, I investigated the factors of mis-recognition. On the other hand, I unfortunately have not constructed the training system for speech recognition system user. But, I have constructed some speech recognition system, which are a navigation system on potable system, a introduction system of university on personal computer, and so on.

Free Research Field

音声情報処理