AAMを用いた顔方位にロバストな唇領域特徴抽出と音声特徴による構音障害者の音声認識

摘要

We investigated the speech recognition of a person with articulation disorders resulting from athetoid cerebral palsy. The articulation of speech tends to become unstable due to strain on speech-related muscles, and that causes degradation of speech recognition. Therefore, we use multiple acoustic frames as an acoustic feature to solve this problem. Further, in a real environment, the speech recognition systems do not have sufficient performance due to noise influence. In addition to acoustic features, visual features are used to increase noise robustness in a real environment. However, there is a recognition problem due to the tendency of his/her unsettling head movement. We investigate a pose-robust audio-visual speech recognition method using Active Appearance Model (AAM) to solve this problem.%本稿では，アテトーゼ型脳性麻痺による構音障害者の音声認識の検討を行う.アテトーゼ型の構音障害者の場合，筋肉の緊張のため発話が不安定になりやすく，発話時に頭が動いてしまう場合がある.これに対して，音声特徴としてデルタケプストラム係数のセグメント特徴量を用いる.また，発話時の頭部の動きに対しては，Active Appearance Model（AAM）を用いることで画像から顔方位にロバストな唇領域特徴を抽出し，音声特徴と共に用いることで，雑音の影響を受けず発話変動を考慮したマルチモーダル音声認識を検討する.