Computational Auditory Scene Analysis Using Active Audio-Visual Integration in a Dynamically Changing Environment

提出并实现了一种基于机器人摄像头和麦克风的音视频融合框架。此外，该框架被扩展，提出了“主动视听集成（AAVI）”，它提高了质量的音频和视频信息使用主动机器人的运动。自动语音识别和语音活动检测的初步实验表明，AAVI框架有效地工作，即使在视觉和/或嘈杂的条件。

A framework for Audio-Visual Integration (AVI), which can provide optimal integration according to quality of audio and visual information obtained from a robot’s camera and microphone, was proposed and implemented. In addition, the proposed framework was extended by proposing “Active Audio Visual Integration (AAVI)”, which improves the quality of audio and visual information using active robot ’ s motion. Preliminary experiments on automatic speech recognition and voice activity detection showed that the AAVI framework worked effectively even in visually and/or auditorily noisy conditions.