11th Annual Conference of the International Speech Communication Association

Makuhari, Chiba, Japan
September 26-30. 2010

Turn Taking-Based Conversation Detection by Using DOA Estimation

Yohei Kawaguchi, Masahito Togami, Yasunari Obuchi

Hitachi Ltd., Japan

We propose a new method that detects conversation groups when multi-conversation groups exist simultaneously. The proposed method uses hands-free microphone arrays without wearable microphones. It has two main features: (a) We integrate a conventional turn taking-based conversation detection method with Direction of Arrival (DOA) estimation-based Voice Activity Detection (VAD). (b) The proposed method estimates the number of speakers for DOA estimation-based VAD by using turn taking rules. Experimental results indicate that the performance of the proposed method with only microphone arrays setup in rooms is comparable to that of the conventional methods with wearable microphones.

Full Paper

Bibliographic reference.  Kawaguchi, Yohei / Togami, Masahito / Obuchi, Yasunari (2010): "Turn taking-based conversation detection by using DOA estimation", In INTERSPEECH-2010, 3134-3137.