Who spoke when? - automatic segmentation and clustering for determining speaker turns

The problem of labelling speaker turns by automatically segmenting and clustering a continuous audio stream is addressed. A new clustering scheme is presented and evaluated using a clustering e ciency score which treats both agglomerative and divisive clustering strategies equally. Results show an e ciency of 70% can be obtained on both manually and automatically derived segments on the 1996 Hub4 development data. For the task of identifying potentially unknown anchor speakers within broadcast news shows, the frame classication error rate is very important. To re ect this, a frame-based cluster e ciency is de ned and the results show a 90% frame-based e ciency can be achieved. Finally a frame-based comparison between the manually and automatically derived segment/cluster sets shows that approximately one third of the errors are introduced during segmentation and two-thirds during clustering.