Speaker diarization using data-driven audio sequencing
Résumé
In this paper, a speaker diarization system based on datadriven segmentation is proposed. In addition to segmentation and clustering steps, a new module which detects repeated segments between the same shows broadcasted on different dates is added. This process is achieved by using the ALISPbased audio identification system which segments audio data into pseudo-phonetic units. The ALISP segmentation is then used to identify the similar audio segments in the TV and radio shows. The system was evaluated during ETAPE evaluation campaign in 2011 on 11 French TV and radio shows and obtained a DER of 16.23% which was the best result among seven participants