作者: Mike Matton , Karel Braeckman
关键词:
摘要: Broadcast video retrieval is a key issue for media researchers looking suitable material in archives. Current applications use at VRT have proven to be suboptimal solution. In this paper, we explain novel search environment based on the combined metadata from multiple broadcast systems. Furthermore, how speech recognition can facilitate unlocking archive. This illustrated by conducting experiments and measuring keyword rate, rather than pure word error rate. We show that rate sufficient efficient application.