Multilingual Speech and Video Search
Gareth Jones
School of Computing, Dublin City University, Ireland
Abstract
Providing access to spoken and video content presents challenges in terms of indexing this content. This most obviously involves knowing what has been said, but can also involve use of supporting metadata. Manual transcription of spoken content is generally uneconomic, meaning that speech recognition techniques must be used. Searching spoken content in a multilingual environment introduces further issues of translation between the query and document languages involved. The visual data stream of video material is often independent of the spoken soundtrack meaning that it can be useful for language independent indexing and also as a retrieval target in multilingual search.
This tutorial will introduce techniques for speech and video retrieval, including speech recognition and visual indexing. These will then be explored in the context of multilingual search examining its opportunities and challenges.
Course Material
Multilingual Speech and Video Search
Gareth Jones,
School of Computing, Dublin City University, Ireland
