Needs of academic researchers using speech-to-text systems

You are here

Type: 
Blog
Author: 
Luke McKernan
Year: 
2013
Org.: 
BBC [British Broadcasting Corporation], Luke McKernan
Language: 
English
Difficulty: 
Not Applicable
Tags: 
Kennisbank
Topics: 
 

Blog by Luke McKernan in the context of the British Library project Opening up Speech Archives, which has looked at the application of speech-to-text technologies (ASR) for research, particularly in the art and humanities. McKernan introduces shortly the different types of speech recognition systems available: dictionary-based or phonomenes-based. BBC Research & Development tried yet another option: applying the available audio to automatically annotate programmes within the World Service archive with Linked Data web identifiers. The combination of Semantic Web technologies, automatic indexing and crowdsourcing (invited researchers providing feedback on the suggested tags) resulted in the continuous improving of the search algorithm and the navigation in the archive.

With the links to demonstrations of the systems and prototypes inserted, this blog provides a quick insight into the speech-to-text technologies, which is expected to bring about a huge change in how researchers and other target groups discover things.