acm-header
Sign In

Communications of the ACM

ACM TechNews

'googling' Through ­nique Audio Material: Towards a Better Search Result


View as: Print Mobile App Share:
audio icon

Credit: TMCEC.com

University of Twente researchers are working on the CHoral project, which focuses on making spoken audio material from the past more accessible. CHoral is one of 18 projects from the Netherlands Organization for Scientific Research's Continuous Access to Cultural Heritage program.

The researchers note that automatic speech recognition, when combined with search technology, can offer a method for searching through sound files. The researchers, led by Twente's Laurens van der Werff, proposed a way of evaluating the quality of automatically generated transcripts that pays more attention to how end users want to use the search results. They say the method offers the possibility of an improved analysis of where problems occur and provides leads for optimization.

The CHoral project's techniques were used on collections from the Rotterdam Municipal Archive, the NIOD, and the interview archive of Aletta/IAVV. The techniques also have led to the University of Twente's SHoUT program, an open source speech recognition package. The researchers say the program makes each archived audio source accessible without the need for its own in-house specialist.

From Netherlands Organization for Scientific Research
View Full Article

 

Abstracts Copyright © 2012 Information Inc., Bethesda, Maryland, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account