AIST - Paving the way to the Future Media files in www.youtube.com 288 Edits 29.9 % Video 1 fav
CrestMuse: demo video (4/5) Media files in www.youtube.com 143 Edits 21.7 % Video 1 fav
Staying Safe Online (21/10/2010) [Video] All items | LSE Public lectures and events | Video 32 Edits 0.4 % Video
BBC News Overnight 2011 with Gavin Grey Media files in www.youtube.com 429 Edits 74.0 % Video
|44 channels||7994 episodes
(86 have been corrected)
|16653 corrected words (Edits)||2317 searches||28 registered speakers|
PodCastle is a service that enables searching of speech data such as podcasts, individual audio or video files on the web, and video clips on video sharing services (Nico Nico Douga, YouTube, and Ustream). Speech data are converted to text data by using an automatic speech recognition technology. All users accessing the PodCastle service can correct speech recognition errors. For more information, please check "What is PodCastle?".
It takes a few hours or a few days to finish the speech recognition after entering the URLs of speech data at "Register contents". Please refer to "Content registration" for the details about how to delete a content on PodCastle.
Since this beta service is under development, it might be down for maintenance. The transcripts generated by a speech recognizer are not necessarily perfect. Recognition errors tend to increase in speech data that have bad acoustic conditions (loud background noise, multiple speakers, spontaneous speeches, etc.).
The following environment is supported by PodCastle.