Hunter Walk, product manager at Google Video, talked to Beet.tv about advertising and search in Google Video. Google looks at the metadata provided by the video's author, the data provided by the community (labels, comments) and tries to analyze the video in order to understand it. When talking about content analysis, he gives as an example speech-to-text, and that might mean Google will incorporate that explicitly in the future.
November 14, 2006
Subscribe to:
Post Comments (Atom)
Perhaps you meant speech-to-text?
ReplyDeleteText-to-speech has been around since the Mac 128K.
yeah, that must be some speech recognition technology (I dont belive that), or better some audio indexing combined with speech recognition (maybe), or - and that is the easiest way - some limited audio keywords spotting. This technology does not precisely convert speech to text, but very accurately find specified keywords in text (like common tags, words from comments, words from description etc).
ReplyDeleteSearch engines dump data in their records through crawlers and when any search is being requested the data is shown from those records. In the case of pictures or videos normally the picture meta description is taken in record and if any search is requested the meta description matches are shown.
ReplyDeleteSome reseraches are being carried on to produce the search data on voice or speech recognition. In this scenario the speech will be converted to text and then the normal search procedure will be carried out. This technology will really help in getting better search results of video clips and files. This technology does not precisely convert speech to text, but very accurately find specified keywords in text (like common tags, words from comments, words from description etc).
Apnaavenue.com