The index of the sound appeared on the page of Google Labs
page promising developments Google Labs has a new project: GAudi (Google Audio Indexing). It is a recognition technology and indexed the English language, which is extracted from the multimedia files, including video.
/ > Latest development Google a de facto started testing two months ago on a small number of videos from YouTube: see the news "appeared On YouTube a full-text search in video". But it was a kind of "black box": we could just see a new feature, but did not know what was really behind it. Now published as a separate interface to search for videos (in this index in the future, you can load any video content from the Internet), as well as FAQ information.
From the FAQ, we learned that the engine of speech recognition created from scratch special working group of Google employees. Although this region are surveying tens of years by many companies, but GAudi is a completely independent development of Google.
Currently supports only the English language and, of course, makes a lot of mistakes. For example, in this video the word "Czechoslovakia" misinterpreted as "tech also but there" and the word "free" is recognized as "forty", and such errors quite a lot.
On the project page it is reported that the engine of speech recognition will gradually "feed" not only electoral rolls, but other themed YouTube channels, and in the future, perhaps it needs to index video content from other sites.
Article based on information from habrahabr.ru
/ > Latest development Google a de facto started testing two months ago on a small number of videos from YouTube: see the news "appeared On YouTube a full-text search in video". But it was a kind of "black box": we could just see a new feature, but did not know what was really behind it. Now published as a separate interface to search for videos (in this index in the future, you can load any video content from the Internet), as well as FAQ information.
From the FAQ, we learned that the engine of speech recognition created from scratch special working group of Google employees. Although this region are surveying tens of years by many companies, but GAudi is a completely independent development of Google.
Currently supports only the English language and, of course, makes a lot of mistakes. For example, in this video the word "Czechoslovakia" misinterpreted as "tech also but there" and the word "free" is recognized as "forty", and such errors quite a lot.
On the project page it is reported that the engine of speech recognition will gradually "feed" not only electoral rolls, but other themed YouTube channels, and in the future, perhaps it needs to index video content from other sites.
Комментарии
Отправить комментарий