Die automatic speech recognition is for many YouTubers-shower a blessing: Especially hearing impaired help YouTube Captions very. The language in videos YouTube already recognizes wonderfully and writes it often error-free - but: What is with noise? How often are sounds in a clip or the sound is unclear! Soundeffects are a big area, but exactly the recognition of the Videoportal is now announced, as a report on the Google Research Blog announces.
YouTube Captions: applause, laughter, music
YouTube has so far provided three words for the algorithms behind speech recognition: "applause", "laughter" and "music". Further terms will follow, even if recognition and implementation are already a problem. For example, "music" is a very simple term and does not define what kind of melody is currently running. In addition, many words, such as "ringing", have several meanings: does it refer to a door, a telephone or a bell? It probably takes some time until Google's algorithms understand and reflect these subtle differences ...