Interview on Transcription (closed captioning)
J Thoendell stashed this in Film
The company I freelance for has its own voice recognition software. Its engineers run client audio files through the software to produce a rough transcript that can be quite accurate depending on the recording. After being thus processed, the files are placed on an online marketplace, at which point the editor logs on to choose one to transcribe. The editor can see information about each file: the client, running time, and price per minute. An audio and visual preview is also available so that one can avoid the most difficult files—faint recordings seemingly set in a wind tunnel and featuring multiple motor-mouthed speakers with accents that confound the voice-recognition technology, which is set to American English. These files lure in many a young transcriber by offering higher rates, but wily veterans know to search for the hidden gems that maximize one’s dollar-to-effort ratio (which reflects the real hourly rate more accurately than the dollar per audio minute does). My greatest such discovery was a documentary about a wandering yogi wherein three total words were spoken, two of which were subtitled and needed only a [SPEAKING FOREIGN LANGUAGE] tag. I measure every new project against this Platonic ideal.
Stashed in: History of Tech!