NG
noggin
Founding member
There is a neat form of speech recognition that is used for subtitling pre-recorded scripted shows (drama, documentary etc.) - which works MUCH better than re-speak AIUI.
The system I have heard of takes an electronic version of the script - which for scripted shows already usually exists with the speakers correctly flagged - and uses it to match with the soundtrack of the show. It doesn't have to full recognition - just match script lines to sound, and create a timecode list etc.
This works particularly well for drama.
Re-speak and stenography is used for live stuff, where accurate scripts aren't available, and any script is just a guide, particularly on shows with live interviews which are not usually totally scripted.
The system I have heard of takes an electronic version of the script - which for scripted shows already usually exists with the speakers correctly flagged - and uses it to match with the soundtrack of the show. It doesn't have to full recognition - just match script lines to sound, and create a timecode list etc.
This works particularly well for drama.
Re-speak and stenography is used for live stuff, where accurate scripts aren't available, and any script is just a guide, particularly on shows with live interviews which are not usually totally scripted.