Thoughts on Transcription
December 11th, 2007 by Roo ReynoldsI just noticed that the You and Yours transcripts seem to have stopped being produced some time in early October. Perhaps Steve’s discovery about transcriptions for Analysis being cancelled applies more universally across the BBC. If so, that’s a shame.
If the BBC are not going to provide transcripts, maybe the internets can. We could always start running the MP3s through a transcription service like CastingWords, though since this one show would have cost $130, $78 or $39, depending on whether we wanted a 1 day, 6 day or one month(ish) service, I’d like to find another way.
Maybe there’s a space for a volunteer army of transcribers? Perhaps Speechification should host a wiki and anyone who wants to can help out. Maybe you don’t care about transcriptions. What do you think, gentle reader/listener?
December 11th, 2007 at 3:43 pm
Actually, the Analysis transcripts are back - apparently because of my ‘protest’! Only one member of staff will lose his or her job as a result, I am assured. Seriously, though, I think this is a great idea. I guess the ideal way to do it would be to distribute an MP3 to two people for transcription and then compare the results, flagging differences for attention from a third. I think this is how they go about transcribing things like phone books when digital editions don’t exist.
December 12th, 2007 at 7:06 pm
I’m really pleased to here the Analysis transcriptions are back! Do you think it’s part of a broader move to support transcription for more shows (again), or a one off? I’d be interested, if anyone knows, whether individual shows at the BBC have to pay for transcriptions themselves, of if there’s a central transcription-monkey service that does them for all shows it thinks deseve it.
December 12th, 2007 at 7:20 pm
Analysis editor Hugh Levinson made it clear it comes out of his own budget. I think they always provided them because the shows are useful for academics and students. I don’t think there’s a station-wide policy at all.
December 12th, 2007 at 9:11 pm
couldn’t you just run the whole output of Radio 4 through SpinMyBlog and just dump one long continuous transcript on a tumblrblog somewhere?
December 13th, 2007 at 12:35 am
SpinVox is limited to about three minutes I think…
December 19th, 2007 at 6:48 am
When I was at the Beeb, running the team of developers and designers sitting alongside the producers, we had quite a few forays into automated transcription. However, even at circa 95% accuracy, it was generally felt that any errors in that last 5% would be too problematic for the BBC, editorially speaking. (They’re not just being luddite; it is a complex issue, although I was on the side of 95% is better than 0%.) Equally, we hadn’t got to the point where we had automated programme pages upon which to put the automated transcriptions. Thanks to the hard work of that team, the programme pages are now dropping into place, and perhaps the transcriptions will follow. Tristan Ferne, in the R&D team, would be the man to ask …
December 19th, 2007 at 11:11 am
Yes, we did experiment with automated machine transcriptions (not brilliant results, even with a proper BBC accent) and I also sent a couple of podcasts to castingwords (good, still not perfect, some cultural differences I think). I’ll ask around about any plans.
You could use something like Annotatable Audio for creating transcriptions in a wiki-like way - go to http://www.bbc.co.uk/radio4/science/allinthemind.shtml and try one of the “Find Listen Label” links in the right hand sidebar.