Is there a speech-to-text program, for subtitling videos?

Using applications, configuring, problems
Post Reply
Message
Author
User avatar
Aitch
Posts: 6518
Joined: Wed 04 Apr 2007, 15:57
Location: Chatham, Kent, UK

Is there a speech-to-text program, for subtitling videos?

#1 Post by Aitch »

Hi Guys'n'Gals

I have a friend who has perforated eardrums
He no longer watches TV because he complains that there is so much background music 'over the top of' what speech he's trying to listen to,
that it is unbearable for him because of the 'boom chitty boom' stealing his concentration
I recently gave him a link to a piece of video of a talk which I know he would be interested in, but he contacted me to say he couldn't make out what the talk was about & wished he had it on TV, 'because he could turn on subtitles'

Is there a speech to text program, [note: not text to speech, he's deaf not blind] for any linux, preferably puppy?
I wondered if there was something trobin was doing, that could be adapted, to enable subtitles to appear where speech is available
http://murga-linux.com/puppy/viewtopic.php?t=26863
There are other people I know also, who could benefit from this

Thanks,
Aitch

User avatar
Aitch
Posts: 6518
Joined: Wed 04 Apr 2007, 15:57
Location: Chatham, Kent, UK

#2 Post by Aitch »

Does anyone know if this could be made to work with puppy,
or if it will do what I'm after?

http://www.nch.com.au/scribe/linux.html

EDIT: or this:-
http://home.gna.org/subtitleeditor/

What do you have to do with a tar.gz file anyway?

Aitch

User avatar
HairyWill
Posts: 2928
Joined: Fri 26 May 2006, 23:29
Location: Southampton, UK

#3 Post by HairyWill »

Neither of those tools do automatic speech transcription.
Commercially available tools require training to recognise a particular persons voice.

One of the academics in our group works on live transcription without voice training. One of the uses he is tackling is enabling lectures for the deaf. The lecturer should just be able to clip on the microphone and their speech should just appear live on the main screen or students laptop.

Without voice training he has hit a barrier of accuracy (I think it is around 80%). Usually the text is accurate enough to convey understanding though the errors may be confusing or distracting. I think that he has stopped trying to improve the accuracy of the automatic system and is now working on better methods for an assistant or the whole class equipped with laptops to edit the live transcript on the fly, maybe inserting tags/notes/comments for their own personal benefit.

You might find his publication list interesting they are mainly open access using eprints.
http://www.ecs.soton.ac.uk/people/mw
Will
contribute: [url=http://www.puppylinux.org]community website[/url], [url=http://tinyurl.com/6c3nm6]screenshots[/url], [url=http://tinyurl.com/6j2gbz]puplets[/url], [url=http://tinyurl.com/57gykn]wiki[/url], [url=http://tinyurl.com/5dgr83]rss[/url]

Post Reply