What do you use to generate transcripts?

Anybody know of any free programs or web services to generate transcripts from audio files? Ones that work in different languages would be helpful of course. Obviously the transcripts are probably not going to be perfect, but I would imagine that it is still quicker to auto-generate and then correct a transcript than it is to create one manually.

I am personally looking for ones that work in English, Spanish, and Italian, but those who know of ones that work in any other language are more than welcome to post those as well.


Try the auto captioning in YouTube.

so I have to upload an audio file as a video to youtube? Surely there is a more direct way…

If someone finds a free way that is not too complicated and works ok, it would revolutionise LingQ

I personally haven’t uploaded anything to this so I don’t know what is involved in uploading your content but it works with English and Japanese currently:

ABC broadcasting from Australia has many radio articles that are availble as mp3 files with transcripts on many topics every week. Find them at abc.net.au
I think you will find them interesting and useful.
The following is the title of one of these programs. I found it by clicking on the heading ‘latest transcripts’ on the front page of the site.

18 February, 2012 7:14PM AEDT
An English Maori celebration

It could be useful to do a little exploring of the site because there is a huge amount of material available.

PS If anyone knows of anything similar in German, I would be very interested.

@Rae68: Auf meinem Blog habe ich ein paar Links mit Quellen zusammengestellt: http://learninggerman.verasblogs.com/weitere-quellen-zum-deutschlernen/
Interessant dürften vor allem die folgenden Quellen sein.
Artikel zu Wissenschaft und Bildung vom Deutschlandfunk, viele mit Transkript und Audio:
Außerdem findet man Interviews beim Deutschlandfunk, viele davon mit Transkript und Audio:
Montag bis Sonntag - Interviews
Radio Prag (Fortgeschrittene):
Nachrichten aus der Tschechischen Republik | Radio Prague International
Oder Deutsche Welle Nachrichten für Deutschlernende:
Langsam gesprochene Nachrichten | DW Deutsch Lernen

Podcastle looks promising but there a quite a few current problems as far as I can see.

The concept is that you upload audio or video and its automatic speech recognition software creates a text. Then the text appears on a single horizontal line in the middle of the screen… Each word has a few possible alternatives in a droplist and if you think the automatically generated word is wrong you can just click on the correct one. There is also an “empty” word alternative in which you can manually write the correct word. There are also community features so that the correction is crowd sourced. As well, the feedback humans give is supposed to further improve the ASR.

Challenges on this site are: currently only works for English and Japanese, unintuitive interface (I suggested some audio but I still haven’t been able to edit it or even find the ASR transcript), a lot of broken or hanging links.

I found another possible (English Only) option called www.voicebase.com.

I am also planning on experimenting with pumping my audio output back into the microphone jack while using Microsoft’s speech recognition to see how accurate it is with untrained voices.

@veral: Viele Dank Vera. Ich finde dradio und DW sehr interessant und nützlich!