Comments on: Mac Voice Recognition and Transcription

Question: Mac Voice Recognition and Transcription

thirteen — Mon, 12 Jul 2004 14:13:24 -0800

I am looking for a Mac based solution to a transcription problem. I want to have a set up where a Mac could play an MP3 file of an interview (possibly recorded on an iPod with a voice recorder), capture the sound with some sort of voice recognition, and give me back a raw text file. Is this possible, or am I still 5 years too early to be hoping for such a thing?

Barring that, is there transcription software availablle that will slow down the audio, and allow quick rewinds controlled by some sort of pseudo analogue device.

By: wackybrit

wackybrit — Mon, 12 Jul 2004 14:24:05 -0800

This slow down program may be a viable solution for you.

By: neustile

neustile — Mon, 12 Jul 2004 15:00:32 -0800

Have you tried Via Voice for OSX? It certainly does realtime transcription but I am not sure if it can take an mp3 as an input. If not, you can play the mp3 in iTunes and use something like AudioHijack pro to feed it back into Via Voice.

By: silusGROK

silusGROK — Mon, 12 Jul 2004 15:13:55 -0800

Here's the link...

By: squirrel

squirrel — Mon, 12 Jul 2004 22:38:42 -0800

Drop your key words into versiontracker.com's search engine.

By: stavrosthewonderchicken

stavrosthewonderchicken — Mon, 12 Jul 2004 23:28:22 -0800

I used ViaVoice back in '99 (it can only have gotten better, I'd imagine) and even with the 'tuned-for-UK-Oz-accents' version I had access to, after I'd trained it, it achieved better than 85% accuracy with only-slightly slower than conversation-normal speech.

I can't vouch at all for its usefulness with recorded speech, and training it would be an issue, but I imagine it might give you a decent transcript that would be a lot easier to fix on the fly during a second listen than typing the whole thing up.

Dragon Naturally Speaking is the competitor (or was), and is apparently comparable in accuracy.

By: stavrosthewonderchicken

stavrosthewonderchicken — Mon, 12 Jul 2004 23:29:56 -0800

Er, whether either of them might be available on the Mac or not, I leave to you. I dunno nothin' 'bout usin' no macs.

By: zpousman

zpousman — Tue, 13 Jul 2004 07:10:38 -0800

Not to be the naysayer in thread, but automatic transcrption of a speech (i.e., without training) is 5 years out. If not NP-complete or at least NP-veryhard. You're better off trying to come up with some nice controls for the playback speed and using the computer between your ears to actually transcribe the interview.

Maybe it's an excuse to buy a physical jog dial, something I want so bad, though I have no use for.

By: thirteen

thirteen — Tue, 13 Jul 2004 08:39:00 -0800

Thank you all for the advice. I think Via Voice is out cause I am trying to set this up for 20 editors, interviewing any number of people. I fear it (untrained in 8 directions) would be clean enough to be useful. I still like the idea of replacing our microcassette recorders with iPods, and the wheel control seems like it would do the job fine. I should be able to tweek the jog dial to work with something like the slow down program that wackybrit pointed to?

Thanks again all.