A speech recongition program that can convert audio files to text?
December 11, 2014 1:47 PM Subscribe
One that can convert an audio file to text (not looking for dictation to text). Preferably free.
On a mac run this, open _any_ text editor, and enable dictation(you might have to pop it on in system preferences first)
It'll treat it like you're recording with the mic, and use apples dictation(which is pretty good) to convert it to text.
On windows use google dictation but basically do the same thing, like this with something like this.
The second method with chrome would also work on linux, but i haven't played around with stuff like this on linux since alsamixer was a big deal and... yea, someone else would have to explain the best current way to do loopback on linux if you're running ubuntu or something.
posted by emptythought at 3:29 PM on December 11, 2014 [1 favorite]
It'll treat it like you're recording with the mic, and use apples dictation(which is pretty good) to convert it to text.
On windows use google dictation but basically do the same thing, like this with something like this.
The second method with chrome would also work on linux, but i haven't played around with stuff like this on linux since alsamixer was a big deal and... yea, someone else would have to explain the best current way to do loopback on linux if you're running ubuntu or something.
posted by emptythought at 3:29 PM on December 11, 2014 [1 favorite]
Response by poster: Those both look like dictation services.
posted by atinna at 3:34 PM on December 11, 2014
posted by atinna at 3:34 PM on December 11, 2014
Can you code? If so, you may be able to put together a python script using CMU Sphinx to accomplish what you want. Here's an example.
posted by Poldo at 7:22 PM on December 11, 2014
posted by Poldo at 7:22 PM on December 11, 2014
Speech recognition is hard. Really hard. This is something people work on to get their PhDs in computational linguistics and related fields, and then keep researching for years and years.
Here's something explaining why. Granted, this is from a transcription company so they're obviously biased, but nothing they're saying is wrong: these are all problems that any speech recognition software is going to have problems with. Feel free to poke around the Wiki article for speech recognition for more information.
So, with that being said, apparently, YouTube can get you a rough start, but anything that you can get, for free, right now, is going to be rough.
posted by damayanti at 5:29 AM on December 12, 2014 [2 favorites]
Here's something explaining why. Granted, this is from a transcription company so they're obviously biased, but nothing they're saying is wrong: these are all problems that any speech recognition software is going to have problems with. Feel free to poke around the Wiki article for speech recognition for more information.
So, with that being said, apparently, YouTube can get you a rough start, but anything that you can get, for free, right now, is going to be rough.
posted by damayanti at 5:29 AM on December 12, 2014 [2 favorites]
Try Pop Up Archive. It does very good machine-generated transcripts — but they aren't perfect. No speech-to-text software that you can buy or use for fee is perfect, or even close to perfect. They have one hour free in their basic plan, which is good for testing.
Also, the speech-to-text in Android is very good and there are a number of free apps in the Play store that make it easier to use for long passages. Like some of the above recommendations, you would simply play the audio into your phone as if it were lives speech.
posted by Mo Nickels at 11:52 AM on December 12, 2014
Also, the speech-to-text in Android is very good and there are a number of free apps in the Play store that make it easier to use for long passages. Like some of the above recommendations, you would simply play the audio into your phone as if it were lives speech.
posted by Mo Nickels at 11:52 AM on December 12, 2014
Those both look like dictation services
I missed the requirement that it not be, my recommendation then is dragon naturally speaking. And NOT the iOS version. That's just a service as well.
posted by emptythought at 12:10 PM on December 12, 2014
I missed the requirement that it not be, my recommendation then is dragon naturally speaking. And NOT the iOS version. That's just a service as well.
posted by emptythought at 12:10 PM on December 12, 2014
This thread is closed to new comments.
posted by rada at 3:15 PM on December 11, 2014