Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I listen to a daily podcast that is about 4 hours long. I think it would be a cool project if I could come with some way to generate transcripts of it automatically. Is there any software that will "listen" to the mp3s and create text of what they are saying?

I'm not very concerned with differentiating who is talking because I think that would be asking too much. There are 4 main people speaking and others less often.

share|improve this question

closed as off-topic by Tog, Excellll, Kevin Panko, MariusMatutiae, harrymc Jun 6 at 8:05

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "Questions seeking product, service, or learning material recommendations are off-topic because they become outdated quickly and attract opinion-based answers. Instead, describe your situation and the specific problem you're trying to solve. Share your research. Here are a few suggestions on how to properly ask this type of question." – Tog, Excellll, Kevin Panko, MariusMatutiae, harrymc
If this question can be reworded to fit the rules in the help center, please edit the question.

add comment

7 Answers 7

up vote 0 down vote accepted

I'm pretty sure Dragon will let you do this. You might have trouble with the accuracy though.

share|improve this answer
    
Ouch, a bit pricey at $199. –  AndyMcKenna Jul 22 '09 at 6:42
1  
Dragon is not powerful enough. I've tried it. –  studiohack Apr 10 '10 at 6:36
    
Even today, in 2014, Dragon is no different. For it to work well, you need to "train" it to the voice that is speaking by reading as much as 500 words from a text file. Even after that, it will miss punctuation more than half the time. –  fredsbend Mar 29 at 0:28
add comment

Steps:

  1. Download Express Scribe free transcription player and install.
    1. Once the application is installed you will find a example dictation file already loaded. This will give you the opportunity to play around with the controls and options to set the application up to suit your needs.
    2. From the main interface click on the Settings button.
    3. Express Scribe allows you to change or set quite a few parameters including Rewind and Fast Forward Speeds, System Wide Hotkeys (to control the application without using a mouse), set up of a Foot Pedal Control (if you have one), Autoloading of files from a LAN folder or FTP Server, choice of file types to allow (over 20 supported) and set up of a Speech Recognition Engine.
    4. Once you have your options set up you can now load your first audio file. Depending on where you are receiving it from you can either click Load (it will open a dialogue box that will allow you to browse for the file on your PC), Load Now (if you have set up a LAN or FTP connection) or Dock (if you have to digital or cassette dictation recorder with you).
    5. Select the audio file in the main interface window and click the Play button to start. If needed slow down or speed up the audio playback using the speed control in the far bottom right corner. Above this control you can also adjust the playback volume.
    6. If you would prefer Express Scibe to float above your word processing application you can choose Scribe Mini from the main interface toolbar. As the name suggests this option provides a miniture version of the application.
    7. Once you have finished transcribing your audio files you can either mark them as "Done" (removing the file from your list) or "Dispatch" (returning and email to the original sender along with your trext attachment and removing the audio file from your list.

Tips

  • Take some time to explore all the options of the application.
  • If you receive a poor quality audio file you can choose Special Audio Processes from the File menu to try and clean it up.
  • When Express Scribe receives an audio file it is stored in C:\Documents and Settings\username\Application Data\NCH Swift Sound\scribe\Current
  • If you have marked an audio file as "Done" but want to retieve it you can Recover Old Files from the File menu.
share|improve this answer
2  
I don't think the poster wanted to do manual transcription. –  JeffP Nov 26 '09 at 20:23
add comment

i'm using Dragon NaturalSpeaking for transcription. excellent software!

but it ain't free, Amazon offers the actual version 10 for $45, if you look around you'll get DNS9 even cheaper.

share|improve this answer
    
Can it transcribe from a file though? There are many programs that do this, but many will only listen from a live mic –  Paul Betts Oct 2 '09 at 19:56
    
Transcribing a recording with Dragon Naturally Speaking: wiki.wsu.edu/ctowiki/… –  Molly7244 Oct 2 '09 at 20:31
    
Yes it can - though I seem to recall you need to the Preferred version or greater to enable that feature, though I can't seem to prove that... nuance.com/naturallyspeaking/resources/product-matrix.asp –  Goyuix Oct 29 '09 at 15:17
add comment

The problem with Dragon Naturally Speaking in a context like this is that you really wouldn't be able to go through the training process where it learns to accurately interpret the speech. It's not intended for this kind of use. Maybe the NSA has some software to do this sort of thing. :)

share|improve this answer
    
i'm not familiar with the training of voice recognition programs, but it seems like you could "train" it by playing back short snippets of the podcast, over and over -- basically the same way you train it to your own voice. certainly won't be automatic tho. –  quack quixote Apr 10 '10 at 5:08
    
@quack quixote: problem is, it is already hard enough to train it to your voice, let alone 4-5 other voices...When I tried Dragon 9, it wasn't anywhere near where it should have been (or where I thought it should be)... –  studiohack Apr 10 '10 at 6:37
add comment

One thing you could try (although not on a daily basis) is uploading the audio as a video to Youtube, then asking YouTube to transcribe the video, giving you English captions...can take up to 24 hours for YouTube to do this, however...The caption file is even downloadable and you can edit them yourself, so you could copy and paste the text to wherever you wanted...

share|improve this answer
    
Fantastic trick :) Though its not perfect but works quite good.. –  Mohib Sheth Jul 18 '11 at 18:58
add comment

You need a speech recognition software.

Several programs exist for that, but I don't know any that can do batch conversion. I'm pretty sure it's a feature of at least one of them though. Note most of them are far from cheap.

share|improve this answer
add comment

Tried wave to text? It's windows only and has a limited free trial.

How many voices are in these files? Do all the files contain the speech from one (trainable) voice? What's the audio quality? If the answers to these questions is not: 'Just me', 'Yes', and 'CD Quality', @Snark is right, you need to do this essentially manually. Get a good speech recognition program, spend some time training it and play back your files while you try to say what they say.

Expect about 90% - 95% successful transcription, which sounds great until you realize that means 1 or 2 out of every 20 words is wrong.

Other factors that may affect accuracy:

  1. Lots of acronymns in your speech
  2. incomplete sentences - speech to text tends to perform much better with the context clues it gets from complete sentences.
share|improve this answer
add comment

Not the answer you're looking for? Browse other questions tagged or ask your own question.