2

I have lots of audio files that I want transcribed. Dictation is the best transcriber I have found that will run on my MacBook (there exist others like PocketSphinx but this is no where near as accurate).

How can I use Dictation to transcribe audio files in batch?

4
  • Don't know whether or not you've see the following article, but thought I'd mention it may be worth looking at. Have a look at: How to Transcribe .MP3 Audio from Podcasts or .MP4 Movies to Text on Mac OS – user3439894 Oct 7 '17 at 19:26
  • @user3439894 not helpful to me, I specifically need it in batch. Of course I've already realised I could use output>input redirection to transcribe just a couple minutes of audio, but I'm talking days of audio, and I do not want to be unable to use my MacBook for days. – theonlygusti Oct 7 '17 at 19:27
  • I fully understand you want to do it in batch, you did after all say "How can I use Dictation to transcribe audio files in batch?"! Also, if you have days worth of transcribing to do and you also have to use the Mac at the same time, then I do not believe you can do it with Dictation. That said, I'd set up a scripted solution that runs while I'm not using the computer, e.g. while I'm sleeping, then it doesn't interfere with my usage. – user3439894 Oct 7 '17 at 19:37
  • On my Mac, even with eight GB RAM and an SSD, I've found that Dictation needs a bigger buffer. It fails to type until I pause my speech, and if I say too much before a pause, it loses whole phrases. This began after one of the recent updates—a few months ago, it was much better. This problem makes it impossible for me to use on an unedited audio file; I must first load the file into Audacity or Amadeus and insert lots of pauses. Putting them in good locations (so as not to mess up context, which Dictation uses) means pretty much listening to the whole file. – WGroleau Dec 27 '18 at 5:44
2

You can use SFSpeechRecognizer (mirror) (requires macOS 10.15+): this is made for speech recognition and it's an SDK so you can write some code for batch processing.

Perform speech recognition on live or prerecorded audio, receive transcriptions, alternative interpretations, and confidence levels of the results.

Note that from https://developer.apple.com/videos/play/wwdc2019/256/ (mirror):

enter image description here

1

I just tried using eqMac with VLC. In VLC you can select which audio device to use for output in the audio menu.
Then in the dictation settings you can select eqMac as the input and continue from there - as per the blog post mentioned in comments.
For batch processing, just have a VLC playlist of all your files and let it go. It will all go to one text file though.
I tried out with some previous recordings and the recognition was poor. Hopefully that will improve with more articulate and clear speaking.

3
  • And this will take 10 hours to complete assuming there's no hiccups. Not really ideal – theonlygusti Feb 1 '18 at 6:30
  • Also this was discussed on the comments of the quetsion – theonlygusti Feb 1 '18 at 6:30
  • No it was not. If you go to the blog post, the author stated that soundflower can no longer be used. The only alternative was a $99 commercial application to replace it. eqMac + VLC was never mentioned. It still uses Macs native recognizer, true, but at least this can be tried without paying $99 – vv111y Feb 1 '18 at 16:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .