MacOS – Commercial command line utility for transcribing voice memos to text

macossoftware-recommendationspeechvoice-dictation

I'm looking for a commercial command line utility to transcribe my MP3 voice memos into text files. My workflow is to record voice memos through out the day and upload these memos as MP3 files to Dropbox. I would like to convert these MP3 files to equivalent TXT files on a batch basis. My voice memos range from < 1 minute to a max of ~10 minutes.

I tried using Dragon Dictate 4 for Mac and found this application a nightmare to use and very unreliable. I would be happy to purchase a slimed down version of this application that only transcribed audio files but I can't find such a product on the Nuance website. Unfortunately I could not get this application to work reliably enough to look for an API or scripting interface … and I'm a little skittish about this vendor after reading about similar experiences from other users.

I've also looked at cloud services as well, but the ones I've found seem to be rather expensive – $0.20/minute and with limitations on the max audio file size that can be uploaded.

Best Answer

I doubt you can find command line commercial application.

You can better use some open source library like pocketsphinx http://cmusphinx.sourceforge.net

To transcribe a file you can use latest version 5prealpha:

   ffmpeg -i file.mp3 -ar 16000 -ac 1 file.wav
   pocketsphinx_continuous -infile file.wav -logfn /dev/null

that should give you an automated transcription. To increase the accuracy you can adapt to your own voice.