Ever wished you could transcribe the text of an audio file easily just from the command line, without heath robinson playback and dictation software? Its easy to do now from the command line and only takes a few minutes to set up. Just remember, you're the product and nothing is for free so don't send data that might be confidential to be transcribed.
TL:DR – The whisper tool from openai does the job well, certainly for my needs, in UK English and is easy to install and use. YMMV.
Install ffmpeg
You'll need homebrew, to install some tools, (see brew.sh). You'll need a Mac and enough proficiency with the terminal to change to the right folder where your audio file is (hint: type cd and then drop the folder from finder onto the terminal window to auto complete the full folder name. Then run brew install ffmpeg .
% brew install ffmpeg
==> Auto-updating Homebrew...
... lots of output, allow it to finish ...
Install Whisper
Run % pip install -U openai-whisper
% pip install -U openai-whisper
Collecting openai-whisper
Downloading openai-whisper-20231117.tar.gz (798 kB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 798.6/798.6 kB 14.2 MB/s eta 0:00:00
Installing build dependencies ... done
Getting requirements to build wheel ... done
Installing backend dependencies ... done
Preparing metadata (pyproject.toml) ... done
... lots of output, allow it to finish ...
Transcribe text from an audio file on a Mac
Just enter the command from the folder the file is in. whisper Audio.mp4 --output_format txt
Theres a console output showing you the progress through the audio, heres a snippet
[25:17.040 --> 25:23.440] But that's a sort of slightly longer-term thing, perhaps.
[25:23.440 --> 25:26.040] Any more? Any other questions from anyone?
[25:28.600 --> 25:29.600] All right, thanks.
[25:29.600 --> 25:31.760] Thanks for having me. I'll be here.
Go and have a cup of cofffe, when you return your audio transcription will be in the same folder.
Magic really.
