Subtitle Edit is a free and open source tool that does a WHOLE lot of things, but the key piece it is doing for me in my workflow is transcribing audio to text so that I can get SRT transcripts with very little editing required. The accuracy on this has been amazing in the edits I have done so far. 100% accuracy, no edits required by me after the tool runs on my edited video.

My transcription steps

Here are the steps I use in Subtitle Edit that have worked so well so far.

1. Generate an MP4 file of my edited video, fully finished and ready for transcription.
2. Launch Subtitle Edit.
3. From the menu, select Video > Open video file
4. Find and select my MP4 file
5. From the menu, select Video > Audio to text (Whisper)
   • There is also a Vosk/Kaldi option that I haven’t tried, but Whisper has been so good I’m not changing!
6. My audio to text configuration:
   • Engine: Purfview’s Faster-Whisper
   • Language: English
   • Model: medium.en (1.5 GB)
   • Auto adjust timings: True
   • Use post-processing: True
     • For the settings on post-processing, I have “Merge short lines”, “Split long lines”, and “Fix short duration” enabled

The interface won’t win any awards, but it works great. There is a lot of functionality built into this tool that I’m not even using and it makes the interface a little busy so it took some digging for me to find the right buttons to click. For the most part, I left things on the default settings and everything worked out really well.

The installation doesn’t come with all the tools, though. For example, as you make choices on your engine and model the tool begins asking you to download dependent packages. I actually liked this because it meant that the installation wasn’t a huge bloated thing at the beginning, but slowly adds in only the pieces you actually need.

In terms of activity, you can watch the GitHub repo for new releases. The latest release was in December 2023

All in all, great tool, you should add it into your flow!