If you record talking clips and captions are the step you keep putting off, the built-in speech recognition saves most of the typing.

One-time setup: Settings > Configure Kdenlive > Speech to Text. Pick Whisper (slower, more accurate, more languages) or VOSK (fast and light), then download a model for your language. The small Whisper models are a good start on a laptop without a GPU.

Then per video: put the clip on the timeline, open Speech Recognition from the subtitle tools, and run it on the whole timeline or just a zone. It fills the subtitle track for you.

A quick cleanup pass that works well:

• Play it back once and fix only the wrong words. The timing it gives you is usually close enough, so don’t retime everything.
• Split long lines so each subtitle is a few words. On a phone, two short lines read much better than one long one.
• For vertical videos (Shorts, Reels), keep the captions above the bottom area where the app’s buttons and description cover the frame.

Curious how long a 5 minute clip takes for people here with Whisper on CPU only, and which model size you settled on.