Subtitle composer can do that: https://subtitlecomposer.kde.org/
Also Kdenlive has a feature for it as I know, though I never tested: https://docs.kdenlive.org/en/effects_and_compositions/speech_to_text.html
I’d also recommend something based on whisper.
If you’re looking for live transcriptions: https://github.com/abb128/LiveCaptions
And I’ve fiddked around with vosk a year ago.
I wrote a TUI for whisper in bash for a journalist friend of mine for exactly this use case. It’s a bit hackey, but it’s a good place to start.
Whisper is your best bet for FOSS transcription. This is the most efficient implementation AFAIK: https://github.com/guillaumekln/faster-whisper.
How does whisper compare to Mozzilla’s Deepspeach?
From what I’ve heard they’re competitive for English but I’ve never used Deepspeech myself. Whisper has much more community support so it’s probably easier to use overall.
Whisper is pretty good and open source, you just need to write your own script to do the automation.
And then you can also use some summarisation with OpenAI to create short summaries for each lecture or extract highlights or key points.
You can then upload them to Obsidian to make them indexed and searchable and can use any of their plugins to make it even better.
And you can use Syncthing to sync it to your phone.