TranscribeGeek

TranscribeGeek turns audio and video recordings into text on your own computer. It runs OpenAI's Whisper models locally, so an interview, a lecture, a meeting or a medical recording never leaves the machine and is never uploaded to a transcription service.

Drop in a batch of files and it works through the queue. Four model sizes are offered, with the real download size shown before you commit to one, so you can match the model to the hardware you have. Output is a plain text transcript and, optionally, a timestamped SubRip .srt subtitle file.

Once a model has been downloaded it works with no internet connection at all.

There is no account, no advert, no bundled offer, no telemetry and no per-minute charge. Free for personal and commercial use, open source under GPL-3.0. The .NET 8 runtime is bundled, so there is nothing to install first.