TranscribeGeek TechyGeeksHome

TranscribeGeek runs OpenAI's Whisper speech models locally through whisper.cpp, so a recording is transcribed on the machine it is already on. There is no account, no server and no per minute charge, and a two hour recording costs the same as a two minute one. It writes a plain text transcript beside the source file, with or without timestamps, and an optional SubRip .srt subtitle file. Files are queued and worked through one at a time. Twenty two languages are supported, or it can detect the language itself. Speech models are not bundled. You pick one on the Models screen and it is fetched once and kept, from Tiny at 78 MB up to Medium at 1.5 GB. Compressed formats such as MP3 and MP4 need ffmpeg present to decode. Plain 16 kHz mono WAV works either way, and the app says plainly which case it is in rather than failing quietly.