Scriberr

Why we included this project

Scriberr is for people who record meetings, interviews, or voice memos and want a transcript without uploading the audio to anyone. It works fully offline on hardware you own, using Whisper, NVIDIA Parakeet, and Canary models for transcription with word-level timestamps. Beyond plain text, it detects and labels speakers, and it can summarize a recording or answer questions about it by pointing at a local LLM through Ollama or any OpenAI-compatible endpoint. The built-in recorder, notes, and folder watcher fit it into a normal daily routine, and the PWA installs like a native app on desktop or mobile. Docker and Homebrew handle deployment. One thing to know: development is paused while the maintainer recovers from a layoff, though the project is not abandoned and he is open to contributors in the meantime.