Scriberr is an open-source, self-hosted AI audio transcription application designed for people who value privacy and performance. Developed by rishikanthc and released under the MIT license, it has earned roughly 3,110 stars on GitHub and is actively developed in 2026. Everything runs locally and offline: your recordings never leave your machine, which makes it a popular choice for journalists, researchers and anyone transcribing sensitive material.
Scriberr is powered by OpenAI’s Whisper models, with fast local inference and speaker diarization that labels who said what in multi-speaker recordings. It ships with a polished web interface and desktop experience for uploading audio, managing transcripts, searching across them and exporting results. A documented API lets you integrate transcription into other tools, and the whole stack is packaged for straightforward self-hosting, typically via Docker, on anything from a home server to a workstation.
Because transcription happens entirely on hardware you control, Scriberr can be run on a home server, a NAS or a workstation, using smaller Whisper models on modest CPUs or larger, more accurate models where a GPU is available. Transcripts can be searched, read alongside the audio and exported for use in notes, articles or video subtitles, and the API makes it practical to fold transcription into larger personal automation setups. The project is a good illustration of the local-first AI movement: capable models, your data, no subscription.
Scriberr is free and open source under the MIT license. There is no cloud plan or usage fee; the only cost is the hardware you run it on.
Journalists, podcasters, researchers, legal and medical professionals, and privacy-conscious self-hosters who need accurate transcripts without sending audio to a third-party cloud.