Hi,
This is Scriberr, a self-hostable AI audio transcription app. Scriber uses the
Whisper models from OpenAI, to transcribe audio files offline, on your hardware.
By leveraging Whisper.cpp high-performance inference engine for whisper,
Scriberr accelerates transcription significantly making it possible to run it on CPUs (the lower 3 model sizes).
Additionally, whisper.cpp supports hardware acceleration on a wide variety of
platforms (including Nvidia and Intel GPUs) all of which Scriberr consequently enjoys as well. Auto summarization
of transcripts is supported using either Ollama or OpenAIs ChatGPT API (requires OpenAI API Key).
From 0.2.0, Scriberr also does offline speaker diarization using pyannote.
Features
- Offline audio transcription using whisper.cpp
- Offline speaker diarization using pyannote
- Auto-Summarization of transcripts using Ollama or OpenAI ChatGPT
- Use your own custom prompt templates for summarization
- Local on-device computation
- Mobile & Desktop compatible
- Simple & easy to use
Screenshots
Installation
Check out the installation section for instructions on how to deploy Scriberr. Make sure to check the usage page for detailed instructions on how to configure and use Scriberr.
Scriberr is still in its early stages of development. The app will have bugs and updates might include breaking changes. Please bear with me in case of any issues. My time on this project is limited as I’m working on this outside of my full-time job. Apologies for any inconvenience.
About Me
Hi, I’m Rishikanth Chandrasekaran. I recently graduated my PhD in Computer Science from University of California, San Diego. I currently work as an Applied Researcher at eBay. You can learn more about me here