rishikanthc/Scriberr

What it solves

Scriberr is a private, self-hosted audio transcription application. It eliminates the need to upload sensitive recordings to third-party cloud providers and avoids the subscription costs associated with commercial transcription services.

How it works

The application runs locally on your machine and leverages state-of-the-art machine learning models—including NVIDIA Parakeet, Canary, and Whisper—to convert audio and video files into text with word-level timing. It includes a built-in audio recorder and can be deployed via Docker (with CPU or NVIDIA GPU acceleration) or Homebrew.

Who it’s for

It is designed for self-hosters, privacy-conscious users, and individuals who need to transcribe voice notes, meetings, and recordings without relying on cloud-based AI services.

Highlights

• Smart Speaker Detection: Automatically identifies and labels different speakers (diarization).
• Chat with Audio: Integrates with Ollama or OpenAI-compatible providers to summarize transcripts or ask questions about the recording.
• Automation Ready: Features a folder watcher to automatically process new files and an extensive API for integration with tools like n8n.
• Native Experience: Supports PWA installation for a native-like experience on desktop and mobile devices.
• Integrated Note-taking: Allows users to highlight key moments and annotate transcripts while listening.