WHAT IS IT?
Scriberr is a self-hosted audio transcription app that runs entirely on your own machine, never sending your recordings to the cloud. It turns speech into text with word-level timing accuracy, powered by modern ML models (NVIDIA Parakeet, Canary or Whisper). It's the private, free alternative to subscription transcription services charging $100-240/year.
WHY IS IT INTERESTING?
- Fully offline: all processing happens on your hardware, no data ever leaves your server. Perfect for sensitive or confidential content.
- Model choice: Whisper for versatility, NVIDIA Parakeet and Canary for speed and accuracy, all with word-level timestamps.
- Speaker diarization: automatic detection and labeling of the different voices in a conversation.
- Built-in AI chat: plug in Ollama or any OpenAI-compatible provider to generate summaries and ask questions about your transcripts.
- All-in-one: built-in audio recorder with note-taking, folder watcher for batch processing, and PWA support for a native feel on desktop and mobile.
USE CASES
- Transcribe meetings or interviews without exposing their content to a third-party service.
- Generate notes and summaries of podcasts or webinars entirely locally.
- Automate processing of a folder of recordings via the folder watcher.
- Capture voice notes and turn them into usable text straight from the browser.