riffado/riffado
Self-hosted AI transcription interface for Plaud Note devices
What it solves
Riffado provides a self-hosted alternative to manufacturer-locked cloud services for AI voice recorders. It allows users to own their transcripts and recordings by syncing them from the device manufacturer's cloud to infrastructure the user controls, removing reliance on a single vendor's ecosystem.
How it works
The application connects to voice recorder accounts (currently supporting the Plaud Note family) via an OTP flow to sync recordings. Once synced, it transcribes the audio using either an OpenAI-compatible API (such as Groq, Ollama, or Azure) or directly in the browser using Transformers.js (Whisper in WebAssembly). Users can choose their own storage backend, such as a local filesystem or S3-compatible providers, and the system encrypts sensitive data like API keys and transcripts using AES-256-GCM.
Who it’s for
It is designed for users of AI voice recorders who want more control over their data, privacy-conscious individuals who prefer self-hosting, and those who want to use specific AI transcription providers instead of the default ones provided by the hardware manufacturer.
Highlights
- Provider Agnostic: Supports any OpenAI-compatible API and ElevenLabs Scribe for speaker diarization.
- Privacy-First: Features AES-256-GCM encryption at rest and self-hosted deployment via Docker Compose.
- Flexible Storage: Compatible with local storage and various S3-compatible services (AWS, Cloudflare R2, MinIO, etc.).
- Client-Side Transcription: Offers free transcription in the browser via WebAssembly.
- Automation: Includes an automation API with signed webhooks for external integrations.
Related
- Project
- Project
- Project
- Project
- Project