Speakr

Self-hosted AI transcription and intelligent note-taking platform that turns audio recordings into organized, searchable notes with speaker identification.

Open source alternative to:

Rev

Speakr is a personal, self-hosted web application for transcribing audio recordings into organized, searchable, and intelligent notes. It uses AI for high-accuracy transcription with speaker identification, voice profiles powered by voice embeddings, and an interactive chat interface that lets you ask questions about your recordings.

Speakr supports multiple transcription backends including OpenAI Transcribe (with diarization), self-hosted WhisperX ASR, Mistral Voxtral, and VibeVoice ASR — giving you full control over where your data goes. It offers a comprehensive REST API with Swagger UI, SSO authentication via any OIDC provider (Keycloak, Azure AD, Google), internal and public sharing with granular permissions, group management, smart tagging with custom AI prompts and tag stacking, auto-deletion with flexible retention policies, and full internationalization across six languages.

Built with Python/Flask and Vue.js 3 with Tailwind CSS, Speakr runs entirely on your own infrastructure using Docker, ensuring your sensitive conversations remain completely private. Additional features include audio-transcript synchronization with click-to-seek, semantic search across all recordings (Inquire Mode), customizable export templates for Obsidian and Logseq, incognito mode for session-only processing, and support for both SQLite and PostgreSQL.