Speakr

Speakr is a personal, self-hosted web application that turns audio recordings into organized, searchable notes using AI transcription and post-processing. It supports both cloud and self-hosted ASR/LLM backends and is designed for privacy-conscious individuals and teams.

Key Features

• In-browser recording and audio file upload
• AI transcription with optional speaker diarization and audio-transcript sync
• Voice profiles via speaker embeddings when using a compatible WhisperX ASR service
• Interactive chat and semantic “inquire” mode to query recordings using natural language
• Tag-based organization with custom prompts, ASR settings, and prompt stacking
• Sharing and collaboration with granular permissions, groups, and group-scoped tags
• Retention policies and automatic deletion with tag-based protection
• REST API v1 with OpenAPI/Swagger UI
• Single Sign-On via OIDC providers

Use Cases

• Meeting and standup transcription with searchable summaries and action items
• Research, interviews, and personal voice notes exported into a knowledge base
• Team knowledge capture for architecture decisions and client calls with controlled sharing

Limitations and Considerations

• Some advanced features (voice profiles/embeddings) require a separate WhisperX ASR service and typically a GPU
• LLM-powered summaries/chat depend on configuring a compatible text model provider

Speakr combines transcription, organization, and collaboration in a single web UI, while keeping data under your control. Its tagging, sharing, and retention features make it suitable for both personal note-taking and team workflows around recorded conversations.