Speakr
Self-hosted recording and transcription app with speaker labels, searchable notes and configurable local or cloud speech and text models.
Speakr is a self-hosted transcription and note-taking app for meetings, interviews and lectures. Its browser interface can record microphone and system or browser-tab audio, or accept existing files. The app runs on your server and supports Docker deployment.
Speech processing uses the transcription service you configure. For local processing, Speakr connects to a separate Whisper or WhisperX ASR service. Hosted providers are also supported. Summaries and chat use a configurable OpenAI-compatible text endpoint, including local services such as Ollama or LocalAI. Keeping the workflow local requires local endpoints for both speech and text generation.
Transcripts synchronize with audio playback, so you can jump to a passage and review what was said. WhisperX can label speakers and match recurring voices across recordings. Custom vocabulary helps with names and specialist terms. Tags and folders organize the library, while prompts control how recordings become summaries or study notes.
Chat answers questions about individual recordings with timestamp references. Inquire Mode searches and queries the wider library. Teams can use group sharing, differentiated permissions and OIDC sign-on. A REST API and signed webhooks connect recordings to other workflows. Retention rules can remove audio while keeping its transcript.
