Oracy
Self-hosted voice memos with searchable transcripts.
Record audio on your phone. Transcribe it on your own server. Keep the recordings, transcripts, and search index in a database you control — on hardware you control.
Why self-host
If you just want voice memos with transcripts and no setup, Google Recorder, Apple Voice Memos, or Otter will serve you better — install and go. Oracy is for the opposite case: you want your audio and transcripts in a database you control, searchable with full-text and semantic search, without an account on someone else's analytics pipeline.
The honest trade-off: you run a PostgreSQL + FastAPI stack behind Caddy, and you bring your own OpenAI API key. Transcription itself still goes to OpenAI Whisper — audio is sent to their servers for processing, so this is not end-to-end private in the strong sense. What you get is custody of the recordings, the transcripts, the embeddings, and the search index.
What it looks like
