Dictation captures whatever you say, and 'where does the audio go' is the question that decides whether a tool is usable at work. Making that a per-request choice rather than an account setting means a sensitive recording can stay local while routine notes use the cloud. The speaker-attributed segments with timestamps are the other substantive part: a transcript you can quote precisely is worth more than a block of text.
OpenWhispr's endpoint over their dictation and notes product. Twelve tools split between the notes themselves and the transcriptions behind them.
- Notes: list with folder filtering and paging, get, create, update, delete
- search_notes — semantic and full-text together
- Folders: list and create
- list_transcriptions — history filtered by language or note
- get_transcription — including speaker-attributed segments with timestamps
- get_note_transcript — the transcript for a note, as structured segments
- get_usage — word counts and plan details
A bearer key, and a paid Pro plan for API access. Each tool has a documented required scope. Transcription can run on local models, your own keys, or OpenWhispr's cloud, chosen per request.
Paste a URL with a key — add the endpoint to your client