Voice

Voice transcription

A thin pass-through to OpenAI Whisper. Audio is processed in-memory and discarded — we never persist what you record.

Transcribe

POST/api/v1/voice/transcribe

multipart/form-data with a `file` part (webm, mp3, m4a, wav, ogg). Optional `language` field. Returns { text }. Max 10 MB.

Pro only (free returns 402 pro_required). If your agent can transcribe audio itself, prefer that — quik.md doesn’t need to be the Whisper relay. If you do use this, pair the returned text with POST /api/v1/capture to turn the thought into a triaged item.

curl https://quik.md/api/v1/voice/transcribe \
  -H "Authorization: Bearer qk_..." \
  -F "[email protected]" \
  -F "language=en"