VoiceCapture
Voice Capture lets survey respondents answer open-ended questions by voice instead of typing. A researcher creates a project, gets a small HTML/JS widget snippet for each open-ended question on their survey platform (QuestionPro, Alchemer, or any platform that accepts custom HTML/JS), and pastes it once per question. Respondents click the widget, speak their answer, and the transcription is written back into the survey's own answer field — it exports like a typed response, no extra step for the researcher. With this connector, Claude can list and create Voice Capture projects, generate the embed snippet for a question (including the exact paste location for each platform), and read back transcribed responses to check a test answer or review results after fieldwork. It pairs naturally with a survey-platform connector like QuestionPro's: Claude builds the survey and the voice question in the same conversation. Audio is never stored — only the transcript. Sign-in uses the user's own Voice Capture account (OAuth 2.0, dynamic client registration via Supabase Auth). Creating a project consumes no credits; each transcribed response does.
Details
https://api.voicecapture.ai/mcp · streamable-http
