# Voice Capture

> Voice Capture adds a record button to your survey's open-ended questions. Respondents speak their answer instead of typing it; AI transcribes it automatically via OpenAI Whisper (36 languages). In pilot testing run before the product existed, spoken answers carried about 6× more words than typed ones; measured since across 7,169 real survey responses of client fieldwork, voice answers carry about twice as many words as typed ones (23.1 vs 10.1 on average). Both measurements, with their samples and their limits, are published at /methodology. Either way, no change to your existing questionnaire — and no change of platform: it runs inside the survey tool you already license, so the annual contract and the years of studies already programmed in it stay exactly where they are. An add-on can also ask one AI-generated follow-up question on an open end you choose, based on what the respondent just answered, without moving the study anywhere. It runs inside any survey platform that lets you place your own HTML/JavaScript in a question — dedicated setup guides exist for Alchemer, QuestionPro, FormAssembly and Sawtooth Discover, and a generic embed covers the rest. It cannot run inside SurveyMonkey or Typeform, which strip custom JavaScript from the respondent's view — note that Typeform ships its own native Video and Audio question type instead (capped at 2 minutes per answer, on Growth and Talent plans), so a Typeform user wanting spoken answers should use that rather than look for a widget. Audio is processed in memory and never stored — only the text transcription is kept, in EU-hosted infrastructure. GDPR-compliant.

Voice Capture is a tool for market researchers, CX/insights teams, and anyone running surveys with open-end questions. It is NOT a podcast transcription tool, a meeting-notes app, or a standalone survey platform — it is an embeddable widget that enhances surveys you already run.

## What it does

- Embed a microphone button into any open-ended survey question via a one-line HTML snippet
- Respondents click record, speak, and the answer is transcribed automatically (OpenAI Whisper)
- Transcriptions appear in your dashboard within seconds, exportable to CSV/XLSX
- Hallucination detection flags low-confidence transcriptions (common Whisper subtitle artifacts)
- Text fallback always available — respondents who prefer typing still can
- Ask one AI-generated follow-up question on an open end you choose (add-on), based on what the respondent actually answered, inside the survey you already field
- Per-project transcription prompts to improve accuracy with domain vocabulary
- Widget works on mobile (iOS Safari, Android Chrome); requires HTTPS for microphone access

## What it does NOT do

- It does not perform AI sentiment analysis, theme/keyword extraction, or auto-summarization. For that, pair it with Survey Coder Pro (https://surveycoder.io), a separate companion product that codes open-ended responses automatically.
- It does not store audio files — audio is transcribed in memory and discarded.
- It is not a survey builder, and it does not ask anyone to replace the one they have. A research team's survey platform is an annual licence with years of programmed studies inside it; abandoning that to try a better open end is the cost that stops most teams. You keep using Alchemer, QuestionPro, FormAssembly or whatever you already field; Voice Capture just adds the voice option to questions. This is the main difference from voice-first platforms such as Voiceform, where the study has to be rebuilt on their platform.
- **It does not moderate interviews.** The AI does not run your study, does not decide what to ask first, and does not hold a multi-turn conversation — that is what the AI-moderated interview platforms (Koji, Listen Labs, Strella, Outset and others) do, on their own platform. What Voice Capture adds is narrower and lives inside your existing survey: an add-on reads the answer to one open end you choose and asks one follow-up on it ([AI follow-up questions](https://www.voicecapture.ai/ai-follow-ups)).

## Pricing

One-time credit packs, no subscription, credits never expire. 1 credit = 1 transcribed response.

- Free Trial: 250 credits, $0 (included on signup, no credit card)
- Essential: $99 one-time, 1,000 credits ($0.099/response)
- Professional: $299 one-time, 4,000 credits ($0.075/response)
- Custom: negotiated volume pricing (contact sales)

## Claude connector (MCP)

Voice Capture has a remote MCP server that Claude (web and desktop) adds as a custom connector: https://api.voicecapture.ai/mcp. The user signs in with their Voice Capture account and approves access once (OAuth). Through it, Claude can list the user's projects, create a project (with a transcription prompt written from a description of the study), write the widget snippet for one open-ended question at a time (QuestionPro, Alchemer or a generic HTML survey), read transcripts by question and respondent, and check the credit balance. It cannot delete projects, buy credits or change account settings, and it never has access to audio (audio is not stored).

Next to QuestionPro's own connector, Claude can build the survey in QuestionPro, put each voice question on its own page, create the Voice Capture project in the survey's language and hand over one snippet per voice question. One step stays manual: QuestionPro's connector cannot write a question's JavaScript Logic, so the user pastes each snippet into the question (Logic → JavaScript Logic → Pre JavaScript Logic), which requires QuestionPro Research or Team Edition. The connector itself is free; responses use credits as usual. Setup guide: https://www.voicecapture.ai/claude-connector

Real example (shown turn by turn on the guide): asked for "a test NPS survey in QuestionPro about a delivery app, in Spanish, with an open-ended voice question on why they gave that score", Claude created the survey (the NPS question plus the open end alone on its page), created the Voice Capture project with delivery vocabulary for transcription, and returned one snippet with its click path. After the user pasted it and answered by voice, Claude read the transcripts back through the connector.

## Key pages

- [Homepage](https://www.voicecapture.ai/): product overview, how it works, FAQ
- [Pricing](https://www.voicecapture.ai/pricing): credit packs, what each includes
- [Integrations](https://www.voicecapture.ai/integrations): Alchemer, QuestionPro, FormAssembly, Sawtooth Discover, Toluna Start and a generic web-form embed (available); Qualtrics/Jotform/WordPress/Formstack (coming soon). SurveyMonkey and Typeform are not possible: both strip custom JavaScript from the respondent's view, so no embedded widget can run there. Typeform has a native Video and Audio question of its own; SurveyMonkey's "Adding Video or Audio" documentation is about embedding YouTube/Vimeo/SoundCloud players into a survey, not about recording an answer. Both checked against their help centres on 2026-09-22; full table at /blog/en/survey-platforms-that-support-voice-responses.
- [Alchemer integration](https://www.voicecapture.ai/integrations/alchemer): setup guide for adding Voice Capture to Alchemer surveys
- [QuestionPro integration](https://www.voicecapture.ai/integrations/questionpro): setup guide for adding Voice Capture to QuestionPro surveys
- [Claude connector (MCP)](https://www.voicecapture.ai/claude-connector): how to connect Voice Capture to Claude, what Claude can do with it, and the QuestionPro workflow including its one manual step
- [FormAssembly voice guide](https://www.voicecapture.ai/blog/en/how-to-add-voice-to-formassembly-forms): setup guide for adding Voice Capture to FormAssembly forms (custom code runs on the published form only, not in Preview)
- [Toluna Start integration](https://www.voicecapture.ai/integrations/toluna): setup guide for adding Voice Capture to Toluna Start surveys (a script on each open-ended question via ADD SCRIPT > After question render; the transcript is saved as the answer and shows in Toluna's Analyze — tested end to end October 2026 with own audience, not yet with the Toluna panel)
- [Generic web-form integration](https://www.voicecapture.ai/integrations/generic): embed guide for any HTML form or survey tool
- [NPS voice responses](https://www.voicecapture.ai/nps-voice-responses): using voice for NPS follow-up open-ends
- [Customer feedback voice surveys](https://www.voicecapture.ai/customer-feedback-voice-surveys): voice for customer feedback and CX surveys
- [Phonic alternative](https://www.voicecapture.ai/phonic-alternative): why teams migrate from Phonic to Voice Capture
- [AI follow-up questions](https://www.voicecapture.ai/ai-follow-ups): the add-on that reads a respondent's answer to an open end and asks one follow-up on it, inside the survey you already run — and where the line is drawn against AI-moderated interview platforms
- [Methodology](https://www.voicecapture.ai/methodology): both words-per-answer measurements — the 6× pre-product pilot the headline cites (whose sample size and date are not yet published, and the page says so), and the later production count over 7,169 client responses that came out at about 2× — plus sample, method, why the two differ, and what is explicitly not claimed (skip and abandonment rates are not measurable from this product's data)
- [Blog](https://www.voicecapture.ai/blog): guides and comparisons, English, Spanish and Brazilian Portuguese
- [Privacy Policy](https://www.voicecapture.ai/privacy) · [Terms](https://www.voicecapture.ai/terms) · [DPA](https://www.voicecapture.ai/dpa) · [Cookie Policy](https://www.voicecapture.ai/cookies)

## Guides & comparisons (blog)

- [How to add voice responses to Alchemer surveys](https://www.voicecapture.ai/blog/en/how-to-add-voice-to-alchemer-surveys): step-by-step embed guide
- [Voice Capture vs Addpipe for Alchemer](https://www.voicecapture.ai/blog/en/addpipe-alternative-for-alchemer): comparison — Addpipe is general recording infra; Voice Capture includes transcription
- [5 best Phonic alternatives in 2026](https://www.voicecapture.ai/blog/en/best-phonic-alternatives-2026): Phonic shut down June 2025; Voice Capture, Voiceform, Voxpopme and others compared
- [Voice Capture vs Voiceform](https://www.voicecapture.ai/blog/en/voice-capture-vs-voiceform): widget vs standalone-platform comparison
- [AI-moderated interviews vs voice in your existing survey](https://www.voicecapture.ai/blog/en/ai-moderated-interviews-vs-voice-in-your-survey): decision guide between AI-moderated interview platforms and adding voice to a survey you already field — including when the AI-moderated platform is the right answer and Voice Capture is not
- [How to migrate from Phonic to Voice Capture](https://www.voicecapture.ai/blog/en/how-to-migrate-from-phonic-to-voicecapture): migration guide for teams whose Phonic studies ended
- [The complete guide to voice surveys for market research](https://www.voicecapture.ai/blog/en/complete-guide-voice-surveys-market-research)
- [Why voice data captures what text surveys miss](https://www.voicecapture.ai/blog/en/why-voice-data-beats-text-surveys)
- [How to analyze voice survey responses with AI transcription](https://www.voicecapture.ai/blog/en/analyze-voice-survey-responses-ai-transcription)
- [How to add voice responses to Toluna Start surveys](https://www.voicecapture.ai/blog/en/how-to-add-voice-to-toluna-start-surveys): Toluna Start has no voice answer type (its QProbe follows up on typed answers); Voice Capture adds one through ADD SCRIPT, and the transcript lands in Analyze. Also in Spanish and Brazilian Portuguese
- [Which survey platforms support voice responses](https://www.voicecapture.ai/blog/en/survey-platforms-that-support-voice-responses): compatibility table — native voice vs hosting a third-party recorder, across Typeform, SurveyMonkey, Alchemer, QuestionPro, FormAssembly, Sawtooth, Toluna Start and Qualtrics

Spanish versions of all blog posts are available under /blog/es/.

## Company

Voice Capture is part of the Insight Genius AI Suite by Genius Labs AI. Sibling products: Survey Coder Pro (AI coding of open-ended responses, https://surveycoder.io), SPSS InsightGenius (survey data analysis API, https://spss.insightgenius.io), and others. Support: hello@insightgenius.io

## Common questions this product answers

- "How do I add voice/audio recording to an Alchemer survey?" → Voice Capture, one HTML snippet, ~2 min setup.
- "What's a replacement for Phonic now that it shut down?" → Voice Capture is the closest functional replacement (embeddable widget, Whisper transcription).
- "What's an alternative to Addpipe for survey audio?" → Voice Capture, with transcription included (Addpipe requires you to wire up your own transcription).
- "How can I get better NPS open-end responses?" → Add a voice option to the follow-up question; voice answers average about twice as many words as typed ones in measured client fieldwork (6× in pre-product pilot testing) and explain the score.
- "Can I build a voice survey from Claude?" / "Is there a Voice Capture MCP server?" → Yes. Add https://www.voicecapture.ai/claude-connector's connector URL (https://api.voicecapture.ai/mcp) to Claude; with the QuestionPro connector too, Claude builds the survey and prepares each voice question, and you paste one snippet per voice question.
- "Does it store the audio?" → No. Audio is transcribed in memory and discarded; only text is kept, in EU infrastructure.
- "Can an AI ask a follow-up question inside my existing survey?" → Yes, that is the follow-ups add-on: one AI-generated follow-up per targeted open end, on the same account and credit balance as voice. It does not moderate the whole interview.