Open-ended questions, answered out loud.
Put a video or voice question in a survey, a diary study or a UX session, and respondents answer from the phone or laptop they're already on. Each answer arrives as a file tagged with the respondent and the question, ready for transcription and coding, with the consent and retention controls your ethics board will ask about.
No app, no account for respondents. Works inside Qualtrics, Alchemer, Nfield, Jotform and your own platform.
Think about the last time you switched brands. What made you do it?
Up to 90 seconds. Say as much or as little as you like.The recording layer for insights platforms and panels
Voice, video or screen, per question.
Each mode has a different cost, a different completion rate and a different data-protection footprint. Pick per question, not per study.
Voice
The cheapest and least intrusive. Respondents talk more than they type, and no face means less PII to protect. Opus in, AAC out, with a warning when the mic is too quiet.
1¢ / minVideo
For expression, tone and showing things to the camera. Up to 4K, with background blurring for privacy and a framing outline so the face is usable.
from 2¢ / minScreen, with voice or face
Think-aloud usability sessions, unmoderated: the participant's screen with their commentary and, optionally, their face in the corner. Desktop only. Screen recorder.
PRO planWhat a research recorder has to get right.
01Completion
Every extra step costs respondents, and panels pay per complete.
- Records in the browser on any phone, tablet or laptop; nothing to install, no account.
- Streams while recording, so pressing next is immediate; a dropped connection is recovered rather than failing the response.
- Time limits with
mrt, a countdown, replay and re-record, all optional per question.
02Structure
Hundreds of respondents times a dozen questions has to land as a dataset, not a folder of files.
- Several recorders per page, each tagged with respondent and question IDs in
payload, returned in the webhook and REST API. - Duration, device, user agent and codecs stored as metadata with each file.
- CSV export per environment from the dashboard, or pull everything as JSON.
03Analysis-ready
The answer is only useful once it's text, or at least playable in the coding tool.
- Every response converted to mp4 or AAC that plays in any analysis platform or browser.
- Send files from the webhook to Amazon Transcribe, ElevenLabs Scribe or Whisper; step-by-step tutorials for each.
- Audio extracted from video responses when the transcript is all you need.
04Ethics and protection
The IRB, the DPO and the client's procurement team all get a say.
- PII controls: collect IP and user agent, or don't. Audio-only and background blurring reduce what's captured in the first place.
- Lifecycle rules delete responses at the end of the retention period; deletion on demand for withdrawn consent.
- Ingest in the EU or US, EU silo available, or push to your own storage with nothing kept on AddPipe. SOC 2 Type I.
05Scale
A panel launch is a spike, not a trickle.
- 200k+ recordings a day per region in production; two zones per region in the EU and US.
- Recordings queue under load rather than failing, and every one is processed.
- Volume discounts above 10,000 minutes a month; audio at 1¢/min keeps large samples affordable. Pricing.
In the survey tool you already use.
Step-by-step guides for adding a recorder question to the major platforms. For a research product of your own, the same recorder has full APIs.
Building your own platform? Embed code, npm, a React hook, the JS control API for your own question UI, and REST and webhooks for the data.
// one recorder per open-ended question, tagged for the dataset PipeSDK.insert("q7", { size: { width: 480, height: 120 }, accountHash: "YOUR_ACCOUNT_HASH", eid: "study_2026_brand", ao: 1, // voice only for this question mrt: 90, payload: respondentId + ":q7" }, (recorder) => { recorder.onSaveOk = (id, name, duration) => { answers.q7 = { file: name, seconds: duration }; enableNext(); }; }); // webhook on your server: video_converted → transcribe → code
From question to coded dataset.
Ask
A recorder in the question, voice or video, with the respondent and question IDs attached.
Answer
The respondent records on whatever device they have; the answer streams to the nearest region.
Process
Converted to mp4 or AAC, metadata attached, stored in your region or pushed to your bucket.
Transcribe
The webhook hands the file to your transcription service; text comes back keyed by respondent and question.
Code
Responses and transcripts in your analysis tool, playable inline, exportable as CSV or JSON.
Studies that use it.
Open-ended survey questions
A spoken "why" after a rating, in a quant survey, at panel scale. Richer than a text box, cheaper than an interview.
voice · panelsVideo diaries and ethnography
Participants record entries over days or weeks from their own phones, wherever the moment happens.
video · mobileUnmoderated usability tests
Screen and think-aloud on a prototype or a live site, with no session to schedule.
screen + voiceConcept and ad testing
Show the stimulus, then record the reaction on camera, in the same page.
videoAcademic and clinical studies
Spoken responses, language samples and self-reports, with the consent, anonymisation and retention controls an ethics board requires.
voice · videoLanguage and speech research
48 kHz Opus capture with the original file kept, for phonetic and linguistic work that needs the raw audio.
voice · original keptQuestions research teams ask.
Do respondents need an app or an account?
No. They answer in their browser, on any phone, tablet or computer, from inside your survey. They never see AddPipe as a service.
Can I offer voice and video in the same survey?
Yes, per question. Voice for most open-ends, video where expression matters, and a typed fallback in your survey tool for respondents who'd rather not record.
How do I know which response belongs to which respondent and question?
Pass both IDs in payload when the recorder loads. They're stored with the file and returned in every webhook and REST call, and included in the CSV export.
Does AddPipe transcribe?
Not itself, but the webhook fires the moment a file is ready and there are step-by-step tutorials for Amazon Transcribe, ElevenLabs Scribe and Whisper. Many teams transcribe every response automatically within a minute of it being recorded.
How do we handle consent withdrawal and retention?
Delete any response through the dashboard or REST API, and set lifecycle rules so responses expire at the end of the retention period without anyone remembering to do it.
Can we keep participant data in the EU?
Yes. Pin the study's environment to the EU and enable the EU silo; ingestion, processing, storage and even the recorder script are EU-hosted. Or push to your own EU bucket and keep nothing on AddPipe. GDPR details.
What does a study cost?
Voice is 1¢ per minute, video from 2¢. A $50 Standard plan covers about 5,000 voice minutes a month before anything extra is billed; volume tiers apply above 10,000 minutes. Pricing and calculator.
What if I need help integrating?
Contact us and an engineer will reply. For work beyond AddPipe, our custom development partners can help.
Something else? Email us.
Add one spoken question to your next survey.
Drop a recorder into a test survey, answer it from your phone, and watch the response land with your IDs attached. Free for 14 days, no card required.