Submit a ticket

Voice interviews

Let respondents talk instead of type, choose how the interviewer sounds, and know which browsers support it.

A voice interview is an AI interview that respondents answer out loud. They see a single orb and a Start button. They talk, the interviewer listens and asks its follow-up questions out loud, and the conversation is saved as a written transcript.

Turn on voice

  1. Add an AI interview block and open Interview setup.
  2. Under How respondents answer, choose Spoken.
  3. Set up the brief, objectives and limits exactly as for a typed interview.

Everything about how the interview decides what to ask, when to follow up and when to stop is the same as a typed interview, and it costs the same 1 credit.

How it sounds

  • Voice. Marin (clear and warm), Cedar (low and calm), Gleam (bright and natural), Meridian (steady and grounded), Willow (soft and gentle), Vesper (measured), Quartz (crisp and friendly) or Stone (plain and relaxed).
  • Pace. Unhurried, Natural or Brisk.
  • Interviewer name. Optional. The interviewer introduces itself with this name, for example "Maya".
  • Start and end tones. Soft tones when the conversation starts and ends.

The orb takes its colour from the tone and the voice.

What respondents need

  • A recent Chrome, Edge or Safari browser. Other browsers show "Voice needs a recent Chrome, Edge or Safari with a microphone."
  • A working microphone, and permission for the survey to use it. If they block the microphone, they see a message asking them to allow it and press Start again.
  • A quiet place. Background noise and other voices can be picked up as answers.

Respondents press Start to begin and End to stop. While it runs, the orb shows Listening when it is their turn and Thinking while the next question is prepared.

Language

Choose Match the respondent or a fixed language under Language. On a published survey, how well speech is recognised in a given language depends on the respondent's browser and device, so test with someone who speaks it before you launch.

Results

Spoken answers are saved as text. In Results, a voice interview appears as a transcript of questions and answers, the same as a typed one, and it is included in Themes and Analysis and in exports. Audio recordings are not shown in Results.

Before you launch

  • Run the interview in Preview on the device and browser your respondents are most likely to use.
  • Mention in the block's description that the question is spoken and needs a microphone.
  • Tell respondents in your consent text that they will answer out loud. See Consent pages.

Was this article helpful?