> ## Documentation Index
> Fetch the complete documentation index at: https://docs.listen.land/llms.txt
> Use this file to discover all available pages before exploring further.

# Text vs Voice: Pick the Right Mode for Your Interview

> ListenLand supports text and voice interview modes. Understand the differences, tradeoffs, and which mode best fits your research goals.

ListenLand supports two interview modes — **text** and **voice** — and the choice shapes both the respondent's experience and the data you get back. Text mode is lower friction and works anywhere; voice mode produces richer, more spontaneous responses and includes audio recordings. You set a default per interview, and respondents can optionally switch before they start.

<Tabs>
  <Tab title="Text Mode">
    In text mode, respondents type their answers and the AI replies in text — a back-and-forth written conversation in their browser.

    **How it works**

    * Respondents type; the AI reads and responds in text
    * No microphone or special hardware required
    * Works in any modern browser on any device

    **Best for**

    * Written feedback and long-form qualitative research
    * Async responses — respondents can take their time and come back
    * Anonymous respondents or low-trust contexts where asking for mic access adds friction
    * Surveys distributed broadly (email lists, communities, public links)

    **Transcript quality**
    Because everything is typed, the transcript is always 100% accurate — there is no speech recognition involved and nothing is lost to audio quality.

    **Friction**
    Text mode has the lowest barrier to entry. Respondents don't need to grant any browser permissions, don't need a quiet environment, and can complete the conversation on mobile just as easily as desktop.
  </Tab>

  <Tab title="Voice Mode">
    In voice mode, respondents speak their answers and the AI responds in natural speech — a live, real-time audio conversation that runs entirely in the browser.

    **How it works**

    * Respondents speak; the AI listens and replies in natural speech
    * Runs entirely in the browser — no app, plugin, or download required
    * Low-latency responses make the conversation feel natural and fluid
    * Native interruption handling: the AI pauses and listens if the respondent speaks over it

    **Best for**

    * Research where emotional tone and spontaneity matter
    * Faster, more fluid responses — speaking is quicker than typing
    * Usability research, customer discovery, and exploratory interviews
    * Contexts where you want the audio recording alongside the transcript

    **AI voice**
    The AI's voice can be customized per interview. Available voices include: Alloy, Ash, Ballad, Cedar, Coral, Echo, Marin, Sage, Shimmer, and Verse.

    **Transcript**
    A full text transcript is automatically captured from the voice conversation — you get both the audio recording and the written record side by side.

    **Duration cap**
    Voice conversations have a hard cap of **20 minutes** per session. This keeps costs predictable and encourages focused conversations.

    **Audio recordings**
    Three audio tracks are stored for every voice conversation: the respondent's audio, the AI's audio, and a combined mixed track for easy playback.
  </Tab>
</Tabs>

## Choosing a default

Every interview has a **Default Mode** setting — the mode respondents land in when they open the link. You set this when creating or editing the interview.

If you want to give respondents a choice, you can allow them to select their preferred mode on the start screen before the conversation begins. This is useful when you're reaching a mixed audience (some on desktop with a mic, some on mobile without).

<Note>
  Voice mode requires a modern browser (Chrome, Firefox, or Safari 17+) and microphone permission granted in the browser. Some corporate or restricted networks may prevent voice mode from connecting; text mode works as a reliable fallback in those environments.
</Note>

<Tip>
  Not sure which to pick? Start with text mode — it has no device requirements and is easier for respondents to complete asynchronously. You can always create a separate voice-mode interview once you've validated your questions.
</Tip>