New

Now in Claude, ChatGPT, Cursor & more with our MCP server

Back to docs
Interview Experience

Voice Interview Experience

What participants see and hear during a voice interview — from microphone permission to natural conversation.

Voice interviews produce richer, more natural responses because participants can speak freely rather than typing. Here is a complete walkthrough of what participants experience from the moment they choose voice mode to the final thank-you screen.

Starting a Voice Interview

When a participant arrives at your interview landing page, they see two separate start buttons — Start Voice Chat and Start Text Chat. The participant clicks Start Voice Chat to begin.

If you have configured your project to offer only voice mode, the text option does not appear and participants go straight to the next step.

Microphone Permission

Before the conversation begins, the browser prompts the participant to grant microphone access. This is a standard browser permission dialog that looks slightly different on each browser and operating system.

A few things to note:

  • First-time visitors will always see this prompt. Once they grant permission, the browser remembers it for future visits.
  • HTTPS is required. Microphone access only works on pages served over a secure connection. If your interview link uses HTTP, the browser will block the request.
  • If permission is denied, the participant is offered the option to switch to text mode instead.

The Conversation

Once microphone access is granted, the interview begins immediately. Here is what the participant experiences:

The AI Interviewer Speaks First

The conversation opens with a warm greeting from the AI interviewer. The participant hears a natural-sounding voice that introduces the topic and asks the first question. The AI uses ElevenLabs Conversational AI for lifelike, expressive speech.

Always-On Microphone

There is no push-to-talk button. The microphone is always listening, and the system automatically detects when the participant is speaking and when they have finished. It feels like a real phone call or video chat — just talk, pause, and the interviewer responds.

Real-Time Follow-Up Questions

The AI interviewer listens carefully and asks follow-up questions based on what the participant says. If someone mentions an interesting detail, the interviewer probes deeper. If a response is vague, the interviewer asks for clarification. This dynamic back-and-forth is what makes voice interviews so effective for qualitative research.

Visual Feedback

While the conversation is happening, the participant sees:

  • An animated orb that responds to audio state — it pulses and moves when the interviewer or participant is speaking
  • A mute button for moments when the participant needs to cough, speak to someone else, or take a brief pause
  • Conversation messages displayed on screen alongside the orb, so participants can follow along with the text of the conversation as it happens

Messages Are Visible

Unlike a phone call, voice interviews in Koji display the conversation messages on screen in real time. Participants can see both what the AI interviewer said and what they said, which provides helpful context and makes it easy to reference earlier parts of the conversation.

Switching Between Modes

At any point during a voice interview, participants can switch to text mode. This is helpful if:

  • Their microphone stops working mid-conversation
  • They move to a noisy environment
  • They simply prefer typing for a particular answer

The switch is seamless — the conversation history carries over, and the AI interviewer picks up right where it left off, now in text form. Participants can also switch back to voice mode if they prefer. See Text Interview Experience for details on the text interface.

Interview Duration

Voice interviews tend to run faster than text interviews because speaking is quicker than typing. A typical voice interview takes around 10 minutes, though this depends on the complexity of your research questions and how much the participant has to share.

The AI interviewer manages the pacing automatically. It will cover all the key topics in your research brief and naturally wind down the conversation when enough ground has been covered.

Ending the Interview

Automatic Completion

When the AI interviewer determines that it has gathered sufficient responses across all the topics in your research brief, it wraps up the conversation naturally — thanking the participant and saying goodbye. The AI can also trigger an automatic end-of-interview signal when all research questions have been covered.

Manual Completion

Participants can end the interview at any time by clicking the Done button in the header. Either way, the conversation moves to the completion flow.

Audio Quality Tips

To help your participants have the best experience, consider sharing these tips in your outreach:

  1. Use headphones. This prevents echo and improves audio clarity for the AI interviewer.
  2. Find a quiet space. Background noise can interfere with speech detection.
  3. Use a stable internet connection. Voice interviews require real-time audio streaming, so a strong connection helps avoid interruptions.
  4. Use Chrome or a Chromium-based browser. These tend to have the best support for real-time audio features.

What Researchers See

As the study owner, you do not hear the interview in real time. Instead, you see completed interviews in your project dashboard, each with:

  • A full text transcript of the conversation
  • A quality score assigned by Koji's analysis
  • AI-generated insights and themes

Voice and text interviews produce identical output in your dashboard — the transcript, score, and insights look the same regardless of which mode the participant used.

Next Steps

Further reading on the blog

<!-- further-reading:blog -->