AURA51.aiBetaRequest a pilotRequest

How does an AI avatar get onto the stage?

In short

Through the mixing desk. The stage microphones feed our laptop, speech recognition transcribes, and when the moderator calls the AI avatar by name, it answers on the projection screen or a display. We run the technology at the event ourselves. Plan on about two weeks of lead time, mostly for coordinating with your sound team. The briefing can come last.

Request a pilot

Last updated 2 October 2026

  1. 01

    How does the moderator direct the AI avatar?

    Through a tablet. Every answer from the AI avatar shows up there as text, usually just over a second before the room hears what it says. One tap on STOP ends the answer. The moderator can also hand it the last sentence someone said as a question, or type one herself. It never speaks on its own.

  2. 02

    Where are the limits of an AI avatar on stage?

    Some things the AI avatar leaves out on purpose. It does not speak unprompted or make up numbers. Other things the technology does not allow. The round with it runs in standard German and only with internet. Its first word comes 1.5–2 s after a question, measured live; later if someone tacks on a second question.

  3. 03

    What does the sound engineer need for an AI avatar?

    Your sound engineer provides the moderator’s microphone as a separate channel and an aux send with the other stage microphones, pre-fader and without the avatar return. Add a return path for its voice and a screen or monitor for its picture. We bring the laptop with audio interface and the tablet and run them ourselves. Internet is a must.

01Signal path

The path to the screen starts with a pause

Whoever moderates can take the time to finish the thought. The laptop releases an answer only once there is a brief quiet after the call. If a sentence sounds unfinished, it waits a moment for the rest. If the moderator keeps talking, it waits too, and a request to wrap up becomes part of the question.

It is built that way because a fragment was once enough. A “Hey”, its name, a “no”, and the avatar started answering right away. Its bare name alone had the same effect. When someone wanted to wrap up, it was already talking.

  1. 01

    From the desk into the laptop

    The moderator’s microphone comes into our laptop on a separate channel, the other stage microphones via an aux send on the mixing desk. The AI avatar’s own voice is on neither path.

  2. 02

    Transcribing

    Speech recognition transcribes every sentence and keeps the speakers apart. The transcript runs along on the moderator’s tablet.

  3. 03

    Is it being called on?

    If a sentence opens by addressing it by name, it is being called on. Otherwise a language model decides whether the name was a call or just a reference to it.

  4. 04

    The tablet first

    A language model composes the answer. It goes out as text to the avatar and to the moderator’s tablet, and shows up there before the avatar says it.

  5. 05

    Through the desk into the room

    The avatar speaks the answer on screen. Its voice comes back into the desk on a return line, and your sound team puts it on the PA.

Allow for about two seconds from the last word of the call to the first audio. The pause the laptop waits for is already part of that.

02Preparation

The two weeks belong to your sound team

Allow about two weeks between request and event. Nearly all of that time goes into coordinating with your event technicians. The avatar itself is set up in minutes once the briefing is in. That is why the briefing can wait until everything else is in place. It has to be there by the dress rehearsal.

  1. 01

    Request

    We hold the date and settle which character comes. Name the one you have in mind, or tell us about your discussion and we will recommend one. That is decided before you invite the guests and the moderator. At the same time you decide whether its image carries a permanent AI avatar label.

  2. 02

    Coordination with your technicians

    With your sound team we work out how your desk connects to our audio interface, one way for the stage microphones and back for its voice. That is what the two weeks are for.

  3. 03

    Voice sample

    Whoever calls on it at the event may pronounce its name differently than we do. A short recording is there so that it recognizes its name however it is pronounced.

  4. 04

    Briefing

    The topic, the participants and the position it should argue. The names of the moderator and the guests are entered on the tablet; that is how it addressed the guests correctly in our live tests.

  5. 05

    Dress rehearsal

    The whole path runs once, from the stage microphone through your desk to the screen, with the briefing and the names entered. It is the first time the path runs through your desk.

  6. 06

    At the event

    One of us sits at the laptop next to your sound engineer. The moderator only operates the tablet, and the AI avatar speaks only once the moderator calls on it.

For the request, we first need the event and its date. Who sits on stage and which position the AI avatar should take can still be open.