((w))ebinaric

Search documentation

Search the Webinaric docs

Start free

Presenters, moderators and voices

Create personas for the presenter and moderator roles, set their rules and AI badge, and choose an OpenAI or ElevenLabs voice.

On this page

A persona is an AI host: a name, a face, a voice and a way of speaking. Every webinar uses two different personas, one as the presenter and one as the moderator:

  • The presenter talks over the slides and answers questions out loud.
  • The moderator welcomes attendees, runs the chat, answers quick questions directly and lines up the rest for the presenter. It also speaks the welcome and outro.

Personas belong to the workspace, so you can reuse them across webinars. New organizations start with one presenter and one moderator. Pick them for a webinar when you create it or under Settings → Hosts.

Create or edit a persona

Go to Personas and click New persona, or open an existing one. Click Create persona or Save changes when you're done.

The presenter of the example webinar.

Profile

  • Avatar: click the avatar to upload a photo (PNG, JPEG or WebP), or use initials and a colour.
  • Name: shown to attendees in the player and chat.
  • Description: role and expertise, e.g. "Content strategist with 12 years in B2B marketing."
  • Knowledge bases: extra knowledge only this persona has. It's merged into the persona's prompt in every webinar that uses it. For knowledge both hosts share, attach knowledge bases to the webinar instead; see Rules and knowledge.
  • Show AI badge: on by default. It shows an "AI" badge next to the persona's name in captions, chat and the transcript.

Personality & style

Rules are free-text instructions the persona follows when speaking and chatting, for example "Speak naturally, like a live presenter" or "Never discuss pricing; offer a call with sales instead". The wand writes or enhances them once the persona is saved. See Rules and knowledge for what to put where.

Voices

Under Voice, choose a provider:

  • Voice: one of OpenAI's voices.
  • Model: the text-to-speech model.
  • Speaking style instructions: how the voice should sound, e.g. "Warm and upbeat, medium pace."

Click Test voice to hear the persona introduce itself. Webinar content can be in any language the voice supports.

Changing a persona's voice marks its rendered audio as outdated in every webinar that uses it. Render the audio again, or let Publish do it; see Scripts and presenter context.

Your own ElevenLabs key

To use your own ElevenLabs account, including cloned and custom voices, add your API key under Settings → Voices. With a key, the voice picker lists your account's voices instead of ours, and ElevenLabs bills your account for them.

With your own key, each ElevenLabs persona also gets a Fallback voice. It's used if your key runs out of ElevenLabs credits during a webinar. Without a fallback, that host goes silent.

If you remove your key, voices switch back to ours. Personas that use a voice from your account fall back in live sessions, and rendering new audio for them fails until you pick another voice.

Video avatar

Video avatar holds a reference photo for generating a talking-head intro video from the moderator's welcome. This is experimental; see Intro and outro.

When changes reach attendees

A published version keeps a copy of its personas. Edits to a persona reach attendees after you publish the webinars that use it again.