Tavus

Tavus creates personalized videos from Personas (AI avatars) and Replicas (voice clones). Speeches are scripts; Videos are rendered outputs. Conversations track viewer interactions.

Data Model

• Personas — AI video avatars.
• Replicas — voice clones from sample audio.
• Speeches — scripts for video generation.
• Videos — rendered personalized videos.
• Conversations — viewer interactions with videos.

Gotchas

• Replica training: Replicas require training from voice samples; creation is async.
• Persona + Replica: Videos typically use a persona and replica; ensure both exist before generating.