Understand how Dialogue Animation selects clips, drives the four-layer animator stack, and uses emotion and dialogue state to animate AI characters.
The Dialogue Animation module drives body and head gesture animations on AI characters using dialogue state and detected emotion as inputs. Instead of hard-coded animator states for every clip, it uses an AnimatorOverrideController to inject clips from a pooled library at runtime — swapping candidates as dialogue progresses. The module runs entirely inside Unity. No animation data is sent to Convai.
Each frame the module reads two signals: the current dialogue state (idle, listening, thinking, speaking, reacting) and the current emotion reading from the Emotion module. It uses those signals to select and crossfade clips from a DialogueAnimationLibrary across a four-layer animator stack.
The four layers are additive and masked: the base idle plays continuously, the idle overlay adds variation, and the two talk layers fade in when the character speaks. Talk layer clips are routed by DialogueTalkBodyCoverage — clips can target head only, body only, or both simultaneously.
The module is driven by four ScriptableObject types:
DialogueAnimationLibrary
Pool of idle and talk clips, each tagged with emotion affinity and character gender
DialogueAnimationRuntimeConfig
Timing, blend durations, layer weights, and selection parameters
DialogueAnimatorContract
Maps layer indices and state names to your Animator Controller
ConvaiDialogueAnimationProfile
Bundles the three assets above into a single character preset
Three bundled profiles ship with the SDK: Balanced, Expressive, and Subtle. They cover most characters without custom authoring.
Follow the Quick Start to get the module running on your first character, then read Animation Libraries & Profiles to understand clip pools, timing, and the full field reference.
Dialogue Animation quick startAnimation libraries and profilesLast updated
Was this helpful?
Was this helpful?
