NEW V4.3 - Audio & Facial Animation Generation
Generate voice audio (ElevenLabs) and facial animation (MetaHuman Animator) from dialogue graph nodes in the editor. Assets are saved to Content and play automatically at runtime — no custom abilities needed.
Flow: Project Settings → Voice Config (per character) → Character setup → Dialogue graph → Generate TTS → Generate Facial Animation
Before you start
Every speaker uses a
Character.*Gameplay Tag — it must match on the character, in DA_VoicesConfig, and on each dialogue node.Voice = per character (Voice Config). Audio + face = per line (graph node).
Facial animation needs audio first — generate TTS (or assign a WAV) before generating the face.
Step 1 — Project Settings
Edit → Project Settings → Plugins → Ascent Dialogue Settings
Voice Generation
Set TTS Voice Gen Key (ElevenLabs API key)
Keep Auto Assign Generated Audio on
Output paths: defaults (
Audio/TTS/Generated/) are fineVoice Config Data Asset → assign here in Step 2 after creating/configuring the asset
Facial Animation
Confirm Visualization Mesh, Output Path, Default MetaHuman Performance (sample defaults work)
Save settings.
Step 2 — Voice Config Data Asset (the important part)
This is where you map each character to an ElevenLabs voice. Project Settings only holds the API key and a pointer to this asset — all per-character voice data lives here.
Where the asset lives
The ACF sample already has one ready to use:
Content/Integrations/ATSIntegrations/Dialogue/DA_VoicesConfigOpen it from the Content Browser. If you're starting from scratch:
Content Browser → right-click → Miscellaneous → Data Asset
Pick class
ADSVoiceConfigDataAssetName it (e.g.
DA_VoicesConfig) and save it somewhere sensible underContent/(e.g.Content/Dialogue/)
Where to plug it in
Go back to Project Settings → Ascent Dialogue Settings → Voice Generation and set Voice Config Data Asset to your asset (DA_VoicesConfig).
This is the only place that links the global dialogue settings to your character voice map. If this field is empty, TTS generation won't know which voice to use for anyone.
Project Settings DA_VoicesConfig (Data Asset)
┌─────────────────────┐ ┌──────────────────────────────┐
│ TTS Voice Gen Key │ │ Voice Configs: │
│ Voice Config DA ────┼──────────► │ Character.Shopkeeper → voice│
│ Final TTS Path │ │ Character.Guard → voice│
└─────────────────────┘ │ Character.Player → voice│
└──────────────────────────────┘
▲
│ lookup by Participant tag
┌──────────┴──────────┐
│ Dialogue graph node │
│ Participant: Character.Shopkeeper
│ Text: "Welcome!"
└─────────────────────┘How to configure the asset
Open DA_VoicesConfig. You need the ElevenLabs API key already set in Project Settings (Step 1).
1. Load the voice list
In the Details panel, category Voice Generation, click Refresh Voices. This fetches all voices from your ElevenLabs account. The dropdown won't work until you do this.
2. Map characters to voices
Expand Voice Configs — this is a map of Gameplay Tag → Voice Settings. Add one entry per speaking character:
Click + to add a row
Key (left side): the character's participant tag, e.g.
Character.ShopkeeperMust be a tag under the
Charactercategory (same tag you'll use on the character and on dialogue nodes)You can pick it from the tag dropdown or type it — but it must exist in your project's Gameplay Tags (
Project Settings → Gameplay Tags)
Value (right side): expand it and set Selected Voice Name from the dropdown
Voice ID fills in automatically once you pick a name
If your voice isn't in the list, use Manual Voice ID instead (copy the ID from your ElevenLabs dashboard)
Example setup for three speakers:
Character.Shopkeeper → "James - Professional British Male"
Character.Guard → "Adam - Dominant Male"
Character.Player → "Rachel - Calm Female"3. Save the asset
Every time you add or change an entry, the Voice IDs update automatically.
How the lookup works at generation time
When you click Generate TTS on a dialogue node:
The node reads its Participant tag (e.g.
Character.Shopkeeper)The system opens the Voice Config Data Asset assigned in Project Settings
It looks up that tag in Voice Configs
It uses the matched Voice ID to call ElevenLabs with the node's Text
If the tag isn't in the map → "Dialogue node has no voice config". If the tag on the node doesn't match the character → wrong voice or no voice.
Verify before moving on
Preview Voice (on the asset) — hear a sample of the selected voice
Clear Cache — if you changed API keys and voices aren't updating
Confirm every character who will speak has an entry in Voice Configs
Confirm the Participant Tag you'll set on characters (Step 3) and dialogue nodes (Step 4) matches the map keys exactly
Step 3 — Character setup
On the Character Data Asset, add ACFDialogueFragment:
Partecipant Tag → same tag as in Voice Config (e.g.
Character.Shopkeeper)Dialogues → your dialogue asset(s)
Optional: Default Camera Config
On the character Blueprint:
Tag the face skeletal mesh with
head(matches Facial Skeleton Component Tag on the participant component)
Without the face mesh tag, facial montages won't play at runtime.
Step 4 — Dialogue graph
Open your dialogue asset. For each line:
Participant → speaker's
Character.*tagText → line content (UI + TTS input)
ADS|Facial Animation → mood / blink settings (used at generation time)
Animation → optional body montage (manual, no generator)
Leave Sound To Play and Facial Animation empty — Steps 5 & 6 fill them.
Tip: Export / Import Dialogue to Text on the asset to build the graph from a screenplay file.
Step 5 — Generate TTS
Select a dialogue node
Details → AI Generation → Generate TTS
Repeat per line, save project
Audio lands in Content/Audio/TTS/Generated/ and auto-assigns to Sound To Play.
Step 6 — Generate Facial Animation
Requires Sound To Play (from Step 5 or a manual WAV).
Select node, set Facial Anim Mood if needed
AI Generation → Generate Facial Animation
Repeat per line, save often (slow — runs in editor)
Montage auto-assigns to Facial Animation. Skip this step for voice-only lines.