Hedra (Character-3)
Audio-driven avatar avec expressivité supérieure, 20M+ utilisateurs, $0.05/min Live API
Latence TTFR
~500ms
temps réel
Coût / minute
à partir de $0.000/min
selon l'abonnement souscrit
Qualité visuelle
8/10
score estimé
Protocoles
WebRTC, REST, LiveKit
Direction émotionnelle & mise en scène
Le rendu suit surtout l’audio expressif fourni ; aucune commande d’émotion, geste ou intention visuelle n’est documentée pour l’API.
Pas de contrôle expressif documentéParamètres et mécanisme
`audio`, `start_image`, `prompt`, `aspect_ratio`, `resolution`, `duration_ms`.
Limite à garder visible
Le classer comme audio-driven : il dépend d’un TTS expressif ou d’un audio préproduit, et non d’une régie live.
Personnalisation de l'avatar
RAG / Base de connaissances
External knowledge base connection via API. Omnia API for enterprise brand consistency. Behavior driven by input prompt.
Comportement & personnalité
Tone, personality via input prompt and voice selection. Omnia API for enterprise persona consistency.
Langage corporel & gestes
Procedural AI-generated body language synchronized with audio. No manual gesture scripting.
Expressions faciales
Character-2/3: dynamic micro-expressions auto-synchronized with audio. Superior emotional expressivity.
Voix & clonage vocal
Large voice library + voice cloning (audio sample: 30–60s WAV/MP3). Speed and emotion adjustments.
Fine-tuning du persona
Omnia API for enterprise brand consistency. Persona via prompt + voice selection.
Entraînement de l'avatar
Bonnes pratiques
- 01.Front-facing portrait, uniform lighting
- 02.Neutral background
- 03.For voice cloning: 30–60s clean audio (WAV or MP3, no background noise)
- 04.Enterprise: 2–5 min reference video for full character model
Analyse API
Protocoles
SDKs
Sessions simultanées
Plan-dependent
Limites de débit
10-minute max session. Credits don't roll over month-to-month.
Fonctionnalités clés
- GET /public/models — list available models
- GET /public/voices — list voices
- POST /public/assets — create asset
- GET /public/generations/{id}/status — streaming_url in response
- Hedra Live Avatars: $0.05/min real-time mode
- Character-3: 6 credits/second (360 credits/min)
Contraintes API
- 10-minute max session duration
- Credits don't roll over
- No webhook support
- US hosting only
- No on-premise standard option
Modèle tarifaire
| Plan | Prix | Minutes incluses | Dépassement |
|---|---|---|---|
| Free | $0/mo | 3 min/mo video | N/A |
| Basic | $15/mo | 1500 credits (~4 min Character-3) | ~$0.40/min |
| Creator | $30/mo | 5400 credits (~15 min Character-3) | ~$0.40/min |
| Professional | $75/mo | 14400 credits (~40 min Character-3) | ~$0.40/min |
| Live Avatars | Usage-based | $0.05/min | $0.05/min |
Coûts cachés / à surveiller
- Credits don't roll over
- Character-3 consumes 6 credits/second (very fast consumption)
- Enterprise private deployment on request
Souveraineté & Hébergement
Score de souveraineté
Hébergement
AWS US
GDPR
YesOn-premise
NoDétail souveraineté
US company. AWS US. GDPR principles respected. No EU hosting, no on-premise standard.
Contraintes & Limites
- 10-minute max session duration
- Credits don't roll over month-to-month
- No manual gesture scripting
- US hosting only
- No on-premise standard option
- No webhook support
Pertinence pour GamiWays
Score
7/10
Lowest cost real-time option ($0.05/min via Live Avatars). Character-2/3 expressivity is excellent. One-shot from image enables rapid prototyping. Main limitations: US hosting, 10-min session cap, and no on-premise option.