Actions3
Video → Streaming Inference
AI-generatedSummary
Generate an AI avatar video in real-time by streaming video chunks as the avatar speaks the provided text with specified emotion, avatar character, and language.
Inputs
- Text (required) — The text to be spoken by the avatar.
- Avatar (required) — The avatar character to use (e.g., African Man, Japanese Woman).
- Emotion (required) — The emotion or facial expression of the avatar (Happy, Neutral, Serious).
- Language — Optional language code for speech synthesis (default is 'en' for English).
Output shape
A single binary file containing the streamed avatar video in MP4 format.
The video is returned as binary data under the 'video' property with MIME type 'video/mp4'. The output includes an empty JSON object. Errors throw node exceptions unless 'Continue On Fail' is enabled.