supertonic2-tts-unity

Unity 6 + Sentis (com.unity.ai.inference) demo for Supertonic TTS 2 ONNX inference.

Reference: https://huggingface.co/onnx-community/Supertonic-TTS-2-ONNX

WebGPU Demo

https://huggingface.co/spaces/Sky-Kim/supertonic2-tts-unity-demo

Environment

• Unity: 6000.3.6f1
• Sentis: com.unity.ai.inference 2.5.0

Project Structure

• Assets/Scenes/SampleScene.unity: Demo scene
• Assets/Supertonic2/Supertonic2TTSRunner.cs: Core TTS runner (warmup, inference, clip generation/playback)
• Assets/Supertonic2/Supertonic2TTSUITest.cs: Legacy uGUI demo controller (Dropdown/InputField/Button)
• Assets/Supertonic2/Models/: ONNX model assets
• Assets/StreamingAssets/Supertonic2/: tokenizer/config/voice files

Quick Start

1. Open Assets/Scenes/SampleScene.unity.
2. Select the GameObject with Supertonic2TTSRunner.
3. Assign all 3 ModelAsset fields:
   • Text Encoder Model
   • Latent Denoiser Model
   • Voice Decoder Model
4. Verify Assets/StreamingAssets/Supertonic2 files exist.
5. Enter Play mode.
6. Wait until warmup is finished.
   • Generate button stays disabled until warmup completes.
7. Select Language / Voice / Step and click Generate.

Notes

• Supported languages: English (en), Korean (ko), Spanish (es), Portuguese (pt), French (fr)
• Supported voices: F1-F5, M1-M5