VoxCPM2 Burmese ONNX Final
VoxCPM2 Burmese Text-to-Speech pipeline with ONNX DiT acceleration.
Features
- Burmese Text-to-Speech
- VoxCPM2 based voice generation
- ONNX Runtime DiT inference
- 48kHz audio output
- Long text generation support
Pipeline
Text
↓
VoxCPM2 Language Model
↓
UnifiedCFM
↓
ONNX DiT Runtime
↓
AudioVAE Decoder
↓
48kHz WAV
ONNX Validation
PyTorch FP32 vs ONNX FP32:
