Fix: CUDAExecutionProvider was not being used to infer Style-Bert-VITS2 ONNX models even though CUDA was available

style_bert_vits2/utils/__init__.py

@@ -0,0 +1,15 @@
+from typing import Any, Sequence, Union
+
+
+def torch_device_to_onnx_providers(
+    device: str,
+) -> Sequence[Union[str, tuple[str, dict[str, Any]]]]:
+    if device.startswith("cuda"):
+        # cudnn_conv_algo_search を DEFAULT にすると推論速度が大幅に向上する
+        # ref: https://medium.com/neuml/debug-onnx-gpu-performance-c9290fe07459
+        return [
+            ("CUDAExecutionProvider", {"cudnn_conv_algo_search": "DEFAULT"}),
+            ("CPUExecutionProvider", {}),
+        ]
+    else:
+        return ["CPUExecutionProvider"]