Feature request - Support CUDA 12.x with onnxruntime-gpi 1.21.x

Feature

Support CUDA and onnxruntime to run the silero models on GPU. It works on CPU but fails on GPU currently.

Motivation

The motivation is to support the newest available infrastructure.

Pitch

Provide the onnx model which makes this possible. Unfortunately i can not export it myself alone with just the jit model.

Alternatives

Not really there, since no everyone can change the environment to support this.

Additional context