Describe the feature request
Like #23306, but for C#.
I hope we can set the seed used in onnx's PRNG from our C# code when running onnx models.
Describe scenario use case
I am a contributor to OpenUtau singing synthesizer's AI synthesis functionality. It uses a diffusion model, which produces different result each time. Currently we cache the generated audio on the user's PC to ensure the audio remained unchanged when they opens the project file next time. However, the cache is very large. Also the same file won't produce the same output on another user's PC. So I hope that the user can manually specify a random seed to produce deterministic output audio.
Describe the feature request
Like #23306, but for C#.
I hope we can set the seed used in onnx's PRNG from our C# code when running onnx models.
Describe scenario use case
I am a contributor to OpenUtau singing synthesizer's AI synthesis functionality. It uses a diffusion model, which produces different result each time. Currently we cache the generated audio on the user's PC to ensure the audio remained unchanged when they opens the project file next time. However, the cache is very large. Also the same file won't produce the same output on another user's PC. So I hope that the user can manually specify a random seed to produce deterministic output audio.