I've created a Gradio demo app for the MP-SENet model, hosted on HuggingFace Spaces with ZeroGPU support. It allows users to try out the model immediately in their browsers without requiring any local setup.
Additionally, a simple segment feature that splits long audio files into segments has been implemented for the app. For ZeroGPU (which uses an A100 with 40G memory under the hood), the maximum segment length is 10 seconds. When running locally, the segment length is limited to 3 seconds to prevent blowing up the memory on my MBP.
Hello,
I've created a Gradio demo app for the MP-SENet model, hosted on HuggingFace Spaces with ZeroGPU support. It allows users to try out the model immediately in their browsers without requiring any local setup.
Additionally, a simple segment feature that splits long audio files into segments has been implemented for the app. For ZeroGPU (which uses an A100 with 40G memory under the hood), the maximum segment length is 10 seconds. When running locally, the segment length is limited to 3 seconds to prevent blowing up the memory on my MBP.
The demo app is here: https://huggingface.co/spaces/JacobLinCool/MP-SENet And the repository is here: https://github.com/JacobLinCool/MP-SENet-Gradio
Any feedback is welcome!