snnithya/GaMaDHaNi · Apply for community grant: Academic project (gpu)

I am a first year PhD student at MIT, and am trying to build an interactive generative model for the Hindustani (North Indian Classical) vocal music. I am interested in making it interactive and would love to have people play with it and give me feedback about their experiences. I have a paper accepted in ISMIR 2024 about the generative model (https://arxiv.org/abs/2408.12658). More examples of samples generated from the model are available here: https://snnithya.github.io/gamadhani-samples/. We are in the process of open-sourcing the code now and I thought a huggingface spaces + gradio demo would be a great way to make this work accessible to people from any background. Unfortunately the model runs too slowly on the free CPU tier (takes > 5min to process a few seconds of audio). This work is trying to develop generative models for an underrepresented form of music in the generative AI space, any form of help would be greatly appreciated!!