Spaces:
Sleeping
Apply for a GPU community grant: Personal project
Hi Hugging Face team,
I’m building IRIS, a Gradio demo for searching inside videos by spoken content. Users can upload a video, transcribe it once, and then search the transcript semantically to jump to relevant timestamped moments. I’m also adding real pre-transcribed demo videos so visitors can try the app immediately.
The project currently runs on CPU, but transcription and embedding/index creation are slow enough that the live demo feels blocked for first-time users, especially when processing videos with Whisper. A community GPU would help make the Space responsive enough for demos, testing, and iteration.
GPU would be used for:
- faster Whisper transcription
- faster sentence-transformer embeddings
- optional visual/video frame search experiments
- keeping the public demo usable without long wait times
This is a personal/open demo project intended to be publicly available on Hugging Face Spaces. I’m happy to keep the Space lightweight and use the GPU only for the interactive demo workload.
Thanks for considering the grant!