Monkeypatch lowpass_filtering_prepare_inference boundary bug

Resample low-sample-rate input to 48kHz before inference

add multiband ensemble control, clean up debug prints

add torchcodec

add server name

add startup debug prints

add startup debug prints

add startup debug prints

add startup debug prints

add startup debug prints

lazy load model on first request

add flush=True to print statements

add gitignore

use super_resolution_long_audio from repo

remove latent_t_per_second hardcode, use library default

add guidance_scale and seed controls, fix author, fix ddim steps range

document hardware requirements

pre-download model weights at build time

remove numpy pin

add git to dockerfile

switch to docker

Update README.md

fix numpy version

initial commit

initial commit