document hardware requirements

Dockerfile

@@ -21,8 +21,6 @@ COPY requirements.txt /app/requirements.txt
 RUN pip install --no-cache-dir gradio==5.28.0
 RUN pip install --no-cache-dir -r /app/requirements.txt
 
-RUN python -c "from audiosr import build_model; build_model(model_name='basic', device='cpu')"
-
 COPY app.py /app/app.py
 
 ENV PORT=7860

README.md

@@ -8,4 +8,18 @@ app_port: 7860
 pinned: false
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# AudioSR — HARP Endpoint
+
+Deploys [AudioSR](https://github.com/haoheliu/versatile_audio_super_resolution) as a [PyHARP](https://github.com/TEAMuP-dev/pyharp) endpoint, accessible from within DAWs (Logic Pro, Reaper, Ableton) via the [HARP plugin](https://github.com/TEAMuP-dev/HARP).
+
+## What it does
+
+Upsamples any audio to 48kHz using audio super-resolution. Files already at 48kHz or above are returned unchanged.
+
+## Usage
+
+Load this Space's URL into HARP as a custom endpoint. Upload an audio file and adjust DDIM Steps (more steps = better quality, slower processing).
+
+## Hardware Requirements
+
+This Space requires at minimum a **T4 Small** GPU instance due to the AudioSR model size (~6GB). It will time out on free CPU hardware.