Upload 2 files

README.md

@@ -1,12 +1,80 @@
 ---
-title: Dl GenAI Project
-emoji: 📊
-colorFrom: gray
-colorTo: red
+title: Messy Mashup AST Genre Classifier
+colorFrom: blue
+colorTo: green
 sdk: gradio
-sdk_version: 6.11.0
+python_version: "3.10"
 app_file: app.py
-pinned: false
+suggested_hardware: cpu-basic
+fullWidth: true
+header: default
+short_description: Fine-tuned AST genre classifier.
+models:
+  - MIT/ast-finetuned-audioset-10-10-0.4593
+tags:
+  - audio-classification
+  - music
+  - gradio
+  - zero-gpu
+  - transformers
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Messy Mashup AST Genre Classifier
+
+This folder is ready to use as the root of a CPU-only Hugging Face Space.
+
+It contains:
+
+- `app.py`: Gradio app with CPU-only inference
+- `working_ast_best.pth`: your fine-tuned checkpoint
+- `config.json` and `preprocessor_config.json`: local AST config so the Space does not need to fetch them at startup
+- `requirements.txt`: Python dependencies for the Space
+- `.gitattributes`: Git LFS rules for the checkpoint
+
+## What The App Does
+
+The app accepts an uploaded audio clip or microphone recording and predicts one of these 10 genres:
+
+- blues
+- classical
+- country
+- disco
+- hiphop
+- jazz
+- metal
+- pop
+- reggae
+- rock
+
+The preprocessing follows your AST notebook flow:
+
+1. Convert to mono
+2. Resample to 16 kHz
+3. Pad or crop to 10 seconds
+4. RMS normalize and clamp
+5. Run the AST feature extractor
+6. Predict the genre with the fine-tuned checkpoint
+
+## Deploy To Hugging Face Spaces
+
+1. Create a new Hugging Face Space and choose the `Gradio` SDK.
+2. Copy the contents of this folder into the Space repository root.
+3. Make sure `git-lfs` is installed before pushing because `working_ast_best.pth` is large.
+4. In the Space settings, use `CPU Basic` or `CPU Upgrade`.
+
+## CPU Deployment Notes
+
+This app is set up for standard CPU execution:
+
+- no `spaces` package is required
+- no `@spaces.GPU` decorator is used
+- the model is loaded on CPU and inference runs on CPU only
+
+For a first deployment, this is simpler than ZeroGPU and avoids queueing and GPU quota issues.
+
+## Local Run
+
+```bash
+pip install -r requirements.txt
+python app.py
+```

working_ast_best.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e2bdd925462b7f57d1ea0695736f865db331f517453fd20d45e46296ac8be9e
+size 344878330