Audio Classification
Transformers
ONNX
English
audio
music
audio-spectrogram-transformer
linear-probe
Eval Results (legacy)
Instructions to use kaaaaan/live-vs-studio-classifier with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use kaaaaan/live-vs-studio-classifier with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("audio-classification", model="kaaaaan/live-vs-studio-classifier")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("kaaaaan/live-vs-studio-classifier", dtype="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "architecture": "ast-backbone + linear-probe", | |
| "backbone": { | |
| "type": "AST", | |
| "hf_id": "MIT/ast-finetuned-audioset-10-10-0.4593", | |
| "embedding_dim": 768, | |
| "params_M": 86.2, | |
| "onnx_quantized_file": "ast_backbone_int8.onnx" | |
| }, | |
| "probe": { | |
| "type": "linear", | |
| "in_dim": 768, | |
| "out_dim": 2, | |
| "labels": ["studio", "live"], | |
| "weights_file": "probe-weights.json", | |
| "params": 1538 | |
| }, | |
| "preprocessing": { | |
| "sample_rate": 16000, | |
| "clip_duration_sec": 30, | |
| "n_windows": 3, | |
| "window_positions": [0.25, 0.5, 0.75], | |
| "voting": "majority" | |
| }, | |
| "evaluation": { | |
| "auto_split_test_acc": 0.9446, | |
| "auto_split_test_n": 2401, | |
| "adversarial_source_quality_acc": 0.9393, | |
| "adversarial_source_quality_n": 2043 | |
| }, | |
| "version": "1.0.0" | |
| } | |