You can deploy this model and start earning money today!

willychenwii/
pig-noise-ast-finetuned

Name: pig-noise-ast-finetuned
Availability: InStock

safetensors

audio-spectrogram-transformer

Views

No views yet

Code to run test on model

python

import torch
from transformers import ASTForAudioClassification, ASTFeatureExtractor
# 1. Configuration - Choose which model to load
repo_id = "MIT/ast-finetuned-audioset-10-10-0.4593" # Use original AST
# repo_id = "willychenwii/pig-noise-ast-finetuned" # Finetuned model for breathing noise clasiification
# repo_id = "willychenwii/pig-condition-ast-finetuned" # Finetuned model for breathing condition classification
print(f"--- Loading model from Hub: {repo_id} ---")
model = ASTForAudioClassification.from_pretrained(repo_id)
feature_extractor = ASTFeatureExtractor.from_pretrained(repo_id)
# 2. Inspect Model Parameters
total_params = sum(p.numel() for p in model.parameters())
trainable_params = sum(p.numel() for p in model.parameters() if p.requires_grad)
print(f"Total Parameters: {total_params:,}")
print(f"Trainable Parameters: {trainable_params:,}")
print(f"Hidden Layer Size (d_model): {model.config.hidden_size}")
# 3. Dummy Input (10 seconds of silence at 16kHz)
sample_rate = 16000
duration = 10 
dummy_audio = torch.zeros(sample_rate * duration)
# Feature extraction
inputs = feature_extractor(dummy_audio, sampling_rate=sample_rate, return_tensors="pt")
print(f"\nInput Shape (to Transformer): {inputs.input_values.shape}") 
# Expected: [batch, time_frames, freq_bins] -> [1, 1024, 128]
# 4. Inference & Hidden States
model.eval()
with torch.no_grad():
    outputs = model(**inputs, output_hidden_states=True)
# 5. Output Shapes
logits = outputs.logits
last_hidden_state = outputs.hidden_states[-1]
print(f"Output (Logits) Shape: {logits.shape}")         # [1, num_labels]
print(f"Last Hidden Layer Embedding Size: {last_hidden_state.shape}") 
# Expected: [batch, sequence_length, hidden_size] -> [1, 1214, 768]