Update README.md

README.md

@@ -36,4 +36,102 @@ model-index:
         value: 0.9692
       - type: recall
         value: 0.9728
----
+---
+
+# AST Fine-tuned for Fake Audio Detection
+
+This model is a binary classification head fine-tuned version of [MIT/ast-finetuned-audioset-10-10-0.4593](https://huggingface.co/MIT/ast-finetuned-audioset-10-10-0.4593) for detecting fake/synthetic audio. The original AST (Audio Spectrogram Transformer) classification head was replaced with a binary classification layer optimized for fake audio detection.
+
+## Model Description
+
+- **Base Model**: MIT/ast-finetuned-audioset-10-10-0.4593 (AST pretrained on AudioSet)
+- **Task**: Binary classification (fake/real audio detection)
+- **Input**: Audio converted to Mel spectrogram (128 mel bins, 1024 time frames)
+- **Output**: Binary prediction (0: real audio, 1: fake audio)
+- **Training Hardware**: 2x NVIDIA T4 GPUs
+
+## Training Configuration
+
+```python
+{
+    'learning_rate': 1e-5,
+    'weight_decay': 0.01,
+    'n_iterations': 10000,
+    'batch_size': 8,
+    'gradient_accumulation_steps': 8,
+    'validate_every': 500,
+    'val_samples': 5000
+}
+```
+
+## Dataset Distribution
+
+The model was trained on [012shin/fake-audio-detection-augmented](https://huggingface.co/datasets/012shin/fake-audio-detection-augmented) dataset with the following class distribution:
+
+```
+Training Set (80%):
+- Real Audio (0): 43,460 samples (63.69%)
+- Fake Audio (1): 24,776 samples (36.31%)
+
+Test Set (20%):
+- Real Audio (0): 10,776 samples (63.17%)
+- Fake Audio (1): 6,284 samples (36.83%)
+```
+
+## Model Performance
+
+Final metrics on validation set:
+- Accuracy: 0.9662 (96.62%)
+- F1 Score: 0.9710 (97.10%)
+- Precision: 0.9692 (96.92%)
+- Recall: 0.9728 (97.28%)
+
+## Usage
+
+Here's how to use the model:
+
+```python
+import torch
+import torchaudio
+from transformers import AutoFeatureExtractor, AutoModelForAudioClassification
+
+# Load model and processor
+model = AutoModelForAudioClassification.from_pretrained("your-username/ast-fakeaudio-detector")
+processor = AutoFeatureExtractor.from_pretrained("your-username/ast-fakeaudio-detector")
+
+# Load and preprocess audio (ensure 16kHz sampling rate)
+audio_path = "path/to/audio.wav"
+waveform, sample_rate = torchaudio.load(audio_path)
+if sample_rate != 16000:
+    resampler = torchaudio.transforms.Resample(sample_rate, 16000)
+    waveform = resampler(waveform)
+
+# Process audio
+inputs = processor(waveform, sampling_rate=16000, return_tensors="pt")
+
+# Get prediction
+with torch.no_grad():
+    outputs = model(**inputs)
+    probabilities = torch.sigmoid(outputs.logits)
+    is_fake = probabilities > 0.5
+    
+print(f"Probability of being fake audio: {probabilities[0][0]:.4f}")
+print(f"Prediction: {'FAKE' if is_fake else 'REAL'} audio")
+```
+
+## Limitations
+
+Important considerations when using this model:
+1. The model works best with 16kHz audio input
+2. Performance may vary with different types of audio manipulation not present in training data
+3. Very short audio clips (<1 second) might not provide reliable results
+4. The model should not be used as the sole determiner for real/fake audio detection
+
+## Training Details
+
+The training process involved:
+1. Loading the base AST model pretrained on AudioSet
+2. Replacing the classification head with a binary classifier
+3. Fine-tuning on the fake audio detection dataset for 10000 iterations
+4. Using gradient accumulation (8 steps) with batch size 8
+5. Implementing validation checks every 500 steps