Audio Spoofing Detection

This repository contains code for training and evaluating audio spoofing detection models. The codebase includes various models and datasets for detecting spoofed audio.

Training

To train the models, use the provided training scripts. Each script is designed for a specific dataset and model configuration.

Example: Training on IndicFake Dataset

To run the training script directly, follow these steps:

1. Run the Training Script:

   python train_models.py --model_name SSLModel --dataset_name Bahubhashak --language SETA --train_variant SETA --language_test SETB --test_variant SETBC --num_epochs 5
   

Example: Training on DECRO Dataset

To run the training script directly, follow these steps:

2. Run the Training Script:

   python3 train_models.py --model_name RawBMamba --dataset_name DECRODataset --language Chinese --language_test English --num_epochs 9 --lr 0.00001
   

Evaluation

To evaluate the trained models, use the provided evaluation script.

Example: Evaluating on WaveFake Dataset

python evaluate_models.py --model_name RawBMamba --dataset_name DECRODataset --language Chinese --language_test English