| from model import DID_Model | |
| import torch | |
| device = torch.device("cuda" if torch.cuda.is_available() else "cpu") | |
| aud_path = r"uploads\L.wav" | |
| wave2vec_model_path = r"model_snapshots\wav2vec2_model.pth" | |
| model_path = r"model_snapshots\Marathi_Model_Snapshot.pth" | |
| if __name__ == "__main__": | |
| # Load the Wav2Vec 2.0 model from torchaudio pipelines | |
| # Load custom dialect identification model | |
| model = DID_Model() | |
| model.load_weights(model_path, wave2vec_model_path ) | |
| # Predict dialect | |
| predicted_dialect = model.predict_dialect(aud_path) # | |
| print("Predicted Dialect:", predicted_dialect) | |