End-to-end modeling for speech spoofing and deepfake detection