The concept of robust detection technique to detect the serious threats in the speaker verification systems is present at higher level of the synthetic speech creation capability. In the present introducing the actual facts to detect the synthetic speech and countering spoofed speakers to verify the same. In the present paper the concept of two databases has been introduced which is CNN, RCNN. The proposed approach is TCNN which generates better results as compare to the CNN and RCNN. Feature extraction techniques which use MFCC and LPC which amalgamates the spectral features and serving as a necessary tools for the identification of artificial speech from the human speech. The paper demonstrates that effective anti-spoofing technology consists of two approaches which are dynamic acoustic feature analysis and raw wave deep neural networks. It has been observed that Deep learning solutions have shown superior effectiveness in synthetic speech detection capabilities. Research and development in the future should work on enhancing speaker verification system robustness by opposing adversarial attacks while achieving better results across distinct datasets.