As in recent few years, voice biometric-based authentication systems are commonly used in various system applications. This is a critical to ensure that these systems are protecting user data against threats like spoofing attack and voice synthesis attack. We found that traditional approach of voiceprint authentication methods work well, they are still vulnerable to these types of attacks and not safe. To counter this vulnerability, in this paper, we proposed a hybrid approach of authentication framework that will combine voiceprint recognition system, using machine learning techniques, with One-Time Passwords (OTPs) to improve overall security of system. The voiceprint authentication component leverages advanced machine learning-based models, including Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, to effectively identify speaker-specific features from voice inputs. The addition factor of OTPs provides an extra layer of security by using a dynamic six-digit numeric random value-based time-sensitive password. This new integrated approach offers strong protection against unauthorized access of user account, including impersonation and deepfake threats. We can assess the system’s effectiveness based on various attributes like accuracy level, security or robustness, and real-time performance of system, demonstrating a significant improvement in system security as compared to traditional approach of voiceprint or OTP-only methods. This new hybrid MFA solution is a scalable and secure option, well suited for a variety of system applications that require both convenience and high level of security.