The objective of this paper is to propose a method for safeguarding user privacy in a designated environment by utilizing speech patterns to differentiate genuine users from fraudulent or cloned speech users, or to grant access to specific features. Our approach entails training a model that accurately distinguishes between male, female, male-tempered voice, and female-tempered voice recordings. To achieve this, we systematically classify a dataset of seven-hundred and fifty audio recordings where multiple speakers, including the authors of this paper, articulate the same word for an identical duration. The key factor that distinguishes each speaker is their unique speech delivery pattern, frequency, and pitch. The distinct characteristics of AI-generated voices are critical in identifying their origin. The findings of this study will contribute significantly to creating a secure environment for users to interact in.