Sign language is a visual language used by millions of people worldwide to communicate. However, creating accurate and expressive sign language gestures for digital applications remains a challenge. One potential solution to this problem is sign language generation using a 3D avatar. Since recently, there has been increasing demand for using 3D avatars to generate sign language gestures, as they offer a more lifelike representation of sign language sentences. In this paper, a new approach is proposed to generate sign language gestures using a 3D avatar. The proposed approach has several potential applications, including improving accessibility for individuals who use sign language as their primary means of communication, Recent advancements in computer vision have led to the development of techniques that use MediaPipe landmarks to generate 3D avatars of sign language gestures. The suggested method integrates motion capture data with MediaPipe landmarks. It uses the MediaPipe model to produce gesture sequences as key points or landmarks from video based on the given text. These generated gestures are then applied to a 3D avatar to create realistic sign language animations. This method was assessed with a dataset of Indian Sign Language (ISL) alphabets and words, demonstrating that it generates precise and expressive sign language gestures.