Enhancing Virtual Yoga Training: Real-time Pose Assessment with Voice-guided Accuracy Feedback
Ashutosh Tripathi, Apar Tyagi, Aditya Dubey, Surbhi Vijh, Pradeep Verma · 2025
By offering real-time voice feedback on asana accuracy, this article presents a cutting-edge AI-based virtual yoga trainer that enhances the user experience and revolutionises autonomous practice without the need for a physical trainer. Our solution surpasses human evaluation capabilities by utilising cutting-edge computer vision and machine learning to provide prompt and helpful advice for improved posture accuracy. This study not only closes a significant gap in the field of virtual trainers, but it also highlights how AI has the ability to outperform human teachers in assessing and directing yoga poses. An important development in virtual yoga training is the addition of real-time audio feedback, which makes it more accessible and customisable to each student’s demands. Unlike existing virtual yoga trainers that rely solely on visual correction, our system introduces a real-time, voice-assisted correction mechanism based on joint angle deviations. This unique integration of audio feedback creates a hands-free, interactive experience that significantly enhances autonomous yoga practice.