This study introduces the design and testing of an AI-enabled virtual office hours platform in which voice cloning technology is applied to resemble the presence of instructors in distance education settings in real-time. The system was deployed in an undergraduate Innovation, Entrepreneurship, and Sustainability course where the spoken answers to the user-posed questions were given by the text-to-speech synthesis of the voice of the cloned instructor using the large language models (LLMs), voice synthesis (ElevenLabs), and Retrieval-Augmented Generation (RAG). In this study, usability, engagement, comprehension, and emotional connection have been measured using a mixed-methods experimental design consisting of 29 students: 15 are included in the pilot group. Findings were that there was high weekly use (83%), positive change in understanding (71%), and emotional support (79%). Students considered the voice very realistic and preferred it to text-based solutions. The multilingual feature of the system and its placement in Moodle also contributed to accessibility. Such results indicate that voice assistants powered by AI can introduce tremendous levels of humanity to online learning in a scalable format.