IIT student builds world's first speech-to-speech AI Model that can sing

A remarkable breakthrough in artificial intelligence has emerged from Jaipur, where 25-year-old entrepreneur Sparsh Agrawal has launched one of the pioneering speech-to-speech AI models capable of singing and conveying emotions. Developed through his startup PixaAI, the innovative model, named Luna AI, processes audio directly to generate natural speech, eliminating the traditional text conversion process. This unique approach enhances the fluidity and emotional depth of conversations. Agrawal's dedication to creating Luna AI without relying on significant tech support or venture capital funding showcases his innovative spirit. The model supports features such as whispering and tone adjustments, making interactions feel more human-like. In a recent meeting with Union IT Minister Ashwini Vaishnaw, Agrawal received commendations for his groundbreaking work, which he proudly shared on social media, calling Luna the answer to the question of India's presence in AI. Initial performance benchmarks suggest that Luna surpasses established systems like OpenAI’s GPT-4 TTS and ElevenLabs, boasting 50% lower latency and delivering more authentic-sounding speech. Agrawal reflected on his resourceful journey, stating, “I didn’t have a research lab or a USD 100 million runway. I borrowed GPUs and cloud credits, even incurred credit card debt, to bring Luna to life. This proves that world-class technology can originate from India, built on ingenuity rather than just resources.” An alumnus of IIT-BHU, Agrawal collaborates with teammates Nitish Kartik, Apoorv Singh, and Pratyush Kumar. Supported by investors like Kunal Shah and Nikhil Kamath, PixaAI aims to embed Luna into various sectors, including entertainment, wellness, and automotive. Early interest has already been shown by automakers and gaming companies, highlighting Luna’s potential. Unlike typical voice models designed primarily for customer service, Luna is crafted to resonate emotionally. “It’s not just about responding — it’s about feeling,” Agrawal emphasized. Industry leaders, including HCL co-founder Ajai Chowdhry, have lauded this achievement, recognizing Sparsh's contribution as a significant milestone for India’s tech landscape. Sudarshan Kamath, a prominent voice AI researcher, echoed this sentiment, calling for more support for innovators like Agrawal who are reshaping the AI frontier. Selected as the only solo founder from over 15,000 candidates for the WTFund, Agrawal’s vision with Luna is clear: to position India as a leading hub for emotionally intelligent AI advancements.