Out of more than 2 million artificial intelligence projects worldwide, an Arlington High School junior’s work now ranks in the top 0.001%. Yatharth Sharma created LuxTTS, an AI model that recently climbed into the Global Top 25 on Hugging Face, the world’s leading open-source platform for AI development and collaboration. LuxTTS has garnered more than 1 million views, and over 4,000 people have already started using LuxTTS to build their own applications.
LuxTTS is a text-to-speech model that converts written words into natural-sounding audio. While text-to-speech technology isn't new, Sharma’s model is highly efficient and doesn't need expensive hardware to run. LuxTTS can generate approximately 150 seconds of high-quality audio in just one second.
"I wanted to help the community by building something that everyone can use without tens of thousands of dollars in GPUs and devices," Sharma explained.
Because LuxTTS creates human-sounding voices quickly and inexpensively, it can be used to:
Run automated call centers that communicate naturally with customers.
Generate real-time video game dialogue that adapts to a player's actions.
Assist visually impaired individuals who rely on high-quality text-to-speech tools.
Yatharth began independently exploring AI programming a few years ago. He learned by experimenting and asking questions within the tech community. While he credits his math classes for his success, he also notes that his English classes played a significant role. "English class helped me with appealing [to audiences] and writing good posts to describe these models in detail," Sharma said, highlighting the importance of being able to communicate complex ideas to the public.
His technical skills earned him the prestigious RIT Computing Award. While he is unsure where he will attend university, the award includes a significant scholarship if he chooses to attend RIT. While LuxTTS focuses on text-to-speech, Yatharth plans to expand his research into large language models (the tech behind chatbots) and image-generating AI.
"My goal is to create a more global impact," Sharma said. "And release more powerful models that fix bigger problems.”
Watch a video of an interview with Yatharth by clicking here.