UMD Students Launch 'Saved' to Turn Twitch Streams into Viral Social Media Clips 

In the world of streaming, where around seven million creators vie for viewer attention across platforms, the ability to repurpose content efficiently can make the difference between obscurity and viral success. For Twitch streamers, translating hours of live footage into digestible, shareable clips is daunting. A group of University of Maryland students has taken on this challenge with their startup, Saved. It uses machine learning to identify and repurpose key moments from streams, turning them into ready-to-post highlights for TikTok, Instagram and YouTube Shorts.

Saved, one of four teams in the 2024-25 Mokhtarzada Hatchery Program, is spearheaded by students Aayush Gupta (B.S. ’25, computer science), Akash Moorching (B.S. ’26, computer science; mathematics) and Eric Chen (B.S. ’26, computer science; mathematics). Their platform uses machine learning to simplify content creation, targeting a niche within the gaming community.

Inspiration 

The startup concept emerged from its founding members' daily interactions and observations. Moorching initially conceived the idea after noticing his roommate's challenges as a Twitch streamer.

"I came up with this idea last February from my roommate, a Twitch streamer," Moorching explained. "I would watch him stream all the time but maintain a similar average number of viewers. So I was like, how can I make it easier for him to post content and market himself? Because I noticed that was a difficult area for him."

The initial idea gained momentum during a startup showcase at the university, where Moorching met Gupta and Chen. Their diverse backgrounds complemented the project, and an essential addition to their team was Chen, a former professional gamer who brought insights from the competitive gaming world and his expertise in computer vision research.

How it Works

Saved utilizes an emotion recognition model called Speaker Emotion Recognition to identify stream highlights.

"Our platform identifies three areas of someone's voice: valence, dominance and arousal," explained Gupta. "Researchers believe all human emotion can be plotted on those three axes. We can pinpoint the most exciting moments in a stream using those emotions."

The process incorporates additional Twitch-specific metrics, such as rapid chat movement, to confirm viewer engagement. This system helps find about 20 highlights per stream. The team is working on enhancing their technology by incorporating a new AI model that leverages computer vision and audio analysis to provide more targeted results. The goal is to prompt the model with specific queries to find exciting, humorous or high-energy content, offering users complete creative control over the output.

Overcoming Hiccups

Building their startup tested their resilience as they navigated early challenges and sought opportunities with collegiate esports programs across the U.S., receiving promising initial feedback.

"We emailed all the collegiate esports teams in the U.S., and we got a lot of responses. We had demos, we had meetings and eventually, a few of them agreed to buy our product, but before they signed the contract, they just ghosted us," Gupta recalled.

This challenge led to a reevaluation and subsequent improvements to the product.

"We put our heads down and built a better product," Moorching said.

The revamped approach secured partnerships with organizations such as the NBA 2K League and UMD’s own esports team. The team is now looking to expand its reach, aiming to collaborate with major entities such as Monumental Esports, which owns the Washington Wizards and Capitals.

University Support 

The founders credit the Mokhtarzada Hatchery Program for providing valuable guidance throughout their entrepreneurial journey.

"The biggest part of the Hatchery is the advice from the advisors, Zeki, Haroon and Adam," Gupta shared. "There are a lot of different directions to take when you're a startup, but it's hard to distinguish between good and bad ideas. As someone without much experience, it's easy to fall into common pitfalls."

Biweekly meetings with mentors help the team stay on track, providing candid feedback on their progress.

"We had a meeting just this Monday, and they told us we must focus on developing our product and not just on sales. That focused us back on the core of our product," Gupta added.

Future Goals

Looking ahead, the team behind Saved is setting ambitious objectives.

"Our goal is to be able to do this full time and not have to take on a job or internship after college," Gupta said. "This school year, we hope to reach $50,000 monthly in revenue. But really, we just hope to help other programs grow their following and connect people interested in esports with those making interesting content."

Challenges and milestones have marked the journey for Saved, but the founders remain committed to their vision.

"It's been an incredible experience," Chen concluded. "Every day is exciting, even with the sleepless nights and hard work. It's worth it to be building something unique."

—Story by Samuel Malede Zewdu, CS Communications