Edge Sound Research Raises $2M to Help Physical AI Make Sense of Sound

Edge Sound Research (Techstars 2021) has raised $2M in new funding, bringing its total funding to $9M.

Edge builds audio infrastructure for physical AI. Its technology picks out the sounds that matter in a noisy environment and pairs them with tracking, camera, and other sensor data, so machines can understand what's happening around them in real time. The new funding will help the company finish integrations with its enterprise customers and bring its platform to new industries.

Momentum Behind the Round

Live sports gave Edge its first commercial market and a demanding place to prove its technology. The company's Virtual Sound Engine, which automatically captures and isolates sound in complex environments, is already deployed across the sports industry.

Edge counts Shure, the National Basketball Association, Fox Sports, and the United States Tennis Association among its customers and partners. That work has also helped Edge train and refine the models behind its platform.

Why Sound Is Hard for Machines

Picture a basketball game. The ball hits the court, fans stomp in the stands, and bass pounds through the arena speakers, all at the same moment. Every microphone in the building picks up that mix, which makes it hard for an AI system to pull out a single dribble.

The root of the problem is how audio gets captured. Microphones record by location, so each one hears everything around it without knowing which sound came from which object. Something else has to connect a sound to its source before a machine can use it.

People make that connection without thinking, because we take in sight, sound, and touch all at once. Physical AI has mostly leaned on cameras, since they're already everywhere. Edge is betting that machines need to hear as well as they see.

How Edge's Platform Works

Edge's approach has two layers. The first is a digital twin of every sensor in a space. Customers import a 3D model or a 2D layout with coordinates and place their microphones and other sensors inside it, creating a spatial map of the environment.

The second layer is a set of machine learning models that separate individual sounds from the mix. In an arena, the system uses microphone locations, pickup patterns, and player tracking to choose the most useful audio. It then pairs that audio with visual data to tie each sound to its source, whether that's a player, the ball, a coach, or a referee's whistle.

Timing matters, because what a machine hears has to line up with what it sees. Sports broadcasting needs results in under 33 milliseconds, while other settings can allow 100 to 200 milliseconds. Edge starts with a general model and fine-tunes it for each customer.

Edge is also building its software and models to be hardware-independent. That lets the platform work with the microphones and cameras customers already have, across robots, media, and other systems.

Beyond the Arena

The same technology that follows a player through a crowded arena could help robots and other machines make sense of the world around them, including things their cameras can't see yet.

Take autonomous vehicles. A driver often hears an ambulance before seeing it, and a rising pitch signals that it's getting closer. With the right audio perception, a self-driving car could pick up a siren coming from a blind corner and know which direction it's approaching from, well before its lights show up on camera.

What's Next

With this round closed, Edge will complete integrations with its existing enterprise customers. The team will also make the platform's components transferable across industries as it works toward a Series A.

The People Building It

Edge Sound Research was co-founded in 2019 by Valtteri Salomaki and Ethan Castro, who met at the University of California, Riverside. The company went through the Minnesota Twins Accelerator by Techstars in 2021, where its audio technology got its first live run at Target Field.

Since then, the team has taken what it learned in stadiums and arenas and turned it into a foundation for machine perception. Congratulations to Valtteri, Ethan, and the whole Edge team on this milestone.

SiliconANGLE has the full story on the round and Edge's technology.

Visit edgesoundresearch.com to learn more about the company.