Voice AI is apparently so hot right now that investors are flinging billions of dollars at anything that can listen, talk back, or take meeting notes. Into this gold rush steps Treble, an Iceland-based startup founded in 2020 by acoustic engineers Finnur Pind and Jesper Pedersen, which just raised $18 million in an extension of its Series A. The round was led by Paladin Capital Group, with existing investors KOMPAS VC, Frumtak Ventures, EIC, and Omega ehf. also chipping in. That follows a $12 million raise in 2024, bringing Treble's total to over $40 million. The company counts Amazon and Logitech as customers, presumably because even giant corporations need someone to tell them how their gadgets sound.

Treble's pitch is essentially: all these new voice AI models and devices need testing, and testing needs data. Lots of it. Rather than scraping recordings off the internet like everyone else, Treble generates synthetic audio data for speech enhancement, noise suppression, and model training. It also evaluates voice AI models under different conditions and feeds that back to labs. Earlier this year, it partnered with Hugging Face to launch a benchmark for speech recognition models across realistic conditions - because someone has to grade the machines.

"Audio AI is really a data challenge, and this is where the most opportunities to enable next-generation models and hardware lie," Pind told TechCrunch. "To date, pretty much all sound-related AI has been made from recordings and data scraped from the internet. We believe that accurate physics simulation can be an alternative way to create data for sound." In other words, why rely on the messy real world when you can simulate it?

Beyond software, Treble does hardware design and testing from a voice perspective. It helps headphone and speaker companies virtually prototype how their products might sound, tests whether a smart speaker can understand commands based on where it's placed, and has lately moved into simulation testing for smart glasses and AI devices. Pind is particularly excited about wearables that could give users "superhuman hearing." "Maybe you are in a restaurant, and you only want to hear people within two meters of range, or you are in a seminar, and want to mute people around you," he said. Because nothing says progress like selectively ignoring the people around you.

The company is also setting its sights on physical AI - robotics, automotive, and drone companies - to enable sound-based functions through testing and simulation. Francois Ruether, VP of Paladin Capital Group, said Treble's platform stands out because it simulates different models and devices, and its importance will only grow. "Our thesis is that, as more products depend on understanding sound, this infrastructure becomes increasingly valuable across voice AI, wearables, robotics, and physical AI," Ruether said. "Customers retain ownership of their models, products, and development workflows, while benefiting from a shared foundation of a simulation-native acoustic infrastructure layer." Which is a fancy way of saying: everyone gets to keep their stuff, but they all need Treble's simulated sound to make it work.