Iceland-based Treble raises $18 million for its voice simulation platform
TechCrunch Ivan Mehta ● Covered by 3 sources
Treble just raised $18 million to test voice AI with simulations, not real-world guesswork. It’s betting sound models, smart glasses, and robots all need the same acoustic sandbox.
Based on reporting by TechCrunch, Ivan Mehta — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Voice AI is hot enough that investors are throwing serious money at it, but the less glamorous problem is obvious: these systems still need a way to be tested before they’re out in the wild. That is the lane Treble wants to own. The Iceland-based startup is building a simulation platform for model makers, robotics firms, and consumer hardware companies that need to know how audio products behave before they ship.
Treble said it raised $18 million in an extension of its Series A. Paladin Capital Group led the round, with existing backers KOMPAS VC, Frumtak Ventures, EIC, and Omega ehf also taking part. The company, founded in 2020 by acoustic engineers Finnur Pind and Jesper Pedersen, had already pulled in $12 million in 2024 and says its total funding is now above $40 million. Amazon and Logitech are among its customers.
The startup’s pitch starts with data. For voice AI companies, it offers synthetic data generation for things like speech enhancement, noise suppression, and model training. It also evaluates voice models under different conditions, which is the sort of unglamorous work that tends to matter more than a flashy demo. Earlier this year, Treble teamed up with Hugging Face on a benchmark for speech recognition models across realistic conditions.
But Treble is not only chasing software teams. It also works with headphone and speaker makers on virtual prototyping, so they can hear how a product might sound before building hardware. The company says it can test how a smart speaker responds to commands depending on where it sits, and it has recently moved into simulation testing for smart glasses and other AI devices. Pind said he is especially excited about wearables that could support something like superhuman hearing in tricky environments.
Treble is also pushing deeper into what it calls physical AI, including robotics, automotive, and drone companies. Paladin’s Francois Ruether said the platform gets more valuable as more products depend on understanding sound, while customers keep ownership of their own models and workflows. That’s the real pitch here: not a single voice app, but a shared layer for the messy acoustics under all of them.
My take — AI-written commentary, not fact-checked reporting
This is the part of AI that usually gets skipped in the hype cycle: not the demo, but the test bench. The companies that can make sound systems measurable, repeatable, and slightly less magical are often the ones that end up mattering most. Everyone loves the model; fewer people want to pay for the boring physics underneath it.
Read more about this at: TechCrunch