Treble secures $18 million Series A extension

The Icelandic startup Treble has closed an $18 million Series A extension and claims Amazon and Logitech as clients. The company focuses on acoustic simulation and synthetic data generation to test models, devices, and wearables, with an increased emphasis on physical AI.

Paladin bets on an infrastructure brick that clients retain

For François Ruether, Vice President of Paladin Capital Group, Treble's platform, capable of simulating different models and devices, stands out, and its importance is expected to grow as it ventures into new areas. According to him, the more products rely on sound understanding, the more valuable this infrastructure becomes in voice AI, wearables, robotics, and physical AI. He specifies that clients retain ownership of their models, products, and development streams while relying on a common foundation of acoustic infrastructure native to simulation. Treble already counts Amazon and Logitech among its clients.

Synthetic data and model evaluation at the heart of the offering

Treble offers a synthetic data generation platform aimed at speech enhancement, noise suppression, and model training. The company also evaluates voice AI models under varied conditions to provide feedback to labs and has collaborated with Hugging Face to launch a speech recognition benchmark in realistic environments. Finnur Pind presents audio AI as a data problem and believes that while most systems have relied on recordings and data extracted from the internet, precise physical simulation can serve as an alternative for creating sound datasets.

From virtual prototyping to testing glasses and AI devices

Beyond models, Treble is working on hardware design and testing from a vocal perspective. The company collaborates with headset and speaker manufacturers to conduct virtual prototyping and anticipate the sound rendering of their products, and can test a smart speaker's understanding of commands based on its positioning. More recently, it has expanded its simulation testing to smart glasses and AI devices.

Wearables and targeted hearing, scenarios discussed by Treble

Finnur Pind indicates that the company is excited about working on wearable devices that could enhance hearing. He mentions the next generation of headsets and smart glasses that could offer "superhuman hearing" in challenging acoustic environments. For example, he cites the possibility of only hearing people within a two-meter radius in a restaurant or cutting out surrounding sounds during a seminar.

Focus on physical AI and funding details

Treble plans to intensify its efforts in so-called physical AI, targeting companies in robotics, automotive, and drones to activate sound-based functions through testing and simulation. On the funding side, the startup has raised $18 million in Series A extension, led by Paladin Capital Group with participation from existing investors including KOMPAS VC, Frumtak Ventures, EIC, and Omega ehf. It received $12 million in 2024, bringing the total funds raised to over $40 million. Based in Iceland, Treble was founded in 2020 by acoustic engineers Finnur Pind and Jesper Pedersen.

A booming voice market demanding testing loops

Voice AI is experiencing a massive influx of investments aimed at uses such as support automation, sales calls, meeting note-taking, and voice-controlled smart glasses. At the same time, new AI models are being rapidly released, and device manufacturers are working to optimize the user experience. This dynamic reinforces the need for testing and feedback loops to improve the technologies brought to market.