Iceland-based Treble has raised an additional $18 million Series A investment led by Paladin Capital Group to develop its sound simulation platform. Existing investors KOMPAS VC, Frumtak Ventures, EIC and Omega ehf. also participated in the round. The company, which raised $12 million in 2024, has now secured more than $40 million in total funding to date. Treble’s customers include Amazon and Logitech.
Founded in 2020 by acoustic engineers Finnur Pind and Jesper Pedersen, the company develops simulation and data solutions for sound AI, robotics and consumer electronics companies. The platform can generate synthetic data for speech enhancement, noise reduction and model training; it also evaluates sound AI models under different conditions and provides feedback to developers. This year, Treble launched a benchmarking system with Hugging Face that compares speech recognition models under realistic conditions.
The company says that, in addition to audio recordings and data collected from the internet, physics simulations could provide an alternative for generating audio data. Treble also carries out virtual prototyping for headphone and speaker manufacturers, testing how products will sound and to what extent smart speakers can understand commands depending on their positioning.
Treble has recently turned to simulation testing for smart glasses and other AI devices. The company is interested in working on wearable devices that could improve hearing ability. Finnur Pind said that, in the future, devices enabling users to pick out specific sounds in noisy environments such as restaurants or seminars could become increasingly important.
Treble aims to focus on robotics, automotive and drone companies in the field of physical AI, equipping these systems with audio-based functions through testing and simulation. Paladin Capital Group Vice President Francois Ruether said the platform, which simulates different models and devices, would become increasingly valuable in the fields of audio AI, wearable devices, robotics and physical AI.
Why it matters
Treble’s approach allows teams developing audio AI to test different acoustic conditions without relying solely on real-world recordings and data compiled from the internet. This could broaden the scope of testing processes, particularly for companies developing speech recognition, noise reduction and voice-command-enabled devices. Virtual prototyping enables headphone, speaker and smart glasses manufacturers to assess detection performance before a physical product is produced. The focus on robotics, automotive and drones also indicates that audio could be used in physical AI systems to perceive their surroundings and respond to them. However, it remains unclear to what extent data generated through simulation will represent the diversity of sounds found in real-world conditions.