Iceland-based Treble has raised $18 million in a Series A extension led by Paladin Capital Group, TechCrunch reported on September 16. Existing investors KOMPAS VC, Frumtak Ventures, EIC, and Omega ehf also participated. The company develops acoustic simulation tools and synthetic audio data for voice AI and hardware engineering.

Founded in 2020 by acoustic engineers Finnur Pind and Jesper Pedersen, Treble is targeting a practical problem: voice systems need testing in environments that differ from the recordings used to develop them. Its platform supports speech enhancement, noise suppression, model training, and evaluation under different acoustic conditions.

Simulating the listening environment

Treble's public Far-Field ASR benchmark on Hugging Face provides a concrete example of that work. Its documentation describes a speech-recognition benchmark covering clean, noisy, and reverberant audio. That makes environmental variation an explicit part of evaluation, rather than treating speech recognition as a single aggregate task.

The company also works on virtual prototyping for headphones and speakers. According to TechCrunch, simulations can help engineers assess how devices sound and how a smart speaker understands commands from different positions. Treble has expanded into testing smart glasses and other AI devices.

Beyond voice models

Pind told TechCrunch that Treble plans to increase its focus on physical AI, including robotics, automotive, and drone applications. Those are intended expansion areas, not evidence that every proposed sound-based capability is already deployed.

The investment backs infrastructure for developing and evaluating audio systems, rather than a new consumer assistant. Its relevance is the testing layer between a voice model and the acoustic conditions it encounters in use.