About Treble
Treble is a voice and sound simulation platform that generates synthetic data for speech enhancement and noise suppression, evaluates AI models across acoustic conditions, and supports audio hardware design. Customers include Amazon and Logitech.
What is Treble?
What is Treble?
Treble develops sound simulation and spatial audio technology. Its platform simulates how sound behaves in rooms and devices, which it uses for several jobs: generating synthetic data to train speech enhancement and noise suppression models, evaluating voice AI models across many acoustic conditions, and designing and testing audio hardware, including smart glasses and AI wearables.
Co-founder Finnur Pind summarized the thesis to TechCrunch: "Audio AI is really a data challenge." Real-world recordings covering every room, microphone, and noise condition are expensive to collect, so simulated audio offers a way to scale training and testing data.
Treble serves voice AI developers as well as robotics, automotive, drone, wearable, and consumer hardware companies. TechCrunch names Amazon and Logitech as customers.
The company was founded in 2020 in Iceland by acoustic engineers Finnur Pind and Jesper Pedersen. In September 2026 it raised an $18 million Series A extension led by Paladin Capital Group, with KOMPAS VC, Frumtak Ventures, EIC, and Omega ehf participating, bringing total funding to more than $40 million.
Pros
- Generates synthetic audio data for training voice models
- Evaluates models across many simulated acoustic conditions
- Supports hardware design for wearables and devices
- Used by companies such as Amazon and Logitech
Cons
- Specialized tool that requires acoustics and ML know-how
- Mostly aimed at enterprises and hardware makers
- Public pricing is not listed
How to Use Treble
- 1Define the scenario
Describe the rooms, devices, and conditions you need to simulate.
- 2Run simulations
Use the platform to simulate sound propagation and acoustics.
- 3Generate data
Export synthetic audio for training or testing voice models.
- 4Evaluate
Test models or hardware designs against simulated conditions.