Treble, founded in 2020 by acoustic engineers Finnur Pind and Jesper Pedersen, builds a physics-based acoustic simulation and synthetic data platform intended for voice AI model developers, robotics firms, and consumer hardware makers. The company has secured an $18 million extension to its Series A round led by Paladin Capital Group, with participation from existing investors KOMPAS VC, Frumtak Ventures, EIC, and Omega ehf. Treble previously raised $12 million in 2024, bringing its total funding to date to over $40 million.
What Treble offers
Treble provides several interrelated services:
- Synthetic data generation for speech applications: producing audio data for tasks like speech enhancement, noise suppression, and model training using physics-based simulation.
- Model evaluation under realistic acoustic conditions: testing voice AI models and providing feedback to labs about performance in varied environments.
- Virtual prototyping and hardware testing: simulating how headphones, speakers, smart speakers, and other devices will sound and how they will interpret voice commands depending on placement and acoustic context.
Earlier this year Treble partnered with Hugging Face to launch a benchmark for speech recognition models that covers a range of realistic conditions.
Customers and focus areas
Treble counts Amazon and Logitech among its customers. The company plans to increase focus on the physical AI space, including robotics, automotive and drone applications, enabling sound-based functions through testing and simulation.
Treble is also working on wearables such as headphones and smart glasses: virtual testing helps manufacturers understand and improve how devices perform in challenging acoustic environments. Co-founder Finnur Pind has argued that accurate physics simulation can serve as an alternative source of training data to recordings and web-scraped audio.
Investors' view
Francois Ruether, Vice President at Paladin Capital Group, told TechCrunch that Treble’s platform stands out because it simulates both models and devices, and that this infrastructure will become more valuable as more products rely on sound. Ruether emphasized that customers retain ownership of their models, products and development workflows while benefiting from a shared, simulation-native acoustic infrastructure layer.
Why it matters
Voice AI is growing rapidly across use cases such as automated customer support, meeting assistants, and voice-driven consumer hardware. Reliable testing and realistic synthetic data are increasingly necessary for both software models and physical devices; Treble’s physics-based simulation approach offers developers and manufacturers a way to evaluate and refine systems without relying solely on recorded audio or scraped datasets.
