Turning Audio Concepts into Hardware Reality

From Idea to Prototype: Pushing Audio Innovation Further with FPGA Technology

We transform advanced audio algorithms into real-time embedded hardware. With rapid FPGA-based prototyping and integrated design tools, we bridge the gap between research and production, delivering next-generation sound experiences faster than ever. This makes ultra-low-latency, high-channel-count audio systems accessible for engineers, researchers, and creators alike.

See Our Technology Contact Us

The Bottlenecks in Real-Time Multichannel Audio

Real-time, multichannel audio systems are pushing the limits of traditional CPUs and DSPs. High channel counts and ultra-low-latency requirements make immersive sound experiences difficult to achieve, especially in live or interactive settings. Even the best low-latency audio setups struggle to achieve sub-millisecond performance, and surpassing a few hundred channels is challenging and costly.

FPGA Innovation for High-Channel, Low-Latency Audio

FPGA technology overcomes these limitations through fully parallel, deterministic, and ultra-low-latency computation. Operating without an operating system and with zero buffering, FPGAs eliminate the main sources of delay found in traditional architectures. This enables unprecedented performance (down to a single sample of latency) while supporting high sampling rates up to 768 kHz for demanding professional applications. Their inherent parallelism makes them ideal for spatial audio systems such as Wave Field Synthesis, where independent computations must occur across hundreds of channels simultaneously. With extensive GPIOs and flexible interfacing, FPGAs provide scalable, efficient, and high-channel-count processing solutions.

Despite their potential, FPGAs have long remained out of reach for most audio developers due to their programming complexity, a gap that Polyphonic is closing. Leveraging ultra-low latency audio converters and FPGA processing, Polyphonic’s platform achieves an impressive 11μs latency from analog input to analog output and handled hudreds of intput and outputs.

Accelerating Innovation Through Exclusive FPGA Technology

Our proprietary compiler and custom prototyping platform allow us to turn complex audio algorithms into real-time embedded systems in record time. The result: rapid prototyping, ultra-low latency, and efficient deployment for research and professional audio projects alike.

A Compiler Built for Sound

At the heart of Polyphonic lies a custom compiler that turns high-level C++ or FAUST audio code into efficient FPGA implementations. This tool abstracts the complexity of hardware description languages, enabling designers to design, test, and deploy real-time audio systems directly from familiar code. By automating the synthesis, interfacing, and optimization processes, Polyphonic allows us to rapidly develop proofs-of-concept for FPGA.

Syfala Repository

A Board Designed for Speed and Flexibility

Our custom FPGA development board is designed specifically for audio applications, combining high-speed I/O, precision converters, and flexible connectivity. It integrates seamlessly with our software environment, allowing instant deployment and testing of real-time signal processing algorithms. With expandable interfaces, synchronized clocks, and direct access to multiple audio converters, the platform supports both quick prototyping and complex multichannel systems.

Real Projects, Real-Time Performance

From immersive spatial audio installations to multi-channel live systems, Polyphonic brings complex audio concepts to life. Our FPGA-based prototyping tools enable rapid development and testing, turning ideas into working solutions faster than ever.

Explore projects that demonstrate both the power of our platform and our expertise in low-latency, high-channel-count audio systems:

Click and drag for more

Ultra Low Latency (ULL) Audio Board

Engineered to exploit the full potential of Xilinx Ultrascale+ FPGAs, the ULL Board supports sampling rates up to 768 kHz and end-to-end latency as low as 11 µs. Featuring 32 balanced analog inputs and outputs, the board can be daisy-chained for scalable, massive multichannel systems.

PAW 2023 ULL Board Demonstration

Demonstrated the Ultra Low Latency (ULL) Board at the PAW 2023 conference in Lyon using an array of amplified speakers. The setup showcased the board’s ability to deliver ultra-low-latency, high-fidelity audio processing across multiple channels in a live environment.

Real-Time Active Noise Control on FPGA

Developed an FPGA implementation of a 14×14 FxLMS active noise control algorithm for École Centrale de Lyon’s 22-speaker listening room. Through advanced HLS and architectural optimizations, the system achieved 24 kHz real-time performance on a Genesys ZU3-EG board, outperforming prior CPU-based implementations running at only 2.5 kHz on a SpeedGoat.

FPGA-Based Minimoog Synthesizer Demo

Implemented the Faust library’s Minimoog example on an FPGA to demonstrate the artistic potential of reconfigurable hardware for musical applications. The system runs a custom embedded Linux environment supporting multiple I/O protocols, with a USB MIDI controller driving the FPGA-based DSP engine in real time.

Adjustable High-Density Beamforming Array

Built from The LINE 8-speaker modules, this scalable array demonstrates FPGA-based beamforming and multichannel audio processing capabilities. The system’s size can be adjusted by combining multiple modules, showcasing flexible hardware architecture and precise low-latency control across a large number of channels.

The LINE: Modular High-Density Speaker Array

Developed a low-cost, modular 8-speaker PCB designed to demonstrate high-density audio array concepts while remaining portable and scalable. Each module can be daisy-chained, enabling up to 32 interconnected boards (256 speakers in total) driven by a single FPGA for synchronized, ultra-low-latency operation.

Frugal Wave Field Synthesis System

Designed and implemented a cost-effective FPGA-based Wave Field Synthesis demonstrator, later showcased at INRIA Grenoble. A custom PCB connects 32 speakers directly to the FPGA, enabling ultra-low-latency audio processing from a single chip, an affordable and easily reproducible alternative to traditional multi-device WFS systems.