FPGAudio Soundboard

Inspiration

We were inspired by DJ boards and the desire to manipulate live audio streams with zero perceivable latency.

What it does

FPGAudio Soundboard is a real-time digital audio processing system built on the DE1-SoC FPGA. It takes live audio input, processes it through five parallel hardware-accelerated DSP cores, and outputs the modified audio. It also features a real-time VU meter mapped to the onboard LEDs for visual volume feedback. It also features a real-time VU meter mapped to the onboard LEDs for visual volume feedback.

Using the board's switches, users can toggle between five distinct effects: Noise Gate: Silences background noise below a specific digital threshold. High Pitch & Low Pitch: Real-time pitch shifting using buffer manipulation. Reverb: A delay-line addition that creates a rich acoustic echo. Muffled: A low-pass filter that mathematically smooths out high-frequency waveforms.

How we built it

We wrote custom I2C and I2S serial communication drivers to interface the FPGA with the onboard WM8731 audio codec. Then we began writing the effect modules by using the app Digital to sketch out our circuits and exporting the files as Verilog, then implementing it into the workflow. Then, we built a robust SystemVerilog testbench featuring automated CI/CD verification via GitHub. The testbench injects simulated 1000 Hz$ sine and 500 Hz sawtooth waves into our modules and calculates real-time time-domain statistics (Min, Max, RMS, and Gain in dB).

Challenges we ran into

Transitioning from a sequential programming mindset (like C, Python, etc.) to parallel, clock-cycle-driven hardware logic was a massive learning curve.

Accomplishments that we're proud of

Stepping into this as a first-year students, just getting the DE1-SoC to successfully talk to the audio chip was a huge win. Beyond that, we are incredibly proud of our automated testing pipeline. Instead of guessing if our effects worked, our testbench calculates RMS energy and gain on the fly to catch hardware overflows. Finally, we are proud that we were able to pitch highly complex hardware architecture clearly and effectively in five minutes.

What we learned

We gained deep, hands-on experience with Verilog, the Quartus Prime toolchain, and low-level communication protocols like I2S. We learned that in hardware design, time-domain statistics are often more effective for catching DSP math errors than frequency-domain analysis.

What's next for FPGAudio Soundboard

The next step is to design a custom PCB using Altium Designer to house the audio codec, a smaller dedicated FPGA, and the necessary physical switches. Our goal is to miniaturize the architecture into a standalone, portable audio pedal.