AUDIO / DIGITAL SIGNAL PROCESSING
Hear the math.
Shape a signal. Inspect the spectrum. Explore fixed-point DSP.
Frequency response
Hamming-window FIR— Floating point— Q1.15 coefficients20 Hz → Nyquist · dB
Signal spectrum
First 2,048 samples · Hann window— Input— Filtered0 Hz → Nyquist · dBFS
Time domain
First 10 ms · normalized amplitudeSample rate
FIR group delay
Fixed-point error RMS
Saturated samples
UNDER THE HOOD
From reference model to RTL.
The browser runs a causal FIR convolution and a signed 16-bit fixed-point model. Compare the numerical error, download the source, and reproduce the reference checks.
This filters frequency bands; it does not isolate vocals or instruments. Delay is calculated from the filter length, not measured end-to-end latency.
↓ Download engineering sourceBrowser & PythonImplemented · reference checks included
SystemVerilogFIR core + comparison testbench
RTL simulationPending · simulator not available here
FPGA resources / timingPending · select a target board
Next hardware stepAudio codec + FPGA interface