I built a tool to measure perceived loudness of presets — curious if this is useful to others
I had a problem I suspect many of you know: record a few presets, compare them in a mix, and even with matched LUFS they sit at completely different perceived volumes. A dense crunch pushes harder than a clean tone at the same K-weighted number — the ear isn't a meter.
I spent a few months figuring out why. The short answer: LUFS K-weighting treats all frequencies equally, but the ear doesn't. A dense midrange-heavy tone triggers more loudness perception than the number suggests.
So I built something. It measures perceived loudness using a psychoacoustic Bark-band model — same principle as Zwicker's loudness model — and gives you:
- The exact dB correction to apply to each preset's output to match perceived loudness to a reference
- A timbre-only "Character" score (presence, dynamics, stereo width) separated from volume — tells you if a difference is a gain issue or an EQ issue
- Cut-through score — how well the spectrum sits in a dense mix
- 8-band spectral balance with concrete parametric EQ suggestions
- Loudness curve over time (N50 sustain / N5 peaks)
Workflow: record the same passage through each preset, drop the files in, pick a reference, press Analyze. Runs in the browser, no install.
I'm selling it but happy to answer questions about the model or the approach first. Has anyone else run into this problem and how did you solve it?