Where the analyser earned its keep
I did not build this as a product. I built it because I was mid-way through making a record and kept hitting decisions my ears could not settle honestly: which desk channels to trust, which mic to put in front of a vocal, whether two options actually sounded different or I just wanted the expensive one to win. These are the two sessions that turned it from a script into something I now reach for every time. Both are lifted from my own production journal.
1. Twenty-four channels of a 1980s British console, one morning
The studio has a 1980s British 24-channel console, and desks that age drift: caps, faders, whole channels. Before committing a tracking block to it I ran pink noise through every channel at unity with the desk warmed up, captured all 24 returns, and put the trimmed files through the analyser.
First finding before any analysis: channels 21 to 24 returned no signal at all. Tested, not skipped. Usable channel count for the session plan became 20, and I knew it before the drummer was on the clock.
For the analysis I matched channels on level-independent metrics only: band-energy ratios, spectral centroid, crest factor, flatness, harmonic ratio and inharmonicity. Gain was not matched per channel, so absolute level would have lied to me; what I cared about was saturation and EQ character, and those metrics isolate exactly that.
- The 20 live channels clustered tightly. The desk is consistent, which is itself worth knowing.
- Channel 4 was the clear outlier: brightest of the set (centroid around 8170 Hz against a cohort near 7850 Hz), highest flatness, lowest crest factor. In plain words, the most saturated channel on the desk. Character, if you want it. A liability under a lead vocal.
- Cleanest channels, ranked: 20, 17, 9, 18, 10. Those got kick, snare, lead vocal and DI.
- Best-matched pairs for stereo sources: 5 with 16 (the tightest on the desk), 2 with 19, 18 with 20. Overheads and rooms went there.
The same characterisation work later fed the capture sessions behind my plugins. One morning of pink noise paid for itself twice.
2. Six vocal mics, twelve takes, one honest answer
Vocal mic shootout at WaterBear, Studio C. Six mics on the same passage, sung the same way every take: once clean into the interface, once through a console-style preamp. Pads out, cardioid on everything, onboard filters flat, and gain re-matched per mic so no take could win by being louder. Twelve bounces went through the analyser.
The finding that reframed the whole exercise: every one of the six mics put 75 to 83 percent of its energy into the core mids, 500 Hz to 2 kHz. The core sound of a vocal mic is barely a choice at all. The decision lives in the edge bands, and z-scoring the six capsules against each other showed each one owning a different band:
- One capsule owned the body, 250 to 500 Hz: the chest and warmth.
- Two owned the core mid and were near-identical to each other, with a complementarity score of zero. Stacking them adds level, not width. I would have kept layering that pair forever on instinct.
- One owned the attack around 2 to 4 kHz, one owned the 4 to 6 kHz presence, and the tube mic owned the air above 6 kHz while also holding the attack band, making it the most spectrally complete single capsule of the six.
The warm preamp measurably added top-end sheen to the condensers (one centroid moved from 4598 Hz to 4975 Hz) and slightly darkened the dynamics. So on this chain the expensive preamp buys air, not low-end weight, which is the opposite of what the marketing folklore says.
The chain got locked from the data: tube mic through the warm preamp for the lead, the body-owning capsule for the double, the presence-owning one optional for grit. My ears made the shortlist; the numbers made the call and killed a redundant mic pairing I would never have questioned.
Why I trust it enough to charge for it
Everything above fed real decisions on a real record, and the plugin line that came out of those sessions is built on the same measurements: measured rather than modelled. You'll hear Box Of Rules plugins on Box Of Rules songs. This analyser is the instrument behind both, and now it is pointed at your recordings too.