From 46d44d494b84f407d0066a511abb19f5bdd3eb14 Mon Sep 17 00:00:00 2001 From: Waters Church Date: Thu, 17 Sep 2026 16:00:34 -0400 Subject: [PATCH] Keep the banding caveat where the band plan is built Co-Authored-By: Claude Opus 5 --- modules/spl-meter/spl-capture.swift | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/modules/spl-meter/spl-capture.swift b/modules/spl-meter/spl-capture.swift index 7c20c87..25a76dc 100644 --- a/modules/spl-meter/spl-capture.swift +++ b/modules/spl-meter/spl-capture.swift @@ -253,7 +253,11 @@ let bandN1 = 207 // ≈ 20 kHz let bandCentres: [Double] = (bandN0...bandN1).map { 1000 * pow(2, Double($0) / 48) } /// FFT bin ranges per band, clamped to what the transform can actually -/// resolve. Below roughly 200 Hz a 1/48-octave band is narrower than one +/// resolve. +/// +/// This is the one caveat a reader of the RTA needs and the gear is not +/// allowed to carry (module rule 5, no explanatory text in a tile's +/// settings): below roughly 200 Hz a 1/48-octave band is narrower than one /// bin, so neighbouring bands start sharing bins and the low end reads /// smoother than it really is. That is the window length talking, not the /// room. Folding to 1/6 or 1/3 puts it right, and a multi-resolution