wip(copycat): pyin bass transcription — Bb1↔F2 vamp, not V chord arp

TODO verify together: bass rewritten from pyin analysis (librosa),
kick LPF raised for click, d3 cut groups for OH ring-over, cello updated.
Bass was wrong — NTO uses sustained Bb1/F2 alternation with chromatic
slides, NOT F/A/C arpeggio. A2 never appears in any segment.

backlog.md

@@ -83,6 +83,7 @@ Instructions
 
 ## 26 lettres de noblesses
 ## Mars Eclosion
+--La Cle des  champs
 Drums Architecte
 ## Fevrier Vide
 ## Janvier A

copycat/la_cle_des_champs.tidal

@@ -30,11 +30,12 @@
 -- ================================================================
 -- Source: ~/Downloads/N'to - La Clé Des Champs [qd_8mm906GA].mp4
 --
--- BASS stem: 86Hz (F2) dominant across all segments
---   Bb1=58Hz (sub band), F2=87Hz (low band), C2=65Hz (passing tone at 120s)
---   Sub/low ratio shift reveals root movement: high sub% = Bb1, low sub% = F2
---   → Bass sits in Bb1-F2 range, NOT Bb2-F2 (deeper than initially coded)
---   → C2 passing tone at breakdown (seg_0120) = scale degree 2
+-- BASS stem (pyin pitch tracking across all segments):
+--   Bb1=58Hz (~45% of frames) + F2=88Hz (~30%) = sustained 2-bar alternation
+--   Chromatic slide transitions: Bb1→Ab1→F2 (descending), F2→C#2→Bb1 (ascending)
+--   Eb2=78Hz for iv (Ebm) sections, F1=44Hz sub-octave in later segments
+--   A2 NEVER appears — bass is Bb1↔F2 vamp, NOT V chord arpeggio
+--   Bass deepens over time: F2 → F1 in segments 250s+
 --
 -- PIANO stem: enters ~40-50s, peaks ~90-100s
 --   237Hz (Bb3) dominant, 53% low + 47% mid
@@ -121,7 +122,7 @@ let gM2 = gMask . gMute2
 let gM3 = gMask . gMute3
 -- Harmonic foundation: 16-bar cycle (each chord = 2 bars)
 -- Main Bbm/F vamp with Db/Ebm variation
--- Demucs evidence: bass at Bb1(58Hz)/F2(87Hz), C2(65Hz) passing tone at breakdown
+-- pyin evidence: bass = sustained Bb1(58Hz)↔F2(88Hz), Eb2 for iv, chromatic slides
 let rt = slow 2 "<bf2 f2 bf2 f2 bf2 f2 df2 ef2>" -- root sequence (for d6 piano, pad, etc)
 d1 $ gF1 $ gMute2 -- KICK: Sub thud, 4otf with flourish (NTO: deepens over time)
   $ midiOn  "^42" (<| "k k k <k k*2 k [~ k]>") -- ON: 4otf + flourish variations
@@ -130,7 +131,7 @@ d1 $ gF1 $ gMute2 -- KICK: Sub thud, 4otf with flourish (NTO: deepens over time)
     superimpose (("0.07" <~) . (|* gain 0.7))
     )
   $ "[techno:0,808bd:2]"
-  # lpf (slow 32 $ range 700 250 saw) -- 32-bar: darkens, more sub over time
+  # lpf (slow 32 $ range 4000 1500 saw) -- 32-bar: kick click present, darkens over time
   # room (slow 32 $ range 0 0.15 saw) -- 32-bar: tiny room growth
   # gain 1.8
 d2 $ gF1 $ gM -- CLAP: Dry quiet snap, contretemps (enters late)
@@ -157,24 +158,27 @@ d3 $ gF1 $ gM -- OH/CY: Subtle offbeat hats, ^44 morphs to cymbal shimmer
   # "h2ogmhh"
   # n "~ 3 ~ 2 ~ 4 ~ <3!3 [3 6]>" -- varying openness per hit
   # legato (range 0.2 0.9 "^52") -- ^52: tightness
+  # cut "~ 3 ~ 3 ~ 3 ~ <3!3 [3 0]>" -- last hit: cut 0 = rings over bar boundary
   # room (slow 32 $ range 0.2 0.45 saw) -- 32-bar: space grows (NOT dry!)
   # sz (slow 32 $ range 0.1 0.3 saw) -- 32-bar: tail lengthens
   # dry 1
   # pan 0.42
   # gain (slow 32 $ range 1.3 1.6 saw * "0.7 0.8 0.65 [0.75|0.85] 0.69 0.8 ~ <0.9!3 [0.9 0.5]>")
   -- 32-bar swell * per-hit velocity (the original subtle dynamics)
-d4 $ gF2 $ gM3 -- BASS: V chord tones (F2/A2/C2), rhythmically alive
-  -- onset analysis: NTO bass = ALWAYS F2/A2/C2 (V chord), never sits on Bb!
-  -- hits every ~2 beats with variation, answer at beat 2.7→3.7→next bar 0.2
-  $ midiOn "^57" ( -- state: octave ghost echo
-    off "e" ((|+ note 12) . (|* gain 0.65) . (# legato 0.3))
+d4 $ gF2 $ gM3 -- BASS: Sustained Bb1↔F2 alternation with chromatic slides
+  -- pyin analysis: Bb1 (58Hz, ~45%) + F2 (88Hz, ~30%) = 2 main notes
+  -- 2-bar holds, chromatic slide transitions (Ab1→F2, C#2→Bb1)
+  -- Eb2 for iv chord, F1 sub-octave in later segments (bass deepens)
+  -- A2 NEVER appears — prior "V chord arp" was wrong
+  $ midiOn "^57" ( -- state: sub-octave layer (NTO later segments)
+    superimpose ((|+ note (-12)) . (|* gain 0.6) . (# legato 2))
     )
   $ midiOn "^89" (struct "[t, [~ t t ~. <t(3,8) t*[2 <2 4>]>]]") -- push: wild mode
   $ note (
-    slow 2 $ "<[f2@5 a2 c2 f2] [c2@5 a2 c2 f2] [f2@5 c2 a2 f2] [a2@5 c2 c2 f2] [f2@5 a2 c2 f2] [c2@5 a2 a2 f2] [f2@5 af2 f2 df2] [f2@5 gf2 ef2 bf1]>"
+    slow 2 $ "<[bf1@7 af1] [f2@7 af1] [bf1@7 af1] [f2@7 af1] [bf1@7 af1] [f2@7 af1] [df2@7 c2] [ef2@7 df2]>"
     )
   # "bassWarsaw"
-  # dec 0.3 # sus 0.8
+  # dec 0.3 # sus 0.8 # legato 2
   # octersubbus 41 (range 0 2.5 "^53") -- ^53: sub-octave depth
   # crushbus 42 (range 16 3 "^33") -- ^33: grit
   # room (slow 32 $ range 0.05 0.2 saw) -- 32-bar: bass space opens up
@@ -212,7 +216,7 @@ d9 $ gF2 $ gM3 -- PAD: Warm evolving harmonic bed
   # room 0.6 # sz 0.8 # dry 0.9
   # legato 4
   # pan 0.4
-  # cut 9
+  -- # cut 9
   # gain (slow 32 $ range 0.7 1.4 saw) -- 32-bar: pad swells gently
 d6 $ gF2 $ gM3 -- PIANO: Sustained Rhodes chords (demucs: Bb3/A3/F3, enters ~40s)
   -- ^31: modIndex sweep (more harmonics = more presence)
@@ -229,8 +233,8 @@ d7 $ gF3 $ gMute3 -- CELLO: Rhythmic bowed answer, tighter complement to bass
   $ midiOn "^91" (ply 2) -- push: doubles for urgency
   $ midiOn "^59" ((|* gain 1.2) . (# "[cpluck:6,cpluck:7]")) -- state ON: pizzicato
   $ midiOff "^59" (struct "t@3 [t t t] ~ ~") -- OFF: rhythmic bow, answer on beat 3
-  $ note ( -- mirrors bass V chord (A/C/F) one octave up, tighter phrasing
-    slow 2 $ "<[f3@3 [a3 c4 f3] ~ ~] [a3@3 [c4 f3 a3] ~ ~] [f3@3 [a3 c4 f3] ~ ~] [c3@3 [a3 f3 c3] ~ ~] [f3@3 [a3 c4 f3] ~ ~] [a3@3 [c4 a3 f3] ~ ~] [af3@3 [f3 df3 af3] ~ ~] [gf3@3 [ef3 bf2 gf3] ~ ~]>"
+  $ note ( -- mirrors bass Bb↔F vamp, one octave up, with answer phrase
+    slow 2 $ "<[bf2@3 [af2 f2 bf2] ~ ~] [f3@3 [af2 bf2 f3] ~ ~] [bf2@3 [af2 f2 bf2] ~ ~] [f3@3 [af2 bf2 f3] ~ ~] [bf2@3 [af2 f2 bf2] ~ ~] [f3@3 [af2 bf2 f3] ~ ~] [df3@3 [c3 df3 ef3] ~ ~] [ef3@3 [df3 bf2 ef3] ~ ~]>"
     )
   # "cbow:2"
   # crushbus 71 (range 16 2.5 "^55") -- ^55: grit