The Appleseed Cast · built 31 July 2026, corrected 6 August 2026

Mare Vitalis: drum tabs for Songsterr, whole album

All ten tracks built from DrumSep 5-piece splits through /stems-to-guitar-pro-drums. Sister page to Illumination Ritual, which is closed at 10 of 10 and 19,609 notes.

10 / 10

tracks with a drum tab

14,277

notes written

1,215

bars

10 / 10

passed all 19 gates

0. What changed on 6 August 2026

Every number on this page was measured on 31 July, before the flam fix of 4 August and before any phantom hit was removed. Nothing regenerated the page, so it kept publishing counts for a file that no longer existed in that form. Three generations of the same twenty songs were live at once: the 31 July build behind these tables, the 4 August flam fix in the output folders, and a phantom strip that had reached only four Mare Vitalis songs. The remaining sixteen were run through the phantom pass today, all twenty are now filed in one folder, and every count below is read straight out of the shipped .gp5.

182 notated hits came out of this album, and the flam count fell hardest, because the first pass wrote grace notes onto rolls a drummer cannot play.

Song	Notes before	Notes now	Cut	Flagged	Lane coverage
01 The Immortal Soul of Mundo Cani	463	463	0	8	toms lane 28%
02 Fishing the Sky	1,707	1,700	7		stripped 5 Aug
03 Forever Longing the Golden Sunsets	2,001	1,945	56		stripped 5 Aug
04 Mare Mortis	1,800	1,759	41		stripped 5 Aug
05 Santa Maria	1,526	1,489	37		stripped 5 Aug
06 Secret	1,424	1,423	1		stripped 5 Aug
07 ...And Nothing Less	1,290	1,290	0	145	kick lane 66%
08 Poseidon	1,102	1,066	36	36	both lanes checked
09 Kilgore Trout	1,206	1,202	4	4	both lanes checked
10 Storms	1,940	1,940	0	200	kick lane 69%

353 flagged hits were left in place on purpose. The stripper refuses to delete from a lane whose best alignment offset does not lock, because deleting on a bad alignment removes real hits. The lane coverage column gives the share of notated hits that line up at the best offset for the lane that failed; below 70% the flagged count records a suspicion the tool could not settle, and those bars want an ear before submission.

The pass covered two of the five lanes. sd_phantom.py defaults to kick and toms, and the run took that default, so the snare, hi-hat and cymbal lanes were never tested against the audio. That is 8,358 of this album's 14,277 notes, 59%, still unexamined. Every count in the table above is real; the phantom column is a floor rather than a total, and the three untested lanes may hold more.

1. The album

#	Song	BPM	Bars	Notes	Bells	Open hat	Ghosts	Flams	Tempo
01	The Immortal Soul of Mundo Cani	172	78	463	0	6	7	6	MEDIUM
02	Fishing the Sky	187	174	1,700	12	24	115	4	LOW
03	Forever Longing the Golden Sunsets	152.75	174	1,945	0	0	143	38	MEDIUM
04	Mare Mortis	136	114	1,759	0	20	191	19	LOW
05	Santa Maria	138.25	116	1,489	8	20	142	7	LOW
06	Secret	87	96	1,423	0	17	63	9	MEDIUM
07	...And Nothing Less	88.25	100	1,290	0	10	100	3	MEDIUM
08	Poseidon	137.25	118	1,066	0	12	65	10	HIGH
09	Kilgore Trout	181.75	106	1,202	0	26	43	0	MEDIUM
10	Storms	95	139	1,940	0	24	142	1	MEDIUM

Album totals: 20 bells, 159 open hats, 0 rim clicks, 1,011 ghosts, 97 flams, 103 crashes, 1,667 rides.

The 1,667 rides are a fallback label rather than a measurement. classify_cymbals in sd_lanes.py opens with out = ["ride"] * len(F) and then overwrites only where a positive test fires, for a crash or for a bell. No branch ever writes ride, so every ride counted here is the initial value that survived. Of this album's 1,790 cymbal-lane notes, 123 carry a positive verdict and 1,667, which is 93.1%, carry the default. On Mare Mortis, the one song here with a human transcription to check against, the human writes 57 bell and 7 bow while this build writes 0 bell and 171 ride. Measured 2026-08-29, evidence in ~/Projects/_outputs/appleseed-cymbal-default-ride-2026-08-29-macbook/.

2. It did not need the external drive

This album was queued as blocked on /Volumes/Black, which is not mounted. All ten tracks were already demucs-separated locally at ~/Projects/_outputs/mare-vitalis-demucs/, every one with a drums.mp3. Only the DrumSep stage was missing, and that runs on local files. Separation ran 02:01 to 05:40 with zero skips, 218 to 474 MB per song, and every hat and cymbal lane cleared the 8-16 kHz air-band reality gate.

3. Two detector bugs this album exposed

The cut did not know which side the ornament was on. The splitter maximized separation without caring which class ended up small, so on six of ten songs it isolated the 5% dimmest cymbals and labelled the remaining 95% "bright". The ornament ceiling rejected that as nonsense, which was the right outcome by the wrong mechanism, and it discarded real bells along with it. Making the cut side-aware recovered 12 bells on Fishing the Sky, 8 on Santa Maria, and open hats on Mare Mortis and And Nothing Less, both of which had been sitting at zero. Those two lanes stand at 20 and 10 in the shipped file, below the 23 and 17 first reported, because the 4 August flam fix removed grace notes that were being counted in the lane.
A capped search made the cap look like a measurement. With the size ceiling in place, any lane preferring a larger class settles exactly on it. Poseidon reported a rim class at precisely 0.450, the ceiling to three decimals, and it passed every other control (body-vs-loudness correlation 0.18, ghost overlap 40%). The splitter now reports saturation and refuses a split whose optimum sat on the boundary. Poseidon went from 108 rim clicks to 0, and its ghosts recovered from 36 to 65, because a runaway rim class eats the snares that ghost detection ranks within.
The control that makes those fixes trustworthy. Illumination Ritual came through all three changes byte-identical, and it still carries its 25 bells, 26 open hats and 75 ghosts today. Its crashes read 25 and its rides 353 in the shipped file, against the 26 and 386 reported on 31 July, and its flams fell from 88 to 30 in the flam fix. Gates of this shape are easy to write in a form that silently deletes everything, and only a known-good song catches that.

4. The grids here are genuinely weak, and are labelled so

Backbeat concentration tops out near 0.29 on this record, against 0.54 on Illumination Ritual and 0.78 on Clearing Life. Confidence came back 1 HIGH, 6 MEDIUM, 3 LOW. Mare Vitalis is a more atmospheric album with less backbeat for the discriminator to lock onto.

The octaves are right. Every one of the five suspect tempos beat both its double and its half on the backbeat test, including the three low ones: Secret at 87, And Nothing Less at 88.25, Storms at 95. These are correct tempos carrying honest low confidence, so check the bar numbers against the audio before submitting anything.

5. Rim clicks: zero across both albums

Twenty songs now, and every one has refused the side-stick class, each for a stated reason: the ghost confound on some, the ornament ceiling on others, constraint saturation on Poseidon. Either these two records have no rim-click parts, or this detector cannot find them on this material. It has not been shown that a rim part exists and was missed.

6. What the gate proves, and what it does not

10 songs, 10 gate-green, 14,277 notes, 1,215 bars

  PASS  every velocity on the 16-step ladder
  PASS  one fixed string slot per drum family
  PASS  kick and side stick never share a slot
  PASS  ride is GM 51, never GM 59
  PASS  zero ghostNote parentheses
  PASS  every requested ghost carries the staccato dot
  PASS  every requested flam survived as a legal grace
  PASS  per-drum counts match the event ledger
  PASS  zero hat_pair / snare_rim / crash_ride / three_hand

That proves each file is well formed and matches its event ledger. It says nothing about whether the detection heard the right thing. The 108-rim Poseidon passed all 19. Whether Guitar Pro and Songsterr draw these is the check that happens in the editor.