Kid mode 🧒

This is the made-up-live part of a song called Back on the Train. Only two of the four players got pulled out of the recording cleanly, so this page is about the bass and the drums. The bass player starts on a note called C and holds it hard. In the middle he slides over to a note called F, and the drummer gets a little louder in every single section, four times in a row, right to the end. Then the bass goes back to C and the singing comes back in. The grown-up breakdown is below, tap Grown-up mode.

ixi-style analysis · believe-you-me

Back on the Train 2025-04-18 jam peak · ixi analysis

An ixi-style read of the improv window, 7:30 to 12:00 within the extracted jam file, which is shorter than the released track, so these timings are file-relative and not LivePhish track positions. This jam's stems never split the guitar or the keys, so Trey and Page are unread here. Everything below comes from Mike's bass and Fishman's drums, plus a vocal lane used only to gate the window. That vocal lane was validated on 2026-07-19 and it does not support a vocal return in this window, so none is claimed. The finding is bounded. Across the window the bass spends 12.8% of its clarity-weighted weight on A, against 6.5% on B-flat and 5.7% on E-flat. C minor pentatonic requires those last two and excludes A, so the bass does not support a strict C-minor-pentatonic reading. That defeats the label for this lane only: the full-mix detector's reading could be carried by Trey or Page, who cannot be checked. What the bass does support is an F and C tonal field with F-major-triad material. E sits at 1.9%, which is what argues against a C major triad, since that chord needs its third. B-flat at 6.5% is equally too thin to confirm full F Ionian, so F major triad over a C-weighted bass is the honest description and 'F major heard from its fifth' stays a hypothesis. The staves are pitch-tracked and then snapped to F major, which makes them a quantized reading and not a transcription in their own right.

Back on the Train (jam peak, 7:30 to 12:00) Phish · 2025-04-18 Climate Pledge Arena, Seattle

C anchored · the floor (7:30 to 8:37) · the bass plants on C and does not negotiate
Mike plants the window on C and holds it without argument. His clarity-weighted pitch histogram gives C a weight of 437 against G at 180 and F at 173, the most one-sided root of the four sections. A Krumhansl read of this section alone comes back C major at 0.768, the only section here that does. Fishman enters at his quietest, drum energy 1382, the floor of the window. The vocal lane measures 52.8, low enough to indicate probable instrumental playing. Trey and Page are unread. Their lanes never separated, so the harmony above the bass is outside what this sheet can reach, and that limit governs every reading below.
C and F contested · the loosening (8:37 to 9:45) · the weighting comes into question
Mike's grip loosens. C falls to 251 and F climbs to 228, close to level, the first point where the weighting is genuinely in question. The section correlation flips to F major at 0.853 and never flips back. The vocal lane bottoms out at 28.4, the most likely purely instrumental stretch of the four. Fishman nudges up to 1438, continuing a climb that runs the whole window.
F takes the most weight · the arrival (9:45 to 10:52) · F outweighs C for the only time
The move the window was building toward. F outweighs C in Mike's histogram for the only time here, 228 against 188, with A close behind at 166. The section correlates F major at 0.880, the highest figure on the page. Fishman keeps climbing to 1584. Across the whole window A takes 12.8% of the bass weight against B-flat at 6.5% and E-flat at 5.7%, so the two notes C minor pentatonic requires both sit below the one it excludes. That defeats a strict C-minor-pentatonic reading of the bass. It does not defeat the full-mix label, which could be carried by Trey or Page.
favorite momentMike letting F outweigh C for the only time in the window, with A riding high enough underneath it to settle a key the full-mix detector could not.
C returns · the crest (10:52 to 12:00) · drums peak and the vocal lane lights up
Mike returns to C and puts it back on top at 294, his strongest root since the opening. The section still correlates F major at 0.806, which reads as C functioning as the fifth. Fishman peaks here at 1731, the top of a climb with no dip in it: 1382, 1438, 1584, 1731. The vocal lane's spectral energy jumps to 104.0, roughly four times the middle of the window, and that rise does not survive testing. A 2026-07-19 validation pass on the vocal lane puts the voiced-frame share at 9.5, 9.9, 3.4 and 10.0% across the four sections, so this section carries no more voiced content than the first two, and by RMS it is not even the loudest section (-35.4 dB against -32.3 dB at the open). The level moves and the voice does not. No vocal return is claimed here.
favorite momentFishman's four-section climb landing at 1731, the drums carrying the crest on their own with no vocal entry underneath it.
diagnostic · the SAME passage unsnapped (8:37 to 9:45) · no key assumption, chromatic spelling, K:C
The four staves above were snapped to F major, and a snapped stave cannot be used as evidence for the key it was snapped to. So here is the measurement instead. Across the window the snap changed the pitch class of 10 of 46 notes (21.7%), a median move of 20.5 cents, against a median source tuning error of 17.6 cents. This stave is the section the snap altered most, 5 of 14 notes (35.7%), redrawn with no key assumption at all. The chromatic notes the snap removed are visible in it, including E-flat, F-sharp and B natural. Read the snapped staves as illustration only. The key argument on this page is built from the unsnapped bass pitch-class histogram, which was never passed through a scale. Full per-section figures in _windows/bott_bass_diagnostic.json.
Built from an autocorrelation and chroma pass over the three demucs lanes this jam produced: Mike (bass), Fishman (drums), and a vocal lane used to gate the window. That lane was validated on 2026-07-19 by testing structure rather than level: voiced-frame share across the four sections runs 9.5, 9.9, 3.4 and 10.0%, and median spectral flatness stays near 0.62, which is noise-like throughout. The loudest section by RMS is the opening one at -32.4 dB, not the closing one at -35.4 dB. So the closing rise in vocal spectral energy is a level change with no matching rise in voiced content, and an earlier draft of this page read it as the song returning. That reading is withdrawn. Guitar, piano and other never separated, so Trey and Page are unread. No lead attribution is offered: with only a percussive lane and a bass lane, spectral flux always favours the drums and measures nothing musical. Full bass pitch-class histogram over the window, as a share of clarity-weighted weight: C 28.0, F 18.5, G 15.2, A 12.8, B-flat 6.5, E-flat 5.7, B 2.9, F-sharp 2.8, D 2.0, G-sharp 1.9, E 1.9, C-sharp 1.8. Krumhansl over the bass alone reads F major 0.835 against C major 0.766. Correlation is a similarity score and not calibrated certainty, and a monophonic bass rich in F, A and C will resemble an F-major profile even where the ensemble hears C as the centre, so treat it as a preference. Two earlier passes are being compared and they are separate systems answering separate questions: the windowed mode detector labelled this stretch C minor pentatonic, and the global key detector read C major 0.782 with F major 0.763. Neither is the same measurement as the bass-stem histogram above. Window from the composite energy, novelty, harmonic-flux and interplay picker at 7:30 to 12:00, score 0.6071, dominant feature energy, file-relative. Scored 5.5 out of 10 by an adversarial ChatGPT deep-research pass on 2026-07-18; this revision is the response to it.
← Jam Library Plasma ixi Golden Age ixi WGTYM ixi Stem player