PERFORMANCE COMPILER 2 / 09 OCT 2026 / SOURCE-BOUND STUDY
Riding
the pitch.
What if matching the tempo makes the transition worse? We put a familiar DJ control in front of the machine — and listened for the moments when it should leave the pitch fader alone.
The original local audition contains 22 rendered takes across seven source pairs. This public field edition preserves their authored alternatives, actual parameter disclosures and provisional reviewer findings. It does not silently turn an engineering result into a favourite record.
WHAT THE FIRST EARS REPORTED
A correction is not automatically an improvement.
Ghost → Dauerversprechen presented the sharpest contradiction. The first blind listener preferred leaving the 120/132 relationship uncorrected. Shared-speed varispeed was criticised for stutter and harshness; pitch-preserving stretch remained possible but had softened transients in a separate comparison. On Rigid Cell → Outil132, a roughly 2.27% adjustment was often too small for reviewers to distinguish reliably. Those are fallible model observations, not universal verdicts.
Some material had no completed independent listening verdict, and one planned comparison was refused by its reservation budget. They belong in the account too.
SOURCE RELATIONSHIP 01 / 6 TAKES
A Ghost Declines → Dauerversprechen
The same scoped instrumental sources: keep their 120/132 friction, find a shared 126-ish pace, preserve pitch, or let a ride arrive late. The old contour-transfer take remains a distinct comparator.
gd-friction
Keep the original 120/132 tension
Agent-selected listening starting point: preferred over the shared-speed and shared-stretch alternatives in blind review23. The preference is provisional, not proof that rhythmic friction always wins.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 5.000000s; upper foreground 5.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: gd-friction/mix.wav (retained WAV identity declared in the publication manifest; playable from the laboratory index)
gd-catching-up
Ride into agreement; let the approach be heard
Rendered expressive pitch ride. The planned stronger comparative review was refused by the remaining reservation budget before a paid call; no listening verdict is invented.
Authored transport parameters
Outgoing rate 1.00000000..1.04834259; Incoming rate 0.95388664..1.00000000; incoming entry 5.000000s; upper foreground 5.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: gd-catching-up/mix.wav (not a public media URL)
gd-misaligned-control
Same tempos, deliberately displaced entry
Deliberately displaced entry control: identical target pace does not guarantee useful phase or phrasing. No artistic verdict follows from its control label.
Authored transport parameters
Outgoing rate 1.04834259..1.04834259; Incoming rate 0.95388664..0.95388664; incoming entry 5.465125s; upper foreground 5.465125s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: gd-misaligned-control/mix.wav (not a public media URL)
gd-echo-bridge
Keep their speed difference; carry a bright remnant
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 5.000000s; upper foreground 5.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: gd-echo-bridge/mix.wav (not a public media URL)
SOURCE RELATIONSHIP 02 / 5 TAKES
Rigid Cell → Outil132
A small 132.02→135.08 difference is the strongest practical pitch-fader question: leave it alone, slow the incoming record 2.27%, keep pitch with WSOLA, or ride the final correction.
ro-friction
Small mismatch, unmodified
No reliable preference established: listeners did not robustly distinguish this original-rate performance from the small matching changes.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 5.000000s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ro-friction/mix.wav (not a public media URL)
ro-vinyl-match
Slow incoming 2.27%: classic pitch
The listener nominated this input but simultaneously called all three variants indistinguishable. That is insufficient evidence of a musical benefit.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 0.97733824..0.97733824; incoming entry 4.974875s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ro-vinyl-match/mix.wav (not a public media URL)
ro-stretch-match
Slow incoming 2.27%: stable-pitch attempt
Technically rendered and retained. Native recognition was too weak to establish better groove or audible damage relative to this pair’s alternatives.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 0.97733897..0.97733897; incoming entry 4.984875s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ro-stretch-match/mix.wav (not a public media URL)
ro-ridden-entry
Let the incoming pace settle over two seconds
Bounded two-second pitch approach is measurable. The listener did not reliably distinguish it from the unchanged performance and withdrawal control; musical benefit remains unproven.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 0.97733824..1.00000000; incoming entry 5.000000s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ro-ridden-entry/mix.wav (not a public media URL)
ro-deliberate-break
Decline the easy match: a one-beat withdrawal
The listener preferred this input but failed to recognise the deliberate withdrawal reliably. Treat its preference as weak evidence, not a final selection.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 8.450000s; upper foreground 8.450000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ro-deliberate-break/mix.wav (not a public media URL)
SOURCE RELATIONSHIP 03 / 2 TAKES
A Mouthful of Roll → Leaning Cell
Near 120 hypotheses with a weaker outgoing pulse: matching is optional. Test whether a small deliberate acceleration adds urgency or merely distorts the incoming identity.
ml-original
Original pace: no invented correction
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 5.000000s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ml-original/mix.wav (retained WAV identity declared in the publication manifest; playable from the laboratory index)
ml-urgency
Incoming+4%: an expressive push
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.04000000..1.04000000; incoming entry 5.000000s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ml-urgency/mix.wav (retained WAV identity declared in the publication manifest; playable from the laboratory index, alongside the incoming record at its untouched original speed)
SOURCE RELATIONSHIP 04 / 2 TAKES
Sine, Then Skin → Rigid Cell
The 120-ish versus 132-ish reading has half-time ambiguity. Test a deliberate timbral handoff against shared-speed transformation without pretending that the analysis identifies the music's true metre.
sr-natural
Natural pace: the bodies disagree
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 5.000000s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: sr-natural/mix.wav (not a public media URL)
SOURCE RELATIONSHIP 05 / 2 TAKES
Sherbert Oxide → Dauerversprechen
A stronger local pulse appears around the older 32 s Sherbert anchor, but metre and phase remain uncertain. Compare original friction with preserving the incoming pitch at a modest slow-down.
sd-friction
Leave the rhythmic difference exposed
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 5.000000s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: sd-friction/mix.wav (not a public media URL)
sd-stretch
Dauer−4% without intended pitch change
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 0.96000000..0.96000000; incoming entry 5.000000s; upper foreground 7.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: sd-stretch/mix.wav (not a public media URL)
SOURCE RELATIONSHIP 06 / 2 TAKES
Antwort ins Leere → Outil132
The 120→135 difference exceeds a one-deck±8% repair. Refuse that premise: compare an explicit rupture with a bounded−8% incoming stretch that deliberately leaves residual tension.
ao-rupture
Withdraw, then answer at its natural speed
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 9.000000s; upper foreground 9.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ao-rupture/mix.wav (not a public media URL)
ao-bounded-stretch
Maximum−8% pitch-preserving attempt; mismatch remains
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 0.92000076..0.92000076; incoming entry 8.250000s; upper foreground 8.250000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: ao-bounded-stretch/mix.wav (not a public media URL)
SOURCE RELATIONSHIP 07 / 3 TAKES
A Ghost Declines → A Mouthful of Roll
Keep the earlier 64–72 s/32–51 s relationship. A one-second gap already survived agent review; test a changing pitch approach against a plainly comparable shorter withdrawal.
gm-one-second
One-second breath, both original pitches
Comparable reprise of the earlier one-second withdrawal. The new three-way review returned malformed content; its reservation remains consumed and it was not retried.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 9.000000s; upper foreground 9.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: gm-one-second/mix.wav (not a public media URL)
gm-rising-answer
An answer that rises through the pitch fader
A changing-rate arrival, with an original comparison and isolated contribution. The attempted new native review failed; no preference is claimed.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.04000000; incoming entry 9.000000s; upper foreground 9.000000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: gm-rising-answer/mix.wav (not a public media URL)
gm-short-gap
Quarter-second breath, no pitch movement
Rendered comparison; independent listening judgement pending.
Authored transport parameters
Outgoing rate 1.00000000..1.00000000; Incoming rate 1.00000000..1.00000000; incoming entry 8.250000s; upper foreground 8.250000s; fixed authored spectral-lane gain 0.52; no normalisation.
Local render: gm-short-gap/mix.wav (not a public media URL)
EARLIER STUDY / PERFORMANCE COMPILER 1
Before the pitch fader: between records.
Its first bank held 24 authored performances over six source pairs and 114 causal-control records. It explored echoes surviving their dry source, loop and contour inheritance, measured answers and deliberate silence. Engineering tests verified the score and WAV bindings; they did not prove a musical relationship was heard.
11-inherited-contour-foreground · 36s
The inherited figure takes the foreground.
A Ghost Declines lends rhythmic contour to Dauerversprechen's material and leaves finite memory after its parent stops. An attributed source-aware reviewer recognised the rhythmic relation but warned of coarse degradation and fatigue.
05-rupture-r1 · 28s
A breath, then another world.
A Ghost Declines stops. One second is left genuinely empty. A Mouthful of Roll arrives. A paired reviewer preferred that to the two-second alternative; the reviewer's time estimates were not accurate, so the score and PCM own the timing fact.
07-ghost-dauer-overlap · 28s
The functional overlap.
A deliberately plain comparison that worked mechanically, while the echo's perceptual contribution remained unconfirmed. Sometimes ordinary mixing is the better test of an elaborate idea.