# Score-note presence audit — `I am the method`

## Scope

This test uses the timing alignment already selected by the two internal rests. For each sufficiently long MIDI note, it samples collision-filtered harmonics at three interior points and compares their magnitude with nearby wrong pitches at the same audio moments. It is a score-consistency control, not a message extraction.

## Result

- 885 of 898 MIDI intervals met the conservative duration, harmonic-collision, and control-coverage criteria.
- 808 / 885 eligible notes (91.3%) have stronger expected-pitch support than their same-time wrong-note median.
- Median log support is `1.179`; the fixed-seed paired sign-flip control gives `p = 0.00010`.

The result supports ordinary presence of the reference-score pitch content. It does **not** expose a selected subset of missing notes, a binary alphabet, or readable output. Low-support entries are expected in dense counterpoint and a lossy synthesized performance; without an author-selected extraction rule they are not a cipher.

## Lowest-support entries (inspection list, not symbols)

| Voice | MIDI note | Score seconds | Video seconds | log support |
| ---: | ---: | ---: | ---: | ---: |
| 2 | 69 | 153.500–153.750 | 134.759 | -2.082 |
| 1 | 77 | 133.000–133.250 | 116.776 | -1.889 |
| 4 | 57 | 144.000–145.500 | 126.974 | -1.250 |
| 3 | 55 | 149.000–150.000 | 131.140 | -1.249 |
| 1 | 72 | 76.750–77.000 | 67.434 | -1.205 |
| 1 | 78 | 151.500–151.750 | 133.004 | -1.115 |
| 1 | 77 | 132.250–132.500 | 116.118 | -1.084 |
| 3 | 55 | 85.250–85.500 | 74.890 | -1.041 |
| 2 | 63 | 149.000–150.000 | 131.140 | -0.965 |
| 1 | 72 | 22.250–22.750 | 19.737 | -0.963 |
| 3 | 55 | 151.000–151.500 | 132.675 | -0.895 |
| 1 | 72 | 73.750–74.000 | 64.803 | -0.893 |

## Structured weak-event check

The following groupings are fixed before interpreting any individual low-support event. They show that weaker entries vary with ordinary musical/acoustic factors rather than forming a uniform selected bitstream.

### By reference-score voice

| Group | Events | Expected stronger | Low / tied | Median log support |
| --- | ---: | ---: | ---: | ---: |
| Voice 1 | 237 | 201 (84.8%) | 36 (15.2%) | 1.077 |
| Voice 2 | 251 | 225 (89.6%) | 26 (10.4%) | 1.211 |
| Voice 3 | 211 | 201 (95.3%) | 10 (4.7%) | 1.200 |
| Voice 4 | 186 | 181 (97.3%) | 5 (2.7%) | 1.223 |

### By pitch register

| Group | Events | Expected stronger | Low / tied | Median log support |
| --- | ---: | ---: | ---: | ---: |
| Low register (MIDI 21–59) | 335 | 323 (96.4%) | 12 (3.6%) | 1.200 |
| Middle register (MIDI 60–71) | 407 | 377 (92.6%) | 30 (7.4%) | 1.201 |
| High register (MIDI 72–105) | 143 | 108 (75.5%) | 35 (24.5%) | 0.994 |

### By score phase

| Group | Events | Expected stronger | Low / tied | Median log support |
| --- | ---: | ---: | ---: | ---: |
| Body (before 140 score s) | 817 | 764 (93.5%) | 53 (6.5%) | 1.206 |
| Terminal cadence (140–156 score s) | 68 | 44 (64.7%) | 24 (35.3%) | 0.308 |

### By note duration

| Group | Events | Expected stronger | Low / tied | Median log support |
| --- | ---: | ---: | ---: | ---: |
| Short (under 0.30 score s) | 429 | 385 (89.7%) | 44 (10.3%) | 1.121 |
| Medium (0.30–0.99 score s) | 279 | 260 (93.2%) | 19 (6.8%) | 1.215 |
| Long (1.00+ score s) | 177 | 163 (92.1%) | 14 (7.9%) | 1.201 |

## Limits

- The timing alignment was originally selected from score/audio chroma and rests, so this is not an independent identification of the composition.
- Collision filtering avoids obvious shared harmonics but cannot isolate physical instruments or voices.
- A noisy or low-support note is not proof of omission. Its dependence on register, voice, note length, and the terminal cadence makes it unsuitable as a standalone binary stream.
- The test does not rule out arbitrary keyed transforms, an unavailable pre-YouTube source, or a different author-specified musical rule.

## Reproduce

```powershell
& '<bundled-python>' scripts\audit_contrapunctus_note_presence.py solution\video\i-am-the-method-kQx2G3x7_WQ\analysis\audio-48khz-mono.wav solution\video\i-am-the-method-kQx2G3x7_WQ\reference-score\mutopia-contrapunctusI.mid --timing-audit solution\video\i-am-the-method-kQx2G3x7_WQ\timing-audit\timing-audit.json --output solution\video\i-am-the-method-kQx2G3x7_WQ\note-presence-audit
```

Audio SHA-256: `bbb2d00633a0c9278b2f0b4de12c9eeffaf8d1aadd4ccc5425d3bedcb0f572fc`
MIDI SHA-256: `9d051a6938b09571846983281c2c7e14a447659e8bf0c74724ce6e07c4070528`
Timing-audit SHA-256: `1f213bd4d3435a9df765e79a29bbdbaba978bc46bb5e8fa5e90b8b45482eac38`