What we tested
We generated a 22,050 Hz, 16-bit mono WAV containing isolated C4, E4, G4, and C5 sine tones. Each note lasted 0.900 seconds with a 0.300-second gap, a short edge envelope, and a peak amplitude of 0.55.
We then made an inverted copy by multiplying every sample by -1. All 115,763 samples in the second file were the exact integer negative of the first. Both files measured the same 0.55 peak and 0.31481 RMS. We converted them at the site's default controls without changing its logic.
| Condition | MIDI events | MIDI size | Intended pitches retained | Observed difference |
|---|---|---|---|---|
| Normal polarity | 6 | 105 bytes | C4, E4, G4, C5 | Reference |
| Inverted polarity | 6 | 105 bytes | C4, E4, G4, C5 | One velocity step |
The note layout stayed the same
Both outputs placed the main C4 at 0.255 seconds, E4 at 1.451 seconds, G4 at 2.648 seconds, and C5 at 3.868 seconds. Their main-note durations were also identical: 0.835, 0.849, 0.918, and 0.883 seconds.
Each output also contained the same two short edge fragments: C4 at 1.092 seconds for 0.081 seconds and E4 at 2.300 seconds for 0.070 seconds. Polarity inversion did not add, remove, move, or resize an event in this run.
Only one reported velocity changed
Five event velocities matched exactly. The short E4 fragment had velocity 48 in the normal output and 47 in the inverted output. The main-note velocities remained 83, 83, 79, and 79.
That one-step difference means the two MIDI files were not byte-for-byte identical even though their event layouts and 105-byte sizes matched. It is safer to describe this result as nearly unchanged rather than identical.
Polarity inversion is not stereo phase cancellation
Inverting one mono file changes every sample's sign but preserves its amplitude over time. The sound normally remains recognizable when played alone. Cancellation is a different case: if a normal channel and its inverted copy are combined, they can sum toward silence. This test sent each mono version to the converter separately; it did not mix them together.
How to compare polarity versions
- Keep one reference file. Do not overwrite the original when testing an inverted copy.
- Hold every other variable fixed. Match sample rate, bit depth, gain, timing, and converter controls.
- Compare MIDI events, not just playback. Check pitch, start, duration, velocity, track count, and extra fragments.
- Test channel combinations separately. A stereo file with opposite-polarity channels can behave differently during mono downmix.
What this test does not prove
It does not test mastered music, lossy streaming audio, stereo downmix, asymmetric processing, mixed instruments, or every transcription model. It records the current tool's output for one synthetic sequence and its exact inverted copy.