What we tested
We generated three mono 16-bit PCM WAV files at 22,050 Hz. Each file contained the same 440 Hz A4 beginning at 0.250 seconds, lasting 2.000 seconds, and reaching the same 0.72 peak amplitude. The only changing variable was 5 Hz amplitude modulation: none, 50% depth, or 95% depth.
All files used the same 0.02-second edge envelope and 0.25 seconds of silence before and after the tone. We ran them through the current local converter at its default note-sensitivity and minimum-length settings, then inspected the exported MIDI files.
| Tremolo depth | MIDI events | Detected pitch | Velocity range | File size |
|---|---|---|---|---|
| None | 1 | A4 | 78 | 59 bytes |
| 50% | 10 | A4 only | 68-79 | 140 bytes |
| 95% | 10 | A4 only | 67-72 | 140 bytes |
What changed in the MIDI
The unmodulated tone produced one A4 from 0.221 to 2.254 seconds. The 50% depth version produced ten A4 events, mostly 0.197 to 0.209 seconds long, ending at 2.253 seconds. The 95% depth version also produced ten A4 events, ending at 2.207 seconds.
No new pitch class appeared. Instead, the roughly five pulses per second were reflected as repeated boundaries at the same pitch. Every run returned HTTP 200, used one musical track, and produced no console errors.
How to clean up tremolo-driven splits
- Inspect repeated notes at one pitch. Regular, closely spaced A4 events can describe one modulated source note rather than ten separate strikes.
- Listen before merging. Keep genuine re-attacks; merge only boundaries caused by the effect.
- Recreate tremolo after transcription. A sustained MIDI note plus an amplitude effect can preserve the musical intent with a cleaner piano roll.
- Test the actual recording. Modulation rate, depth, timbre, noise, and converter controls can change the result.
What this test does not prove
It tests one synthetic A4, one 5 Hz modulation rate, and two depths. It does not establish a universal tremolo threshold, cover pitch vibrato, or predict mixed recordings, direct YouTube-link input, other pitches, or other converters.
For a one-way amplitude change, see the fade-in test and fade-out test. For delayed copies of a note, use the echo test.