MP3 to MIDI short and simultaneous note check: duration and overlap need separate controls.
In this controlled WAV run, the short-note fixtures produced four notes, the simultaneous two-note fixture produced one output note, and silence produced zero.
Input condition
Generated mono PCM WAV fixtures were evaluated with the exact deployed detector and MIDI writer at 11025 Hz. The run compares short separated notes, two notes starting together, and a silent control.
| Input | Observed output | What to inspect |
|---|---|---|
| short-notes.wav | Four output notes at the 100 ms setting | Whether short events remain distinct |
| two-simultaneous-notes.wav | One output note, pitch 48 | Whether overlapping tones collapse |
| silence.wav | Zero output notes | Quiet-input behavior |
The complete recorded results.json and MIDI outputs remain available with the fixtures.
What this does not prove
This run does not establish polyphonic accuracy, a universal minimum note length, or behavior on every recording. It documents only these generated controls and detector settings.
After conversion
Check short-note duration and overlapping-note behavior separately. Compare the generated MIDI against the intended input instead of treating one successful control as a full-song guarantee.
Questions about short and overlapping notes
Did the short-note control produce output?
Yes. The four short-note inputs produced four output notes at the recorded setting.
What happened to the simultaneous two-note input?
The run produced one output note at pitch 48.
Does this prove polyphonic conversion?
No. It is one simultaneous-note control, not a broad polyphonic benchmark.