What we tested
We generated three mono 16-bit PCM WAV files containing the same isolated C4, E4, G4, and C5 sine tones. Each note lasted 0.75 seconds, with a 0.35-second gap. Only the file sample rate changed: 8,000, 22,050, or 44,100 Hz.
All three files went through the current local converter with its default controls. The converter decodes the source in the browser and resamples it to a 22,050 Hz mono buffer before transcription. Each run returned HTTP 200, completed without console errors, and exported one musical MIDI track.
| Input sample rate | WAV size | MIDI events | Detected pitches | MIDI size |
|---|---|---|---|---|
| 8,000 Hz | 74,444 bytes | 7 | C4, E4, G4, C5 | 113 bytes |
| 22,050 Hz | 205,110 bytes | 6 | C4, E4, G4, C5 | 105 bytes |
| 44,100 Hz | 410,174 bytes | 6 | C4, E4, G4, C5 | 105 bytes |
What changed in the MIDI
The 22,050 Hz and 44,100 Hz inputs produced identical note names, starts, durations, velocities, event counts, and file sizes in this test. Each result included the four intended notes plus 0.070-second edge fragments after C4 and E4.
The 8,000 Hz input kept the same four pitch classes, but it started E4 0.023 seconds later, shortened that main event by the same amount, and added a 0.070-second G4 fragment. Its main-note velocities differed by at most two points on the 0-127 MIDI scale from the 22,050 Hz result.
What to do with a low-sample-rate source
- Use the best original you already have. Converting an 8,000 Hz file to a larger number later does not recreate samples that were never recorded.
- Check short edge notes. The extra event in this test was only 0.070 seconds long and used a pitch that was already present.
- Compare timing, not only pitch names. All four intended pitch classes survived, even though one boundary and the event count changed.
- Keep the result as an editable draft. Review note starts, ends, and velocities after transcription.
What this test does not prove
It compares three exact PCM WAV encodings of four synthetic notes. It does not establish a universal minimum sample rate or measure lossy compression, bandwidth-limited speech, instruments, vocals, stereo audio, streamed audio, or dense songs.
For a file-format comparison, see WAV vs MP3 for YouTube to MIDI. If the output contains unwanted fragments, use the extra-note cleanup guide.