Convert MP3 to MIDI free online: Do You Lose Notes vs WAV? (Measured)

The short answer: when you convert MP3 into MIDI, the part that is true is the bass, but not for the reason most people assume. Pitch detection measures frequency in fixed slices, and at the settings this converter uses one slice is 10.77 Hz wide. A semitone is only 5.95% of a frequency, so below 181 Hz (about F#3) two neighbouring semitones fall inside the same slice. That makes low notes hard to read for any input — WAV included. MP3's lossy coding adds a little extra spectral noise on top, but in our measurements (WAV vs MP3 at 128, 64, 24 and 16 kbps) it did not measurably drop detected notes, even in the bass; at 16 kbps it added only a few spurious low-level notes. So: convert from WAV when you can, because it is cleaner — but the converter's bass behaviour is set by the FFT window, not by whether the file is an MP3. It is a midi mp3 converter online that runs entirely in your browser, so nothing is uploaded and the result is yours to check.

What a pitch detector actually measures

Before talking about codecs, it helps to know what the conversion is doing. This converter resamples your audio to 22 050 Hz, then reads it with a 2048-point FFT stepped forward 256 samples at a time. Two numbers follow directly from that:

Each FFT bin is 22 050 ÷ 2048 = 10.7666 Hz wide.
Each frame covers 256 ÷ 22 050 = 11.610 ms, so there are 86.13 frames per second.

A note is not found by reading off the loudest bin. For a single line, the detector scores every candidate fundamental by summing its harmonics — up to 8 of them, weighted by 1 / k^0.7 — and then refines the winner with a parabolic interpolation across three neighbouring bins. That interpolation is what turns a 10.77 Hz grid into a usable pitch estimate, and it is the part that is sensitive to noise.

The 181 Hz dividing line

A semitone is a fixed ratio, 2^1/12, which means the gap in Hz grows with pitch: it is always 5.9463% of the frequency. Set that equal to one 10.7666 Hz bin and you get 181.06 Hz — MIDI note 53.63, roughly F#3. Below that pitch, two adjacent semitones are closer together than one bin.

Semitone spacing measured in FFT bins, at this converter's 22 050 Hz / 2048 settings.
Note	MIDI	Frequency	Semitone gap	Gap in bins
A1	33	55.00 Hz	3.270 Hz	0.304
E2	40	82.41 Hz	4.900 Hz	0.455
A2	45	110.00 Hz	6.541 Hz	0.608
F3	53	174.61 Hz	10.383 Hz	0.964
C4	60	261.63 Hz	15.557 Hz	1.445
A4	69	440.00 Hz	26.164 Hz	2.430
A5	81	880.00 Hz	52.328 Hz	4.860
A6	93	1760.00 Hz	104.655 Hz	9.720

Read the last column. At A6 a semitone spans nearly ten bins, so a small error in the peak position changes nothing. At A1 a semitone spans under a third of a bin, so the same absolute error moves you a whole note. This is not a flaw specific to one tool — it is arithmetic that any FFT-based detector on these settings has to live with.

Where MP3 actually changes things

MP3 is a lossy, psychoacoustic codec: it spends bits on what a listener is likely to hear and discards the rest, leaving coding noise spread across the spectrum. Three consequences matter for pitch detection:

Slightly noisier harmonic ratios. The harmonic-sum score depends on the relative magnitudes of a fundamental and its overtones. Coding noise perturbs those magnitudes a little.
A possible low-end trim. Some encoders apply a high-pass near 20–50 Hz and a low-pass at the top. The top-end trim is outside the 55–1800 Hz region this converter examines, so it barely matters here; the bottom trim only affects true sub-bass.
More detections at very low bitrates. In our tests, 24 and 16 kbps MP3 returned a few more detected notes than WAV — extra low-level artefacts the detector picked up — never fewer real ones.

The point to hold onto: the coarse low-frequency resolution below 181 Hz is a property of the analysis window, which is identical for WAV and MP3. MP3 changes the signal, but in the band this converter reads it does not flip a recoverable note into a lost one.

What we actually measured: WAV vs MP3 on a buried bass line

Claiming a number you cannot reproduce is worthless, so we ran it. We synthesised a clip with eight bass notes (55–165 Hz) played quietly at 0.18 amplitude under a loud sustained pad at 0.60 — the "buried bass" case the codec is meant to struggle with — and ran it through this site's own pitch detector (dsp.js) as WAV and as MP3 at four bitrates:

Bass notes recovered out of 8 (polyphonic mode, identical detector settings).
Source	Bitrate	Notes detected	Bass notes recovered
WAV	—	27	7 / 8
MP3	128 kbps	27	7 / 8
MP3	64 kbps	27	7 / 8
MP3	24 kbps	28	7 / 8
MP3	16 kbps	31	7 / 8

The one note missed in every version was A1 (55 Hz) — the lowest, quietest note, right at the detector's floor. It was missed by WAV and MP3 identically, which tells you the cause is analysis resolution and threshold, not the codec. MP3 at 16 kbps returned four more detected notes than WAV: extra artefacts, not lost ones. A separate realistic song clip (six bass + two mid notes) recovered 6/6 bass and 2/2 mid at every bitrate from WAV through 16 kbps.

Conclusion: for this converter, MP3 does not measurably drop notes you would otherwise get from WAV. The honest reason to still prefer WAV is cleanliness — no codec artefacts, a more stable harmonic structure — not that MP3 secretly deletes your bass.

Why the difficulty lands in the bass for every format

Low frequencies are not singled out by the codec — they are singled out by arithmetic. The harmonic-sum score still works below 181 Hz, but the parabolic interpolation that refines the pitch has less than one FFT bin to work with, so a small error moves you a whole semitone. That is true for a WAV and for an MP3 of the same take. MP3 can only make it worse at the margin: coding noise is one more source of error on top of an already thin margin, so a very quiet, buried bass note sitting at the detector's threshold may be the first to go — but in our measurements that note was missed by WAV and MP3 alike, not by MP3 alone.

How to test it on your own file

You do not have to take the above on faith, and you should not trust any published accuracy figure you cannot reproduce. This takes about two minutes:

Start from a WAV you own. Export an MP3 from it at the bitrate you normally use.
Convert both files here with identical settings — same mode, same sensitivity, same shortest-note value.
Download the .csv for each. The columns are index,midi_note,note_name,start_sec,end_sec,duration_sec,velocity.
Diff the note_name columns row by row.
On a typical take you may see few or no differences at all — that itself is the finding: the formats agree more than they disagree.

If you do see mismatches, sort them by pitch. The ones below MIDI 54 are the resolution limit acting on both files equally; genuinely MP3-specific drops are rare to absent in practice. Either way the cause is the analysis window, not a fault in the converter.

What actually helps

Convert from a lossless file. WAV, FLAC or AIFF. If your source is already an MP3 you cannot undo the loss, but you can stop adding another generation of it.
Do not re-encode on the way in. Exporting an MP3 from an MP3 to "make it smaller" stacks two codecs' worth of noise.
Mono is cleaner than stereo for a single line. The analysis averages channels to mono anyway, so feeding it a mono file removes the channel-relationship step entirely.
Fix the room before fixing the sensitivity. Reverb and noise hurt the same harmonic measurement that codec noise does. Raising sensitivity on a noisy file keeps more noise, not more notes.
Check the piano roll, not the note count. Two conversions can return the same number of notes and disagree completely about what they are.

Once the notes are in, the file is still raw — How to clean up a converted MIDI before you use it walks the five passes (delete ghost notes, fix velocity, quantize, correct the low end, add instrument, key and channels) that turn it into a usable part.

Questions

Is WAV always better than MP3 for this?

For conversion input, yes — a lossless file gives the detector a cleaner harmonic structure to score. Whether the difference is audible in your result depends on the material: sparse bass lines show it most, dense mid-range material often shows none.

Does a higher MP3 bitrate fix it?

It reduces the noise rather than removing it. Above roughly 256 kbps the remaining differences are usually smaller than the ones caused by reverb, room tone, or a busy arrangement.

Why does the converter resample to 22 050 Hz?

It keeps the analysis cheap enough to run entirely in your browser, and it still leaves a Nyquist limit of 11 025 Hz — well above the 4000 Hz ceiling used for peak detection. Resampling is not what costs you bass accuracy; the bin width that follows from the window size is. It is also why your file's own sample rate — 44.1k, 48k or 22.05k — does not change the notes: Does sample rate change what the MIDI picks up? runs the same melody at all three rates and gets an identical note list every time.

Can I just transpose the result up and convert that?

No. Transposing the audio up before conversion moves the material into a region where semitones are further apart in Hz, which can produce a cleaner read — but you then have to transpose the MIDI back down, and any note that was misread is still misread.

If a bass note comes back wrong, can I fix it?

If a low note is misread or missing, re-converting the same MP3 will not help — the information is already thin in that file. The fix is to go back to the original lossless source and convert that. If you only have the MP3, you can re-synthesize the low end by hand in a DAW. Also try a higher sensitivity or a single-line (mono) conversion on an isolated low part, since the 181 Hz resolution limit is what bit you, not the codec.

Does this happen with AAC or OGG too?

The same principle applies. AAC and OGG are also psychoacoustic, lossy codecs, so they add their own low-level artefacts. The exact behaviour depends on the encoder and bitrate, but the dominant factor — the FFT resolution below 181 Hz — is identical regardless of codec, because it comes from the analysis window. WAV, FLAC or AIFF avoid the artefacts because they are lossless.

Convert a file now — runs in your browser, nothing is uploaded.

Is there a website that listens to an MP3 and turns it into MIDI?

Yes — this one, and it does the listening inside your own browser rather than on a server. So if you want to convert MP3 into MIDI without sending the file anywhere, this page is set up to do exactly that. The page decodes the MP3 using your browser's own audio decoder, reads it frame by frame, groups the frames that share a pitch into notes, and writes the result out as a Standard MIDI file you download directly. There is no upload step, because there is nowhere to upload it to.

The part worth knowing before you point it at a finished song: an MP3 and a WAV of the same take usually come back with nearly the same notes, because the converter's bass behaviour is set by its analysis window, not by the file format. Where they differ, the differences concentrate below roughly 181 Hz — the region where one 10.77 Hz analysis bin is wider than a semitone. When you have the choice, convert from the WAV for a cleaner read.

Does converting MP3 into MIDI lose notes that WAV keeps?