MP3 to MIDI Frequency Limits: Where the Ceilings Actually Sit
Two ceilings get blamed when an MP3 will not transcribe: the file's sample rate and MP3's high-frequency cut. Both are converted into MIDI note numbers below. One cannot remove a note at all. The other costs three semitones, on one file rate.
Where the ceilings sit
- The highest note MIDI can express is 127, at 12,543.9 Hz. Nothing in the note range goes higher than that.
- MP3's cut is published at 16 kHz, note 131.2 — above the top of the range. It cannot remove a note fundamental.
- The sample-rate ceiling bites only at 22.05 kHz, stopping at note 124.8 and leaving notes 125 to 127 with no fundamental in the file.
- What the encoder removes instead — masking, transients, high-end stereo — costs note onsets, not pitches.
How a frequency becomes a note number
MIDI note numbers run from 0 to 127, middle C is 60, and A4 is note 69 at 440 Hz. Each semitone multiplies the frequency by the twelfth root of two, so f = 440 × 2^((n - 69) / 12) converts between the two. It puts note 127 at 12,543.9 Hz. Every figure below applies that formula to a published limit.
Ceiling one: the sample rate
Sampling carries frequencies up to half the rate, which is arithmetic rather than a quality setting. At 22.05 kHz the ceiling is 11,025 Hz, note 124.8. The top three notes of the MIDI range, 125 to 127, sit above it, so their fundamentals are not in the file. At 32 kHz the ceiling is 16,000 Hz, note 131.2, already clear of the range. At 44.1 kHz it is 22,050 Hz, note 136.8; at 48 kHz, 24,000 Hz, note 138.2. Only the lowest rate costs notes, and it costs three.
Ceiling two: the MP3 low-pass
MP3 encoding discards high-frequency content, and jukeblocks states where: high-frequency detail above 16 kHz is among the parts most commonly removed. That cut is note 131.2. The highest note MIDI can express, note 127, is 12,543.9 Hz, so the cut sits above the whole range and cannot remove a fundamental.
What it removes sits around the notes. The same page lists, beside that cut, quiet sounds masked by louder ones, fine transient detail in drums and percussion, and stereo width in the high end. songscription describes the effect from the other side, saying MP3's discarded detail can blur note onsets and quiet passages. Onsets are timing, so a blurred onset makes a detector place a note late or run two notes together: a rhythm fault, not a pitch fault.
Bitrate changes the size of that loss, not its kind. jukeblocks suggests 192 kbps for demos and 320 kbps where lossless is not required, and says artefacts become audible at 128 kbps, especially on cymbals and high-frequency content.
What to do with this
Check the sample rate first: it is the only ceiling here that returns a flat no. Do not re-encode an MP3 to WAV expecting the detail back, because jukeblocks states that data removed during encoding is not restored. If the transcription drifts in timing rather than landing on wrong pitches, look at the encoder and the room, not the sample rate.
Questions
- Can a higher sample rate give me more notes?
- Only in one case. A 22.05 kHz file cannot carry the fundamentals of notes 125 to 127. At 32 kHz and above, the ceiling is clear of the whole MIDI note range.
- Does MP3 lose high notes?
- It loses frequencies above 16 kHz, but that sits above the highest note MIDI can express. What goes is harmonics, transients and masked detail, not note fundamentals.
- Is 24-bit better than 16-bit?
- Bit depth sets dynamic range, not pitch: about 96 dB at 16-bit and 144 dB at 24-bit. songscription's guidance is that more is not needed, and that a standard 44.1 kHz, 16-bit WAV is plenty.