Audio Quality and Pitch Shifting: What Happens to Your File

Two different operations

There are two ways to lower the pitch of a recording, and they are not equivalent.

Resampling is what happens when you play a tape or record slower. Pitch and duration move together: lower the pitch by 31.8 cents and the track becomes correspondingly longer. This is mathematically trivial and introduces essentially no artefacts, because you are not altering the waveform — only the rate at which it is read.

Pitch shifting lowers the pitch while holding the duration constant. This is the operation almost everyone actually wants, and it is far harder. There is no simple transformation that does it, because you are asking for a signal that did not exist. The algorithm has to analyse the audio, take it apart, and reassemble it — and reassembly is where quality is lost.

Which one we use

Our converter uses pitch shifting, so your file keeps its original length. A four-minute track stays four minutes. This is the right choice for most purposes, but it means the audio is genuinely reprocessed rather than merely replayed.

How pitch shifting works, roughly

The common approach is a phase vocoder. The signal is chopped into short overlapping windows, each is converted to a frequency representation, the frequency content is scaled, and the windows are reassembled with their phases adjusted so the overlaps line up.

That last step is the difficulty. Real audio contains sharp events — a drum hit, a plucked string, a consonant in speech — where energy across many frequencies arrives at exactly the same instant. Windowed analysis smears these across time, and phase adjustment cannot fully restore the alignment.

What that sounds like

Transient smearing. Percussive attacks lose bite. A snare becomes slightly softer, less immediate.
Phasiness. A faint metallic or "underwater" quality, most audible on sustained tones and cymbals.
Reverb tail artefacts. Decay tails can acquire a subtle flutter.
Stereo image changes. When channels are processed independently, the image can loosen. (Not applicable here — see below.)

At a shift of only 31.8 cents these effects are mild. Pitch-shifting algorithms degrade roughly in proportion to how far you move the pitch, and a third of a semitone is a small move. On sparse acoustic material most listeners will not identify an artefact in blind comparison. On dense, transient-heavy electronic music, some will.

Generation loss: the avoidable problem

The single largest quality risk in practice has nothing to do with the algorithm. It is repeated encoding.

Lossy formats — MP3, AAC, Vorbis — achieve their size by discarding information the encoder judges inaudible. That judgement is made against the audio it is given. Decode a lossy file, alter it, and re-encode, and the second encoder makes fresh discards on top of the first set. Do this repeatedly and artefacts accumulate audibly.

Convert once, from the best source you have. A FLAC or WAV original is preferable to an MP3.
Keep the result. Do not re-convert the same file.
Never convert an already-converted file. Shifting a 432 Hz file again does not take you anywhere useful and doubles the damage.

What our converter does to your file — plainly

We would rather state this directly than let you discover it later.

Stage	What happens
Input accepted	MP3, WAV, FLAC, M4A, AAC, OGG — up to 50 MB
Pitch shift	−31.8 cents (−0.3176 semitones), duration preserved
Sample rate	Preserved — a 44.1 kHz file stays 44.1 kHz
Channels	Stereo preserved
Output	MP3, variable bitrate averaging roughly 180 kbit/s

What this means in practice

Your stereo image is preserved. Both channels are pitch-shifted independently and written back as stereo, so the spatial character of the recording survives the conversion.

The output is still lossy. Variable-bitrate MP3 averaging around 180 kbit/s is comfortably good for listening, but it is not an archival format. If you are mastering or feeding the result into further production work, use a desktop tool such as Audacity — set the pitch change to −31.8 cents — and keep your own lossless output.

Getting the best result available

Start from the highest-quality source you have. FLAC or WAV beats a low-bitrate MP3 — the pitch shift is computed on better data, and that difference survives into the output.
Prefer sparse material. Solo piano, guitar, strings and ambient textures survive pitch shifting better than dense percussive mixes.
Convert once and archive the original. Keep the untouched source so you can redo the conversion if you later want a different result.
Judge on your actual playback system. Artefacts audible on studio monitors may be entirely irrelevant on a kitchen speaker.

Perspective

It is easy to become preoccupied with artefacts that are, in normal listening, inaudible. A 31.8-cent shift is a gentle operation, and for the purpose most people have — pleasant background listening — the result is perfectly good. The lossy re-encode is the change most likely to matter, and it matters most if you intend to process the file further rather than simply listen to it.

Knowing the limits lets you decide when the convenient tool is the right one and when it is not. You can try it on our 432 Hz converter; comparing the result with your original on your own system is the only assessment that counts.