How to Change the Tempo of a Song Without Changing Pitch

Load the song into a tool that offers time stretching — not a plain playback-speed control — set the new tempo as a percentage or a target BPM, and leave the pitch or key control at zero. The track gets faster or slower while every note stays in its original register.

The three numbers that decide the result are the original tempo, the target tempo, and the ratio between them. A 120 BPM track at 75% plays at 90 BPM; the same track at 125% plays at 150 BPM. Everything else in this guide is a matter of staying inside the range where the algorithm still sounds natural, and choosing a tool whose stretcher matches the material.

“Tempo and pitch are two separate dials everywhere except in the analog world. Use the dial you actually want to move.”

Sound is a wave, and pitch is how many cycles of that wave pass in one second. Play a recording faster and you pack more cycles into each second, so the frequency rises — and frequency is pitch. That single relationship explains both of the classic artifacts: the chipmunk voice when you speed a track up, and the sluggish giant when you slow it down.

A record player, a tape machine, and the playback-rate slider in a plain media player all work this way. They change how fast the audio is read, and pitch follows the speed automatically. That link is perfectly fine if you want the effect. It is the wrong tool if you want a song at a different tempo that still sounds like the same song, in the same key, with the same vocalist.

Splitting the two apart required software that treats time and pitch as independent parameters. That process has a name, and it is the feature you should be looking for in any tool you use.

Time stretching changes how long audio plays without changing its pitch. Instead of reading the whole waveform at a different rate, the algorithm slices the audio into very short overlapping fragments and re-spaces them: further apart to make the track longer and slower, closer together to make it shorter and faster. The seams are blended so you do not hear them.

Most modern tools use a phase vocoder, which analyses the audio in short frequency snapshots and rebuilds each snapshot at the original pitch while the timeline moves underneath. Better engines add transient handling, so drum hits keep their attack instead of smearing into a soft thud. That is why the quality of a stretch depends less on how far you go than on whether the algorithm suits the material.

• Transient or beats mode: for drums and percussion. Protects the attack of each hit.
• Monophonic or vocal mode: for a solo voice or single melodic line. Preserves formants so the singer still sounds like themselves.
• Polyphonic or complex mode: for a full mix or a chord-heavy stem. Smooths the spectrum rather than the transients.
• Spectral or texture mode: for pads and sustained material, where phase coherence matters more than attack.

Time stretching is not the same operation as pitch shifting. Thepitch changermoves the key while the tempo stays put, which is the opposite trade. If you need both — a track transposed and re-timed — the two are usually done as separate passes, and the order matters, as the last section explains.

You cannot compute a percentage change without knowing what you are changing from. Detect the tempo first, and treat the automatic reading as a first guess rather than a fact.

Detection gets two things wrong often enough to matter. It frequently reports half or double the real tempo, because a busy hi-hat pattern can look like the main pulse. And it struggles with live or rubato material that never settles on a grid. TheBPM detectorgives you a number from the audio; verify it by ear before you build anything on top of it.

• Tap along to confirm: the BPM and key finder and the beat counter both let you tap the pulse and read the result.
• Watch for the half-time trap: a reading of 140 BPM on a track that clearly drags is almost always 70 BPM counted twice.
• Check the grid, not just the number: play a click at the reported tempo over eight bars. If it drifts by the end of the loop, the tempo is not constant.
• Write it down: both the original tempo and the target, somewhere you can see them while you work.

The arithmetic is simple, and getting it right before you process anything saves a lot of guessing.

• Target divided by original gives the speed factor. 120 BPM to 90 BPM is 90 ÷ 120 = 0.75, or 75% speed.
• Original divided by target gives the stretch ratio. The same move is 120 ÷ 90 = 1.333, meaning the file becomes about 33% longer.
• Duration scales with the same factor. A three-minute track at 75% speed runs about four minutes.
• Doubling or halving is free. A 92 BPM part sits over a 184 BPM track with no stretching at all, because the pulse divides evenly.

If you would rather not do this by hand, theBPM calculatorconverts between tempo and bar or beat length, which is enough to work out how long the new version will run and whether it still fits the slot you are aiming for.

This is the most common reason people change tempo, and the one where quality tolerances are loosest. If you are using the track to learn a part, a slightly processed version is still perfectly useful.

Start at 60% to 80% of the original tempo, pick out the passage you are working on, and raise the speed in small steps as it becomes comfortable. Adding five percent at a time is enough to keep the muscle memory honest; jumping straight back to full tempo is where most practice sessions fall apart.

• Pitch lock on: the whole point of slowing a song down for practice is that it stays in the original key.
• Loop short sections: four or eight bars at a time, not the whole track.
• Use a lossless source if you have one: stretching a 128 kbps file amplifies its compression artifacts along with everything else.
• Learn the part at slow speed, then re-learn it faster: technique practised at 60% does not automatically transfer at 100%. Step it up.

For instrumentalists working from a recording rather than from notation, the same slowed-down file is also a good transcription aid. If you want the part as editable notes rather than as audio, run the slowed version through the audio-to-MIDI tool and clean up the result in theMIDI editor. Working from a slower source usually gives the detector more to work with on fast passages.

Here the tolerance is much tighter, because the result has to sit against another track rather than being listened to on its own. Two jobs come up constantly.

Blending two tracks. Beatmatching is the one place where a small tempo change is invisible: a gap of a few percent between two songs is usually inaudible with a good algorithm. Work out the ratio from the two detected tempos, stretch the track with the larger gap toward the other, and check the result by ear, not by the number on screen.

Changing the tempo of a stem. If you are rebuilding a track around a borrowed vocal, it is worth separating the parts first. Thestem splittergives you vocals, drums, bass, and everything else, and each stem takes a different stretch algorithm far better than the full mix does. If you only need the instrumental, thevocal removergets you there in one step.

When a large change is unavoidable, prefer the half-time or double-time relationship over stretching. A 70 BPM ballad does not need to be stretched to 140 BPM to sit over a 140 BPM track — the pulse simply divides, and nothing gets processed at all. If the target key also has to change, that is a separate pass with thepitch changer, or a new vocal part through theadd vocals workflow.

“If two tempos are related by a factor of two, use the relationship. Do not stretch what you can simply count.”

Two genres are built on deliberately linked tempo and pitch, and both are cases where you should switch off pitch lock on purpose.

• Slowed and reverb: the whole aesthetic depends on the vocal dropping in pitch as the track slows. Pitch lock would destroy exactly the quality that makes it work.
• Sped-up edits: the lifted, brighter voice is the point. A pitch-locked version at the same speed sounds like a plain faster song, not like the genre.

When you do want a linked move, a plain playback-rate change is the cleanest route, because that is precisely what it does. Resampling an already time-stretched file multiplies the processing and the artifacts, so decide which effect you are after before you start rather than after.

Every algorithm degrades, and the honest answer is that the threshold depends on the material as much as on the tool. These ranges are what to expect in practice.

• Up to about 10%: effectively transparent on most material with a modern algorithm. Listeners will not notice.
• 10% to 20%: audible to a trained ear but usually fine for practice, DJ prep, and casual listening.
• 20% to 25%: the practical ceiling for a release. Drums start to flam, sustained sounds develop a watery edge.
• Beyond 25%: no algorithm sounds transparent. Use a half-time relationship, re-record, or accept the artifact as an effect.

What you will actually hear when you push too far is specific: flamming where one drum hit becomes two, warbling on held notes, a loss of punch on the kick, and phasing across a full mix. Each of those points at a different cause, so listen at the transitions rather than in the middle of a chorus where everything masks everything else.

• Start from the best source you have: a WAV or FLAC gives the stretcher more to work with than a 128 kbps stream rip.
• Split before you stretch: processing stems separately beats processing the mix, because drums can use a transient algorithm while a pad uses a spectral one.
• Work in stages: two moderate stretches often sound better than one extreme one. 160 BPM down to 120 BPM is cleaner as 160 → 140 → 120.
• Keep the processing chain short: stretch, then pitch shift, then stretch again accumulates artifacts from every pass. Plan the order once.
• Re-time tempo-synced effects: delay and reverb tails set at the old tempo are now wrong. This is one of the most commonly missed steps.
• Clean the input: trim silence and reduce obvious noise before you process. The algorithm spends effort on signal, not on room hum.

If the source is drifting — a live take with no click, or an older recording — align it to a grid at its original tempo before you change anything. Stretching drifting audio compounds two problems at once: the drift you started with, and the processing you just applied on top of it.

• Using a playback-speed slider: pitch moves with the tempo. Find the time-stretch or tempo control instead.
• Trusting the detected BPM: half and double are the usual errors. Confirm by tapping or by ear.
• Compromising between two tempos: land on the target, not halfway between the original and the target.
• Stretching the whole mix: separate the stems and match the algorithm to each part.
• Ignoring the artifacts: if the drums flam, the mode is wrong or the stretch is too far. Fix the cause, not the symptom.
• Forgetting what else changed: reverb, delay, and compressor settings tuned to the old groove need revisiting.

Does changing the tempo change the pitch?

Not if the tool uses time stretching. Pitch-preserving tempo changes are the default in most modern editors and browsers tools, so a track slowed from 120 BPM to 90 BPM stays in its original key. If the pitch moves with the tempo, you are using a playback-rate control rather than a stretcher.

How do I slow down a song without changing the pitch?

Open the file in a tool with a time-stretch or tempo control, set a speed below 100% (75% is a good starting point), and make sure the pitch or key setting is at zero. Preview, then export. The song plays slower and stays in the same key.

How do I speed up a song without the chipmunk voice?

Use pitch-locked time stretching and set the speed above 100%. The chipmunk effect comes from raising the playback rate, which lifts pitch along with tempo. Time stretching separates the two, so the track gets faster while voices stay natural.

How far can I change the tempo before it sounds bad?

Roughly 10% either way is transparent with a good algorithm. Up to about 25% stays usable for practice, DJ prep, and most casual listening. Past 25% every algorithm introduces audible smearing on sustained sounds and flamming on drums, so use a half-time or double-time relationship instead.

What is the difference between time stretching and pitch shifting?

Time stretching changes the duration or tempo while holding the pitch. Pitch shifting changes the key while holding the tempo. A plain speed change does neither — it moves both together, the way a record player does.

Can I change a song to an exact BPM?

Yes. Divide the target BPM by the original BPM to get the speed factor — 120 BPM to 100 BPM is 100 ÷ 120, or about 83%. Enter that percentage in the stretcher, or enter the target BPM directly if the tool accepts it. Confirm the original tempo first, because the whole calculation depends on it.

Is changing the tempo of a song legal?

Editing a file for private practice, transcription, or listening breaks nothing. Publishing a modified version of someone else's recording is a different matter, because it uses both the composition and the master. Slowing a track down does not avoid that; treat any public release with the same clearance requirements as a remix.