Audio Speed Changer

Tempo and pitch are treated as two separate numbers, not one control with a checkbox attached. That is the difference between a tool you can use for practice and one you can only use to skim a podcast.

• Tempo 0.25x to 4x — by slider, by percentage, or by typing the length you want the result to be.
• Pitch held, following, or shifted — hold the original key, let it rise and fall with the speed, or move it on its own by up to twelve semitones.
• Live preview, no export — every setting is audible the instant you change it, because the preview never encodes anything.
• A-B loop — mark a passage and it repeats, which is how anyone actually learns a part.
• Lossless at 1.0x — leave the speed alone and the audio is copied rather than re-encoded — instant, and bit-for-bit identical.
• Six output formats — MP3, M4A, AAC, OGG, FLAC and WAV, with bitrate and sample rate you control.
• Batch with one ZIP — queue several files on Pro and take them all in a single download.
• Nothing is uploaded — the whole pipeline runs on your device, so your recordings stay yours.

Speeding up a recording the way a tape machine would makes everything higher as well as faster. That is sometimes exactly what you want and usually not, so this tool lets you pick — and the two options are genuinely different processes rather than one setting renamed.

• Hold the original pitch — the audio is time-stretched: it plays faster while every voice and instrument stays at the pitch it was recorded at. This is the default and the right answer for speech, podcasts, audiobooks and lectures.
• Let the pitch follow — the audio is resampled, exactly as a tape or a record played at the wrong speed. It is mathematically exact — nothing is stretched, so nothing can smear — and it is the sound people mean by 'slowed and reverbed' or 'chipmunk'.
• Shift the pitch on its own — the key moves by the number of semitones you choose and the length does not change at all. Useful for dropping a backing track into a range you can actually sing, or matching two recordings that were made a quarter-tone apart.

The three can be combined: speeding a track to 1.25x while dropping it two semitones is one setting each, not a compromise between them.

Learning a phrase means hearing it many times slowly, and then a bit faster, without hunting for the same four bars again after every repetition. Set A at the start of the passage and B at the end and the preview loops between them while you move the tempo. Because the preview never encodes, you can nudge the speed up a few percent at a time and hear it immediately, which is how the tempo actually gets learned.

• Drag across the waveform to set the loop, or press A and B while it plays.
• Zoom to the loop so four bars fill the waveform instead of four pixels of it.
• Drop the tempo to somewhere you can follow — 0.5x to 0.7x is usually the place to start.
• Leave the pitch held so the part stays in its original key.
• Raise the speed a little at a time until you are back at 1.0x.
• Export only when you want a copy to take somewhere else.

If you leave the speed at 1.0x with no pitch shift, and keep the output in the format the file already is, there is nothing to process. The encoded audio is copied straight into a fresh container: no decode, no encode, no generation loss, and it takes milliseconds rather than seconds. Measured on this build, that path runs the engine for 1 ms against 47 ms for the equivalent re-encode of the same file.

When the copy applies

• The speed is 1.0x and the pitch shift is zero.
• The output format matches the file's own container and codec.
• Bitrate and sample rate are left at their source values.

Ask for any one of those differently and the tool re-encodes instead, and says so before you press anything — a copy that quietly carried a bitrate would not be a copy at all.

The speed change itself is identical across formats; what differs is what the container does with the result afterwards.

• MP3 — plays everywhere. The safe default for speech and for anything going onto an unknown device.
• M4A — AAC in an MP4 container — better than MP3 at the same bitrate, and native on Apple devices.
• AAC — the same codec as a raw stream with no container. Useful when something downstream expects ADTS.
• OGG — Vorbis, open and well supported on desktop. A good choice for the web.
• FLAC — lossless and roughly half the size of WAV. The right export for a stretched master you will edit further.
• WAV — uncompressed PCM. Large, universally readable, and the format most editing software prefers to be handed.

Free handles files up to 50 MB and results up to 30 minutes long. The cap is deliberately on the result rather than the source, because slowing audio down is the one operation that makes a file longer: a half-hour lecture at 1.5x is twenty minutes and perfectly fine, while the same lecture at 0.25x would be two hours of processing. The tempo slider stops where the limit is rather than letting you set something that would be refused later.

Pro removes the imposed ceiling, queues up to twenty-five files at once, exports at 48 kHz and up to 320 kbps, and unlocks the independent pitch shift. There is no server-side number: the real limit is what your device can hold, because that is where the work happens.

• Podcasts and lectures — 1.25x to 1.75x with the pitch held gets through an hour in forty minutes without anyone sounding odd.
• Learning an instrument — 0.5x to 0.75x with an A-B loop over the passage that is not working yet.
• Language practice — slow a native-speed recording to 0.7x to hear where one word ends and the next begins.
• Matching a backing track — shift the key by a couple of semitones without the track getting shorter.
• Fitting a voiceover — type the exact length the edit needs and let the tool find the speed.
• Slowed audio edits — let the pitch follow the speed at around 0.85x for the sound that style is built on.

Two numbers describe every request: the tempo ratio, which decides how long the result is, and the pitch ratio, which decides what key it comes out in. Holding the pitch means stretching time while leaving the pitch ratio at one. Letting the pitch follow means resampling, which moves both together. Shifting the pitch on its own means doing the resample and then putting the tempo back where it started. Each of those is a different chain, and the tool picks the shortest one that produces what you asked for — a request that needs no stretching never goes near a stretcher, which is why the tape-speed setting has no artefacts at all.

The pitch calculation is built from your file's own sample rate rather than an assumed one. That sounds like a detail and is not: a tool that assumes 44.1 kHz detunes every 48 kHz recording — which is most phone audio — by nearly a semitone and a half, silently, on top of whatever shift was asked for.

Processing runs on your device. A ten-minute stereo recording at 1.5x completes in about 3.7 seconds on a modern laptop, and leaving the speed at 1.0x copies the audio in milliseconds instead of re-encoding it.

• Pitch shifter
• Audio cutter
• BPM detector
• Audio compressor
• Audio converter
• Podcast trimmer