Audio Cutter

How to use it

1. Drop your audio file on the dropzone.
2. Drag the start handle to where you want the clip to begin.
3. Drag the end handle to where you want it to stop.
4. Hit play selection to check the edges before committing.
5. Click cut and download. The clip downloads as MP3 or WAV.

What actually happens to the audio

The whole file is decoded first. That is the pause you see after dropping a long recording: the browser has to turn compressed audio into raw samples before it can show you a waveform, and there is no shortcut to drawing a picture of something you have not read. Once decoded, only the range between the handles is copied out and encoded. Nothing outside the selection is touched, and nothing is re-encoded twice.

That distinction matters for quality. Cutting an MP3 and saving as MP3 means one further lossy encode of the selected part — a generation loss you can usually not hear on speech and can sometimes hear on cymbals and applause. If the clip is going into an editor or a sampler rather than straight to a listener, choose WAV. It is roughly ten times larger and it is bit-for-bit what came out of the decoder.

Finding the edit point

The waveform is a peak envelope, not a spectrogram, so it shows you loudness over time and nothing about pitch. Speech is easy to cut this way: words are visible as clumps and the gaps between sentences are obvious. Music is harder, because a downbeat and the note before it can look identical. Use play selection rather than trusting your eyes — the ear finds a bad edit instantly and the eye often does not.

Cut on silence wherever you can. An edit that lands mid-word or mid-note produces a click, because the waveform jumps from some non-zero value straight to zero. If the click bothers you, run the result through fade in and out with a 10 ms fade at each end; that is short enough to be inaudible as a fade and long enough to remove the discontinuity that causes the click.

When this is the wrong tool

One selection, one clip. If you need the recording broken into many pieces at once — an hour of practice takes split into individual songs, say — split audio does it in one pass by silence or by fixed length. If you want to remove several sections from the middle and keep the rest, cut the keepers separately and rejoin them with the joiner; there is no multi-region delete here, deliberately, because the interface for it is worse than doing it in two steps.

If what you actually want is to trim the dead air at the very beginning and end of a take, trim silence at the edges does that automatically and gets it right more often than a hand-placed handle does.

Working with video

Dropping an MP4, MOV, WebM or MKV works: the audio track is pulled out and you cut that. What comes back is audio only, always — there is no video output here. If you wanted the video trimmed as well, this is the wrong site, and a local tool like Handbrake or the trim function in CapCut or DaVinci Resolve is what you want. If you only ever wanted the sound, MP4 to MP3 skips the waveform and converts the whole thing.

Size and speed

The practical ceiling is around 500 MB, and it is memory rather than time that stops you: the decoded audio sits in RAM uncompressed, so a 200 MB MP3 can occupy well over a gigabyte once it is samples. A phone will give up sooner than a laptop, and Safari on iOS sooner than Chrome on Android, because iOS caps how much memory a single tab may hold. If a long file fails, converting it to a lower sample rate first with change sample rate shrinks the decoded size proportionally and often gets it through.