Trim audio

Give this page a start and an end and it keeps that section, writing the result as an uncompressed WAV. The decoding and the cutting both happen in this browser tab, so nothing is uploaded and no account is needed. There is a real limit, though, and the page states it: 300 MB per file and 30 minutes of audio on a desktop, half that on a phone. The page also tells you the exact sample the cut landed on rather than repeating the number you typed back at you.

Why a cut cannot land exactly where you asked

Digital audio is a list of measurements taken at a fixed rate — 44,100 per second on a CD, 48,000 in most video. A cut has to land between two of those measurements, so if you ask for 12.5 seconds in a 44.1 kHz file, the nearest boundary is sample 551,250 and the next one is 0.0227 milliseconds later. There is nothing in between, and no tool can put one there.

That gap is far too small to hear, and that is exactly why it is worth being precise about. A tool that quietly reports back the number you typed is easier to trust than one that tells you the truth, and it is also the tool that will hide a real problem from you later. So this page shows the sample index and the actual time it corresponds to, every time.

The same honesty applies to something less obvious. When your source is an MP3 or an AAC file, the decoded audio does not necessarily start at the same instant the container claims it does — the encoder puts warm-up samples in front of the real content. If the two disagree by more than a few milliseconds, this page says so and tells you that every position it reports is measured against the decoded audio, because that is what actually ends up in the file.

Why the output is a WAV, and what that costs

Cutting audio means decoding it. Once it is decoded there are two ways to write it out: compress it again, or store the samples as they are. Compressing again is a second lossy pass, and it damages the audio in a way you cannot undo — the second encoder treats the first encoder’s artefacts as real signal and spends bits preserving them, while adding its own compromises on top.

So this tool stores the samples as they are: 16-bit PCM in a WAV file, the one format that every player, editor and phone reads. There is no quality setting, because there is no quality decision to make.

The cost is size, and it is large. At 44.1 kHz in stereo, uncompressed 16-bit audio is about 10.1 MB per minute. A four-minute clip taken from a 4 MB MP3 comes out around 40 MB. The page shows the exact figure before you press anything, so the trade is one you make deliberately rather than discover in your downloads folder.

To be clear about what this does and does not achieve: no second lossy compression happens here. That is not the same thing as calling the result lossless. If your source was an MP3 or an AAC track, the loss happened when it was first encoded, and decoding cannot undo it. What this tool guarantees is that it adds nothing new.

What you can feed it, and the limits that are real

The decoding is done by your browser’s own audio decoder, so the formats available are the ones it already carries: MP3, WAV, FLAC, AAC inside an M4A or an MP4, and usually Ogg Vorbis and Opus. If a file cannot be decoded, the page says so and names the format it read from the header instead of showing you a generic failure.

There are two limits here and both are technical rather than commercial. Decoded audio has to be held in memory as uncompressed samples, so a file is capped at 300 MB on a desktop and 150 MB on a phone, and total length at 30 minutes on a desktop and 15 on a phone. Those numbers come from what a browser tab can actually hold — about 10 MB of memory per minute of stereo audio, plus the input buffer and the output buffer. We stop before that point rather than let the tab die halfway through and lose your work.

The sample rate and channel count of your file are preserved. This matters more than it sounds: a browser decodes audio at whatever rate the audio context was created with, so the naive way to build this tool silently resamples a 44.1 kHz file to 48 kHz. This page reads the rate out of the file header first and decodes at that rate, so no resampling happens. When the header cannot be read, it says which rate it fell back to.

Nothing is uploaded, and you can check that

Every other online audio cutter sends your file to a server, processes it there and gives you a link. That is why those sites have file size caps, daily quotas and queues — processor time is a bill they pay, so it becomes the thing they ration. It is also why the privacy question is real rather than theoretical: a recording of a meeting, an interview or a voice note is often the most sensitive file a person owns.

Here there is no server in the path at all. Open your developer tools, watch the Network tab and trim a file: your audio appears in none of the requests. The Content-Security-Policy header this site sends restricts connections to our own origin, and it is your browser that enforces it — so a script on this page could not send your file anywhere even if it tried.

After the first visit the page also works with no network at all. The code is cached by your browser, so you can trim audio on a plane.