Taking a piece out of an audio file, step by step
1. Drop the file in
MP3, WAV, M4A, AAC, OGG, FLAC, or the audio inside a video file. It is read straight off your disk. Nothing is uploaded, and nothing is downloaded to make the tool work — the waveform appears in a moment or two.
2. Drag across the part you want
The waveform is the actual shape of your audio, so you can see where the speech starts, where the applause is, where the silence sits. Drag across the stretch you want to keep. Everything outside it dims.
3. Listen before you commit
Press Play the selection and you hear exactly what you have chosen, with a marker running along the waveform. This is the step that turns guessing at timestamps into knowing.
4. Fine-tune to a tenth of a second
The start and end fields stay in step with the waveform and accept seconds or mm:ss. The small buttons underneath nudge each edge by a tenth of a second or a whole one, for when a word is clipped.
5. Choose what comes out
If you brought an MP3 you can have an MP3 back with nothing re-encoded. Any file can be saved as WAV, which is exact to the sample and lets you add a fade in or out.
6. Cut and save
The file is built in your browser and saved straight to your downloads, typically in a few milliseconds. Nothing was sent anywhere at any point, so there is nothing to delete afterwards.
Why this cutter no longer downloads a 30 MB engine
Until today, this page did its work with a copy of FFmpeg compiled to WebAssembly. It is a remarkable piece of engineering and completely the wrong tool for trimming an audio file. The engine is 32,129,114 bytes — about 30.6 MB — and it had to arrive before anything could happen. Measured on the old page: 59 seconds from choosing a 5-second file to getting a 2-second clip back on a first visit, and still 5.6 seconds once the engine was cached.
Your browser already contains an audio decoder. The Web Audio API exposes it through decodeAudioData, which hands back the raw samples of whatever it can open. That is everything a cutter needs: samples to draw as a waveform, samples to play back, and sample positions to cut between. The same 4-second cut now completes in about 7 milliseconds, with nothing downloaded at all.
Cutting an MP3 without re-encoding it
This is the part most online cutters get wrong, and it is worth understanding because it decides whether your file comes back intact.
An MP3 is not one continuous thing. It is a chain of small independent blocks called frames, each holding a fixed number of samples. At 44.1 kHz a frame carries 1,152 samples, which is 26.12 milliseconds of sound. Cutting an MP3 properly means finding the frames that cover the part you want and copying those bytes out untouched. No decoding, no re-encoding, no quality change whatsoever — the audio in your clip is bit-for-bit the audio that was in the original.
Two honest consequences come with that, and this tool states both on screen rather than hiding them.
The first is that the cut lands on a frame boundary. Ask for 12.500 seconds and you will get the frame edge nearest to it, which at 44.1 kHz is never more than 26 milliseconds away. For trimming a song, a lecture or a ringtone this is inaudible. If you need sample-exact placement, use the WAV output instead.
The second is subtler. MP3 allows a frame to borrow unused space from the frames before it, a trick called the bit reservoir. When you cut, the first frame or two of your clip may have been counting on data that is now gone, so the very beginning can be very slightly imperfect. In practice this is a few tens of milliseconds nobody notices in music or speech. It is a real trade-off though, and you should know it exists rather than discover it.
What re-encoding actually costs, and why the old version cost it every time
MP3 is a lossy format. Encoding throws away parts of the sound the encoder judges you will not miss, and that decision cannot be undone. Decode an MP3 and re-encode it and you do not get the same file back — you get a second round of the same judgement applied to audio that has already been through it once. That is generation loss, and it accumulates.
The old version of this page ran that chain on every single cut, whatever you gave it, without saying so. A 320 kbps MP3 came back re-encoded at roughly 190 kbps. A WAV — a format with no loss at all — came back as an MP3.
MP3 or WAV: which one you actually want
Pick MP3 when the source is an MP3 and you want the clip to behave like the original — same bitrate, same size per second, ready to send or upload. This is the right answer for ringtones, for pulling a chorus out of a song, for clipping a podcast segment.
Pick WAV when you need the cut to be exact to the individual sample, when you want a fade in or out, or when the source was never MP3 in the first place. WAV stores audio uncompressed, so the file is much larger: one minute of stereo at 44.1 kHz in 16-bit comes to about 10.1 MB, against roughly 0.92 MB for the same minute at 128 kbps MP3.
One thing WAV cannot do is repair. If the source was a 128 kbps MP3, saving the cut as WAV gives you a large file containing exactly what that MP3 sounded like. It stops any further loss; it does not undo the loss already there.
Reading the waveform
The shape you see is the loudness of the audio over time, drawn from the actual decoded samples rather than an approximation. Tall blocks are loud passages. A flat line through the middle is silence. Speech looks like clusters of bursts with gaps between them, which makes sentence boundaries surprisingly easy to spot; music tends to look like a continuous band with the chorus sitting visibly taller than the verses.
That is enough to find most cut points by eye, and the playback marker settles the rest. What the waveform does not show is pitch or content, so for finding a specific spoken word you will still be listening rather than looking.
The formats your browser can open, and the ones it cannot
This tool decodes with your browser, which means its format support is your browser’s format support, and browsers genuinely differ. MP3 and WAV work everywhere. M4A and AAC work essentially everywhere. FLAC works in current Chrome, Edge and Firefox. OGG Vorbis and Opus work in Chrome and Firefox but have a more complicated history in Safari. WMA generally works nowhere.
Mozilla keeps an up-to-date table of audio codec support if you need to check a specific case. When your browser cannot open a file, this tool says so and names the file rather than producing a vague failure — converting it to MP3 or WAV first will always work.
When this tool stops and tells you something
The version this replaced answered a backwards time range with “Conversion failed: unexpected error. Try a smaller file or a different browser.” Nothing was wrong with the file or the browser. Four things are handled properly now, and each one says what happened:
- An end before the start. The two are swapped and the new range is stated. No error, because this is a typo and not a failure.
- A time past the end of the file. It is pulled back to the real duration, and the message names both your value and the file’s actual length.
- Something that is not a time. The field is restored and the message says which one could not be read.
- A file too large to decode. Above about 250 MB the whole decoded file will not reliably fit in a browser tab, so the tool declines with the size rather than freezing the page.
Nothing you open here goes anywhere
This matters more for audio than for most file types, because audio is often personal — a voice note, an interview, a recording of someone’s family. Your file is read from your disk into this page, decoded in memory, cut, and handed back. There is no upload, no server, no queue, no temporary copy on somebody else’s machine, and nothing to delete afterwards. Close the tab and it is gone.
You can verify that rather than take our word for it: open your browser’s developer tools, switch to the Network tab, and cut a file. You will see no request carrying your audio, because there is none.
Questions people ask when trimming audio
Will cutting my MP3 make it sound worse?
Not on the MP3 setting. The frames covering your selection are copied out unchanged, so the audio is identical to the original. The only caveats are that the cut lands on a frame edge, within 26 milliseconds of where you asked, and that the first moment after a cut can be very slightly imperfect because of the bit reservoir. On the WAV setting there is no loss either, though the file is much larger.
How long can the file be?
Length is not really the limit; size is. Up to about 250 MB works, which is several hours of MP3. The tool decodes the whole file to draw the waveform, so very large files take a few seconds and use a fair amount of memory. Above that ceiling it declines rather than freezing your tab.
Can I cut out the middle and join the two ends?
Not yet. This tool keeps one continuous stretch. Cutting a section out of the middle means keeping two pieces and joining them, which is a different job and a different tool.
Can I use it to make a ringtone?
Yes, and the MP3 setting is the one you want. Trim to the section you like, keep it under about 30 seconds, and save. Because nothing is re-encoded the clip sounds exactly like the original track. What your phone does with the file afterwards is its own business — iPhones in particular want a specific format and a specific folder.
Why did the sample rate stay at 44.1 kHz when my computer runs at 48?
Because that is the correct behaviour and it took deliberate work. Browsers normally resample decoded audio to whatever the sound card is running at, which would quietly turn a 44.1 kHz file into a 48 kHz one. This tool reads the real rate out of the file first and decodes at that rate, so what you put in is what comes out.
Is there a limit on how many files I can cut?
No. There is no account, no daily quota and no queue, because there is no server involved to impose one. The work happens on your machine, so the only limit is your machine.
Does it work on a phone?
Yes. Dragging on the waveform works with touch, and the layout adapts. Very large files are harder on a phone than on a laptop, simply because there is less memory to decode into.
My clip starts with a tiny click. Why?
Two possible reasons. If you cut mid-waveform rather than in a quiet moment, the audio starts partway through a sound wave, and that abrupt start is audible as a click. The fix is the WAV setting with a short fade in — a quarter of a second is usually plenty. The other possibility is the MP3 bit reservoir described above, which produces a much subtler artefact at the very beginning.
Can it cut the audio out of a video?
If your browser can decode the video’s audio track, yes — drop the video in and you will get the sound. The output is audio, so this extracts rather than trims the video itself.
Where has the 30 MB download gone?
It is not needed any more. The old build used a WebAssembly copy of FFmpeg, which had to arrive before it could do anything. Everything this tool does now is handled by the audio decoder already built into your browser.
Other tools people use alongside this one
- Audio Converter — when the format matters more than the length, or your browser cannot open a file and it needs converting first.
- Video to MP3 Converter — for pulling a full soundtrack out of a video before trimming it.
- Video Compressor — the same client-side approach applied to video files.
- Text to Speech — for making the audio in the first place, rather than cutting it.