Split an audio file into parts

Drop an audio file here

or click to choose — MP3, M4A, WAV, FLAC, OGG, Opus, AAC

Drop a recording, a podcast, an audiobook or a long MP3 and choose where to cut it: at the silences between sentences or tracks, into a number of equal parts, every few minutes, or wherever you click on the waveform. The parts keep the original format and are not re-encoded, so they sound exactly like the original. MP3 parts carry gapless info, so played one after another they join with no gap and no repeated sound. Nothing is uploaded.

How to split an audio file

  1. Drop an MP3, M4A, WAV, FLAC or Opus file on the box above.
  2. Choose how to split it. At silences finds the pauses for you; Into equal parts and Every… cut by time. The cuts appear on the waveform and in the list of parts. Click the waveform to add a cut, or click a cut to remove it.
  3. Press Split, then download the parts one by one or all together as a ZIP.

Splitting at the silences

The page measures the loudness of the whole file in 20-millisecond steps. It sets the threshold 30% of the way from the quiet background up to the loud parts, and cuts in the middle of every pause that stays below it for longer than the shortest silence you allow. Cutting in the middle leaves a little quiet at the end of one part and the start of the next, so no word is clipped.

We tested it on a 24-second spoken recording with eight pauses of 1.2 to 2 seconds and faint background hiss. With the default settings it found all eight pauses, and every cut landed in the exact middle of its pause in the WAV, MP3, M4A and FLAC copies of the file (within 67 ms in the Opus copy). Recordings differ, so check the list of parts: if a pause between two sentences is missed, raise the dB number a little; if a song or a sentence is split in two, make the shortest silence longer.

Why the parts fit back together exactly

Compressed audio is stored in small frames, and a file can only be cut between them without re-encoding. An MP3 frame is 26 ms long and often borrows data from the frame before it, so a plain cut garbles the first few milliseconds of every part. Many splitters avoid that by decoding the file and saving each part as a new MP3, which compresses the sound a second time.

This page copies the original frames instead. Each MP3 or M4A part starts two frames before its cut and carries a small note, the LAME gapless header in an MP3 and the edit list in an M4A, saying how many samples to skip at the start and where to stop. Players that read the note start every part on the exact sample. FLAC frames don't depend on each other, so FLAC cuts move to the nearest frame edge instead (at most 52 ms away in our test file), and the parts never overlap.

What we measured

The test recording was split at its eight pauses into nine parts. Each set of parts was decoded by Chrome and separately by FFmpeg, joined end to end and compared with the decoded original, sample by sample:

FileFrames copied unchangedJoined lengthJoined sound vs original
WAV— (samples copied)IdenticalIdentical
MP3, 128 kbpsAll 961IdenticalIdentical
M4A (AAC)All 1,069IdenticalIdentical, or within 2.7% where the encoder used noise substitution (see below)
FLACAll 232IdenticalIdentical
OpusAll 1,2450.6 s longer80 ms of lead-in repeated at the start of each part

The M4A test file was made with FFmpeg's AAC encoder, which by default fills some quiet high frequencies with "noise substitution": the decoder generates that noise itself, randomly, so it comes out slightly different depending on where decoding started. With noise substitution turned off in the encoder, the joined M4A parts matched the original sample for sample too. Opus parts keep 80 ms of lead-in because an Opus decoder needs a moment to settle; it is not marked to be skipped, so you hear it if you play the parts back to back.

Splitting the 24-second file took about 0.1 seconds. The MP3 parts together are about 4% larger than the original (404 KB against 389 KB), because each part repeats two frames and has its own header.

When to choose WAV

Choose WAV · exact when the parts are going into a program that may not read the gapless information, such as an older audio editor or a sampler, or when every cut has to be on the exact sample without relying on the player. The file is decoded once and each part is saved as 16-bit WAV, about 10 times the size of a 128 kbps MP3. To keep a single section instead of splitting the whole file, use Trim Audio; to put files back together, use Merge MP3.

The audio never leaves your device: it is read, analysed and split in your browser.

Frequently asked questions

How do I split an audio file at the silences?

Drop the file and keep At silences selected. The page measures the loudness of the whole recording, finds every pause that is quieter than the threshold and longer than the shortest silence you set (0.7 seconds by default), and cuts in the middle of each one. If it finds too many cuts, raise the shortest silence; if it misses pauses, raise the dB number a little (for example from -40 to -35). Click a cut on the waveform to remove it.

Does splitting lower the quality?

Not with Same format. The compressed audio frames are copied into the parts exactly as they are; the result box counts them and in our tests every frame was a byte-for-byte copy. A splitter that saves new MP3 files has encoded the sound a second time, which loses a little more detail each time.

Will there be a gap or a click between the parts?

Not with MP3, M4A, WAV or FLAC. An MP3 can only be cut between frames of 26 milliseconds, so each part keeps two frames before its cut plus gapless info telling the player to skip them and where to stop; M4A parts do the same with an MP4 edit list. In our test, the parts decoded by Chrome and by FFmpeg and joined end to end matched the original sample for sample, with no gap and no repeated sound.

How do I split an audio file into equal parts or every 10 minutes?

Choose Into equal parts and type how many, or choose Every and type the length, such as 10:00 or 90 for 90 seconds. The last part holds whatever is left. Splitting by time is handy for upload limits of transcription services or for sending a long voice recording over a messaging app.

Which formats can I split?

MP3, M4A and other AAC files, WAV, FLAC, and Opus or Ogg files such as WhatsApp voice messages. Each part keeps the format of the original. Choose WAV if you need every cut on the exact sample or a format that any program opens.

Is my audio uploaded?

No. The file is read, analysed and split inside your browser, so interviews, lectures and private recordings never leave your device. There is no account and no limit on the file length.