Does cutting an MP3 lose quality?
Short answer: yes, when the MP3 cutter decodes your file and encodes the cut part again, which is how browser cutters, including the one on this site, work. How much you lose depends on the export setting:
- Export WAV: no audible loss. In our test every sample matched the decoded MP3 to within 1 step of 16-bit rounding.
- Export MP3 at 320 kbps: small loss. Cutting a 128 kbps MP3 and saving it at 320 kbps kept it at 52.4 dB signal-to-error against the file you started with.
- Export MP3 at 128 kbps: clear loss, and it builds up. The same cut saved at 128 kbps scored 17.7 dB, and after ten cuts in a row at 128 kbps the Mozart test fell from 15.8 dB to 6.2 dB against the original recording.
To cut with the least loss, open the audio cutter, pick WAV or MP3 320 kbps under Export format, and always cut from your original file rather than from an earlier cut.
Why cutting an MP3 changes the audio
An MP3 stores sound as compressed frames, not as samples. A cutter can work in one of two ways:
- Frame copy. The tool copies the MP3 frames between your start and end points without decoding them. Nothing is re-compressed, but a cut can only land on a frame edge. An MPEG-1 Layer III frame holds 1,152 samples (ISO/IEC 11172-3), which is 26.1 ms at 44.1 kHz.
- Decode and re-encode. The tool decodes the file to samples, cuts at any point you choose, and compresses the result again. Cut points are exact to the sample, but a second round of MP3 compression is applied.
Browser tools use the second way because the browser only offers a decoder. So the useful question is how much a second (or tenth) round of MP3 compression costs, and which export setting keeps it small. We measured it.
How we tested
- Music: 30 seconds (1:00 to 1:30) of Mozart's Symphony No. 40, first movement, played by the Musopen Symphony. It is in the public domain on Wikimedia Commons as a 48 kHz, 24-bit FLAC. We converted it to 44.1 kHz, 16-bit stereo.
- Test signal: 30 seconds of synthetic chords, bass, noise bursts and a quiet noise floor, which puts energy up to 22 kHz.
- Encoder: the same lamejs build and settings this site's cutter uses (44.1 kHz, constant bitrate).
- Decoder: FFmpeg 4.4.2.
- Score: signal-to-error ratio in dB after lining the two files up. Higher is closer to the reference. This measures how far the waveform moved. It is not a listening test: MP3 deliberately drops sounds people do not hear, so a low score does not always mean an audible problem. Use the numbers to compare settings.
Cutting a 128 kbps MP3 once: which export setting?
This is the common case: you have a 128 kbps MP3 and cut it once. The score compares the cut with the MP3 you started with, so it shows only the loss the cut added.
| Export setting | Score, Mozart | Score, test signal | File size per 30 s |
|---|---|---|---|
| MP3 128 kbps | 17.7 dB | 17.4 dB | 481 KB |
| MP3 192 kbps | 29.9 dB | 22.9 dB | 721 KB |
| MP3 256 kbps | 44.5 dB | 29.8 dB | 961 KB |
| MP3 320 kbps | 52.4 dB | 37.3 dB | 1,202 KB |
| WAV 16-bit | same samples, max difference 1 step of 16-bit rounding | 5,292 KB | |
Saving at a higher bitrate than the source does not bring back what the 128 kbps file already lost. Against the original Mozart recording, the 128 kbps MP3 scored 15.8 dB, and the same file cut and saved at 320 kbps still scored 15.8 dB. What 320 kbps does is stop the cut from adding much new loss. Saved at 128 kbps, the cut dropped to 13.4 dB against the original recording.
Cutting again and again: how the loss adds up
Here we cut a file, then cut the result, and so on, ten times, saving each round at the same bitrate. Scores compare each round with the original recording.
| Bitrate | Round 1 | Round 2 | Round 3 | Round 5 | Round 10 |
|---|---|---|---|---|---|
| 128 kbps | 15.8 dB | 13.4 dB | 11.8 dB | 9.5 dB | 6.2 dB |
| 192 kbps | 28.3 dB | 23.5 dB | 20.5 dB | 16.7 dB | 11.7 dB |
| 256 kbps | 37.3 dB | 36.0 dB | 35.6 dB | 35.4 dB | 35.3 dB |
| 320 kbps | 41.7 dB | 40.2 dB | 39.9 dB | 39.2 dB | 38.7 dB |
At 256 and 320 kbps the loss nearly stops after the first round. At 128 and 192 kbps every round removes more, and after ten rounds at 192 kbps the file is further from the original than a single 128 kbps copy.
The encoder also keeps the same top frequency, so the high end does not shrink round after round. The highest frequency kept stayed near 16.7 kHz at 128 kbps, 18.8 kHz at 192 kbps, 19.5 kHz at 256 kbps and 20.2 kHz at 320 kbps on the test signal.
Each MP3 export also adds 25 ms at the start
Every time the audio went through the MP3 encoder, the decoded result started exactly 1,105 samples later: 25.1 ms of silence at 44.1 kHz. It was the same at every bitrate and every round. The MP3 encoder adds this delay, and the decoder we used did not remove it.
If you cut from the start of an earlier cut, that silence is part of your file and can build up: ten MP3 rounds that each keep the start add about a quarter of a second. Set the start marker slightly later, or export WAV, which adds no silence.
Settings that keep an MP3 cut clean
- Cut from the original file, not from something you already cut and saved.
- Export WAV if you will edit the clip again or need an exact start. It is about 4.4 times the size of a 320 kbps MP3.
- Export MP3 at 256 or 320 kbps if you need an MP3. It adds far less loss than 128 kbps and it does not keep growing when the file is cut again.
- Use 128 kbps only for a final copy, such as a clip to send or a ringtone, where the smaller file matters more.
All of this works in the audio cutter, in your browser, without uploading your file. To stitch clips back together, use the audio joiner.