Compress MP3 Files

Most MP3 files in the wild are already at 128–320 kbps. People re-compress them for one of three reasons. First, file size: a 60-minute interview at 256 kbps is 115 MB, which won't fit in a single Discord message (25 MB limit) or a WhatsApp file send (16 MB on most regions). Second, attachment limits: Gmail caps attachments at 25 MB. Third, archival storage: a music collection of 10,000 tracks at 320 kbps takes 100 GB; the same library at 192 kbps takes 60 GB.

MP3 is a lossy format. Re-encoding an MP3 at a lower bitrate is called transcoding, and each transcoding pass discards more audio data. The first transcoding step (320 → 192 kbps) is barely audible on most gear. The second (192 → 128 kbps) is harder to spot but starts to introduce subtle artifacts on cymbals, sibilance, and transients. By 96 kbps you'll hear a 'thinner' mid-range. Below 64 kbps for music, the high frequencies start to swirl. Voice content tolerates much lower bitrates than music does.

The practical advice: if you have the original lossless source (WAV, FLAC), encode FROM that to the target bitrate, don't re-encode the existing MP3. If MP3 is all you have, accept that some quality loss is inherent and pick the highest bitrate that meets your size constraint.

If you're trying to make a file smaller, you have two options: lower the bitrate of the MP3 (this tool), or convert to a more efficient codec. AAC at the same bitrate is roughly 30% better quality than MP3, and Opus at 64–96 kbps for voice is a major improvement over MP3 at the same bitrate.

Trade-off: MP3 plays everywhere with zero compatibility issues. AAC works on basically every modern device but can hit problems with old hardware and some web players. Opus is web-native (Discord, WhatsApp, web audio) but has gaps in older non-web players. If pure file size matters and your target playback environment supports Opus, use the [WAV to Opus](/wav-to-opus) or [MP3 to Opus](/mp3-to-opus) tools instead. If universal compatibility matters most, stick with MP3.

Browser-based compression is a real privacy advantage over the server-based competition. We do not upload your MP3 to any server. There is no temporary storage. There are no logs of what you compressed. The browser downloads FFmpeg WebAssembly once (cached for next time), reads your file from local memory, runs the encoder, and writes the result back to a Blob you download. This matters for client recordings, leaked tracks, NDA-covered material, voice memos, and anything else you don't want to hand to a third-party server.