100% on-device
Compress an audio file
Get a recording, a podcast or a voice memo under a size limit — and know the exact size before it starts. MP3, M4A, WAV and FLAC, all on your own device.
How it works
- Drop your audio. MP3, M4A, WAV or FLAC. Its duration is read immediately.
- Name a size. Audio is the one format whose size is not a guess: at a constant bitrate, bytes are simply duration × bitrate, so the tool solves your limit exactly — and tells you it fits — before encoding a single frame.
- Download. Re-encoded on your device, with the title and artist tags carried over.
In our own test, a 2.3 MB file came out at 938 KB — 60% smaller — at 128 kbps stereo.
Re-encoding something already lossy pays the cost twice
MP3 and AAC discard what a psychoacoustic model predicts you will not hear: frequencies masked by louder neighbours, and everything above a cut-off. Decoding gives back a waveform that is not the original, and encoding it again runs a second model over the first one’s artefacts. A 320 kbps MP3 taken down to 128 is audibly worse than 128 encoded from the master.
That is why an uncompressed source is the good case. WAV is raw PCM, so its size is sample rate times bit depth times channels times duration - CD-quality stereo costs exactly the same per second for silence as for an orchestra. There is real redundancy in that for a codec to remove, and nothing already thrown away for it to compound.
Speech and music want different settings, and the codec is part of that
A voice memo recorded in stereo carries two near-identical channels, so mono halves the data and loses nothing. Speech energy also sits low, which means a sample rate chosen to reach 20 kHz is paying for bandwidth that carries no voice at all. Music is the reverse case - cymbals and string harmonics live at the top, and they are the first thing a low bitrate removes.
Codec choice follows from where the file is going. Opus is markedly better than MP3 at low bitrates and is the right answer for a voice recording that has to be small. MP3 is the one that plays in old car stereos and voicemail systems. An .m4a is already AAC, so re-encoding it to MP3 at the same nominal bitrate is usually a step down.
Questions
Is my recording really not uploaded?
Yes — and not as a policy, as a fact of how it is built. There is no upload endpoint: the file is read straight into this browser tab and compressed by code running on your own machine. You can watch the Network tab while it works, or pull your Wi-Fi out mid-compression and it will still finish. Nothing is stored, nothing is queued, and there is no account.
Will it actually land under my limit?
Yes, and it is the one place we can promise that honestly. MP3 bitrates are not a free dial — they come from a fixed table — so a naive encoder rounds your request up and quietly blows the cap it just told you it would meet. This tool snaps down to the next real bitrate instead, which is why the "fits under 8 MB" you see before encoding is still true afterwards.
Should I convert a WAV or FLAC to MP3?
For sharing, almost always. WAV is uncompressed and FLAC is lossless — both keep every bit of the original, and both are enormous. A spoken-word WAV re-encoded as an MP3 typically loses 90% of its size with no audible difference. Keep the lossless original if it is a master; send the MP3.
Are my title and artist tags kept?
Yes. Tags are your own data, so they are carried across deliberately. (Photos are the opposite case — the image tab strips EXIF and GPS on purpose.)
Speech or music — does it matter?
A lot. Speech survives far lower bitrates than music: a podcast at 64 kbps mono is perfectly listenable and half the size of the same thing in stereo, while music at that bitrate audibly falls apart. If your file is a voice memo or an interview, mono is free money.
Other formats
The same workbench handles all of them — pick a tab above, or start from its own page: