The difference in one line
WAV is lossless — it keeps every bit of the audio. MP3 is lossy — it throws away data to make the file smaller. For casual listening you’d struggle to tell them apart. For stems, that thrown-away data starts to matter.
Why the difference shows up with stems: in a full mix, MP3’s compression artefacts hide behind everything else. Isolate a single stem and there’s nowhere for them to hide — the flaws sit right out in the open.
Why lossless helps separation
AI stem separation works from the detail in the file (see how AI stem separation works). The more information the model has, the cleaner it can pull the parts apart. A high-quality WAV hands it everything; a low-bitrate MP3 has already discarded some of that detail before separation even begins, so the resulting stems — especially a bare acapella — can come out slightly rougher.
When MP3 is perfectly fine
Don’t overthink it. If you’re making a karaoke track, practising, or doing quick edits, a good-quality MP3 (256–320 kbps) is fine. The instrumental sits in a full mix where small artefacts disappear. Reach for WAV when you’ll process the stems further — remixing, sampling, or mixing — where every extra edit can compound whatever the format lost.
What to feed the splitter
- Best: a WAV (or other lossless file) at the original quality.
- Fine: a high-bitrate MP3 — 256 kbps or above.
- Avoid: low-bitrate rips and files re-compressed several times — the damage is already baked in.
Whatever you put in, the vocal remover gives you full-quality stems back with no watermark — and you can export them as WAV to keep that quality through the next step.
Frequently asked questions
Is WAV better than MP3 for stems?
For working with stems, yes. WAV is lossless, so it keeps all the detail. MP3 throws away data to save space, and that lost detail can show up as artefacts once a stem is isolated and exposed.
Does the input format affect separation quality?
It can. Feeding a high-quality WAV gives the AI the most information to work with. A low-bitrate MP3 has already lost detail before separation starts, so the stems can come out slightly rougher.
When is MP3 fine?
For casual use — karaoke, practice, quick edits, or listening on the go — a good-quality MP3 is perfectly fine. Use WAV when you'll process the stems further or want the cleanest possible result.
What format should I export my stems as?
Export WAV if you plan to remix, sample or mix the stems — it preserves quality through further editing. Export MP3 if you just need a small file to share or play.
Dan Murtagh is a mixing engineer and audio educator, and the builder of StemConsole. He has spent years separating, mixing and teaching music — StemConsole is the stem tool he wanted to use himself.
More free tools
Vocal Remover
Strip the vocal from any track — clean instrumental + isolated vocal.
Acapella Extractor
Pull studio-quality acapellas for remixing and sampling.
Instrumental Maker
Get a clean, full-quality instrumental of any song.
Karaoke Maker
Turn any song into a karaoke-ready instrumental in minutes.