Downloading and exporting from Suno: stems, WAV, and what to do with them

Stems are your song split back into its parts: the vocal on one file, drums on another, bass, and the rest of the instruments on their own tracks, instead of one flattened mix. You want them the moment a Suno track stops being finished and becomes raw material: to remix it, drop the vocal over a different beat, fix a muddy mix, or duck the music under a voiceover in a video.

What a stem actually is

A regular Suno export hands you one stereo file with everything baked together. Good for sharing, useless for surgery. Once the drums and the vocal live in the same file, you cannot turn the drums down without dragging the vocal down with them. Stems undo that. Each instrument, or group of instruments, comes out on its own track, and every track starts at the same instant, so you drop them into any editor, snap them to zero, and hit play: the same song, now with a fader on every part.

The practical difference is control. With a mix you raise or lower the whole thing at once. With stems you pull the vocal up two dB, mute the bass for a breakdown, replace the drums entirely, or send only the guitar to a reverb. That is the whole reason stems exist: they give back the choices the render took away.

Audio vs stems, and why the format matters

Two separate things get confused constantly. The first is what you get: a single mixed file, or the separated stems. The second is what format and quality it arrives in: a compressed MP3, or an uncompressed WAV.

For listening, an MP3 mix is fine, your ears will not catch the difference on a phone speaker. For working, you want WAV, and ideally stems. Every time an MP3 gets re-encoded it loses a little more; stack three or four edits and the top end starts to smear. WAV holds full quality through as many passes as the project needs. Higher-quality formats and stem export sit on Suno's paid tiers, and those tiers shift over time, so check current pricing at /pricing instead of trusting a number from some article six months old.

If you only ever post the track as is, the free mixed export does the job. The moment you plan to touch it in a DAW, pay for the format that survives editing. Buying WAV stems and then bouncing everything back down to a 128 kbps MP3 to "save space" throws the money away, so keep the project lossless until the very last export.

A workflow for after you export

Say you have four stems: vocal, drums, bass, other. Here is a routine that turns them into a finished track.

1. Line them up. Import every stem and snap each to the project start. If they drift apart, the export went wrong; re-export, do not nudge by ear.
2. Balance first, effects later. Get the raw levels sitting right with nothing but the volume faders. Most "bad" AI mixes are just a vocal three dB too quiet under a loud snare.
3. Fix the one thing that bugs you. Vocal too sibilant? A de-esser on the vocal stem alone, which you could never do on a flat mix. Bass too boomy? A low cut on the bass stem, everything else untouched.
4. Sync to picture, if that is the goal. In a video editor, put the music stems on lower tracks and ride them down under dialogue with a few keyframes, or let the drums land on a cut. Because the drums are their own stem, you can push them louder exactly where the edit wants a hit.
5. Master last, and lightly. One gentle bus compressor and a limiter to bring the whole thing to a sane loudness. Resist the urge to crush it. If the balance in step 2 was right, mastering is a nudge, not a rescue.

The trap here is doing all five at once, reaching for a plugin before the faders are even set. Balance with volume until the song almost works, then reach for anything else.

Clean stems begin at the prompt

Here is the part most people miss, and it is the sunomarket angle: separation quality is decided before you ever hit export. Suno splits a track more cleanly when the arrangement is clean, and the arrangement is clean when the prompt named a small set of distinct instruments instead of a vague wall of sound. Ask for dense layered wall of synths, everything at once and the splitter has to guess where one part ends and the next begins: the vocal bleeds into the pad, the bass smears into the kick. Name four instruments that sit in different frequency ranges and each one comes out with clean edges.

Here is a prompt built for exactly that, a sparse arrangement that separates well.

Style:

minimal indie pop, 90 bpm, one clean electric guitar, upright bass, brushed drums, breathy female lead vocal, lots of space, dry mix

Lyrics (with a vocal anchor so the voice stays in its own lane):

[Verse]
breathy female vocal, close mic, almost spoken
Counting streetlights on the drive back home
Every window dark, every station gone

[Chorus]
Keep the tape rolling, keep the tape rolling
Say it once more before the morning

Exclude:

dense synth pads, heavy reverb wash, layered backing vocals, distorted bass, wall of sound

The lots of space and dry mix in the Style, plus the Exclude list killing reverb washes and stacked vocals, do double duty: they give you the sound you asked for and they hand the splitter clean boundaries to cut along. A reverb tail smeared across the whole stereo field is exactly what makes a vocal stem leak into the music stem.

Common mistakes

• Exporting a mix when you needed stems, then trying to "un-mix" it later. No tool fully un-bakes a flattened render. Decide before you export.
• Editing the MP3. Every save re-compresses. Work from WAV, export MP3 once at the end.
• Nudging stems into alignment by hand. Suno stems share a start point; if yours do not line up, the export is the problem, not your timing.
• Mastering to fix a balance problem. A limiter cannot move the vocal out from under the snare. Go back to the faders.
• Writing a muddy prompt and blaming the splitter. Clean input, clean stems.

FAQ

Does Suno give you stems on the free plan? Stem export and lossless formats live on the paid tiers, and those tiers change, so check /pricing for what is included right now rather than a figure from an old post.

What is the difference between WAV and MP3 out of Suno? WAV is uncompressed, full quality, large files, the format you edit in. MP3 is compressed and smaller, fine for sharing but it loses a little every time it is re-saved. Edit in WAV, share in MP3.

Can I get true multitrack stems, one file per instrument? You get stems along the lines Suno splits, typically vocal and instrumental groupings rather than a fully isolated track for every single instrument. The cleaner and sparser the arrangement, the cleaner those splits come out, which loops right back to the prompt.

Why does my vocal stem still have music bleeding into it? Two usual causes: a dense, reverb-heavy arrangement that never had clean boundaries, or heavy effects smearing parts together. Prompt for a drier, sparser mix and the separation improves. Some bleed is normal; a wall-of-sound track makes it worse.

Should I master inside Suno or in my own tools? If you exported stems to edit, do the final master in your own tools after your edits, not before. Mastering the Suno mix and then editing the stems means you master twice and fight yourself. Edit first, master last.