Why the classic checklists don't map to you
Traditional pre-mastering advice exists because a human mastering engineer can't work well with a mix that's already crushed to the ceiling — so the checklist reaches back into the session and asks for a cleaner bounce. That reach-back is exactly what a file-based workflow doesn't have. A Suno render has no session you can reopen. A bounce from a friend's DAW is whatever they sent. Even your own GarageBand project may be long gone. The good news: an automatic mastering pass works from the file as it is — it analyzes what's actually there and sets its processing accordingly. So the things that matter shift. Headroom numbers matter less; what's in the file, and what you did to it on the way, matter more.
Rule 1: Start from the best source you can get
Everything downstream amplifies what's in the file, so the single highest-leverage decision is which file you start from.
From Suno: WAV downloads are available on the Pro and Premier plans, per Suno's help docs — and the same page notes you need the desktop site, since downloading on mobile defaults to MP3. If you're on the free plan, you get MP3; you can still master an MP3 (more on that in the FAQ), but if you're planning to release the track, the WAV is worth having. One worry you can drop: per Suno's download-limits page, re-downloading a song you've already downloaded doesn't consume your download allowance — so grabbing the WAV of a track you already pulled as MP3 isn't a budgeting mistake.
From GarageBand: Share > Export Song to Disk offers AAC, MP3, AIFF, or WAVE, per Apple's documentation. Pick WAVE or AIFF. (If your export also sounds quiet, that's normal and fixable — we wrote a whole post on why GarageBand exports sound quiet.)
From a collaborator: ask for the uncompressed bounce, ideally the one without loudness processing on the master bus if they have both. That's the one question from the classic checklists that still applies — because they have the session, even if you don't.
Rule 2: Don't re-encode on the way
Lossy encoding (MP3, AAC) throws data away for good; every additional lossy pass stacks more artifacts, and processing a lossy file bakes them in deeper. The practical traps are rarely deliberate conversions — they're transfers. Chat apps that "helpfully" compress audio, quick-share features that transcode, downloading a preview stream instead of the actual file. Move your track by file transfer that doesn't touch the contents — AirDrop, a cloud-drive link to the original file, a USB stick — and it arrives bit-for-bit. One more thing that fools a lot of people: converting an MP3 to WAV doesn't restore anything. The data is already gone; the new container is just bigger. If an MP3 is genuinely all you have, master the MP3 — honestly labeled — rather than laundering it through a format change first.
Rule 3: Don't pre-process it
The second-most-common way to hurt a track before mastering is to "help" it first: an online volume booster, a bass-enhancer app, a normalize pass, a free limiter — and then a mastering pass on top. Two problems. First, peak normalization doesn't do what people hope: it raises the highest sample to a target and changes nothing about how loud the track feels (that's density, not peak — the distinction that explains the entire quiet-export mystery). Second, loudness processing stacks badly: if a booster has already slammed the file into a limiter, the distortion and pumping it added are now in the file, and a later mastering pass can't take them back out — that's squarely in the territory of what mastering can't fix. Give the mastering stage the cleanest, most dynamic version of the track you have, and let one tool do the loudness job once.
Rule 4: Fix upstream problems upstream — before you master
Mastering is a finishing pass on the whole stereo file. If something is wrong inside the track, finishing it louder just gives you a louder version of the problem. So before mastering, spend five minutes on triage — ears only, no meters. We wrote the full version as a three-test triage guide, but the short form:
If words smear or the top end was never rendered, that's a generation problem — regenerate or re-prompt; no downstream tool adds detail that isn't in the file. If one element is wrong while the rest is fine (vocal too far back, pads swallowing the middle), that's a mix problem — with Suno that's what stems are for; with GarageBand, reopen the project if you still have it. Only when the track is coherent but dark, small, and quiet as a whole is the job actually mastering's. Running this triage first is also the budget-friendly order: it costs nothing, and it stops you from spending downloads, credits, or an evening polishing a file that needed regenerating.
Rule 5: Keep the untouched original, and name your versions
Before anything processes the file, duplicate it. Keep a copy that nothing has touched, and name versions so you can tell them apart in six months: song-premaster.wav, song-master-v1.wav, and so on. The pre-master original is the version you'll want when you re-master later with better ears or better tools, when a distributor wants a different loudness target, or when you simply change your mind. Re-mastering a master works much less well than re-mastering the original — each pass of loudness processing narrows what the next one can do.
The 60-second readiness check
You don't need meters to know a file is ready for mastering. Play it once, reasonably loud, on headphones, and ask two questions. One: is it clean? No clicks, crackles, dropouts, or obvious distortion — defects get louder with everything else. Two: is it one coherent balance? Nothing you'd single out as "that element is wrong," per the triage above. If both answers are yes, stop preparing. A finished file doesn't get more ready by being fussed over — over-preparation usually means sneaking in the pre-processing that Rule 3 just warned you about.
Then master it — locally, as many times as you like
If you've followed the five rules, the mastering pass itself is the easy part, and it's the part we build software for. Tonekai is a mastering app for macOS: drop your file on it, and on-device AI analyzes the track and adjusts loudness, tonal balance, and peaks automatically, with an A/B comparison against the original before you export. Because processing is local, the preparation you just did stays meaningful in one more way: the file never gets re-encoded by an upload pipeline, never leaves your Mac, and iterating — master, listen, adjust, master again — costs nothing per pass. Tonekai is free to try; Pro is a one-time unlock, not a subscription.
Download Tonekai on the Mac App Store (free to try)
FAQ
The guides say to leave 6 dB of headroom. My file is already exported — am I stuck? No. That advice targets people bouncing from a session for a human engineer. An automatic mastering pass reads the level that's actually in the file and works from there; what you can't fix after the fact is distortion that's already baked in. A clean file at any sensible level beats a re-export you can't make anyway.
Can I master an MP3? You can, and if an MP3 is all you have (Suno's free plan, for instance), it's still worth doing — loudness and tonal balance improve the same way. Just know the ceiling is lower: artifacts that are in the file stay in the file, and a master made from an MP3 shouldn't be your archival version. When the track matters, get the WAV.
The track I have was already mastered (or loudness-processed). Should I master it again? If you can get an unprocessed version, always prefer it. If you can't, a gentle second pass can still help tone and consistency, but expect less room to work with — and A/B honestly at matched volume to confirm it's actually better, not just louder. Our ears-only checklist covers how.
Do I need to convert sample rates or bit depths first? Don't convert anything preemptively — that's Rule 2 in disguise. Hand the mastering stage the file as it was exported, and deal with delivery formats at delivery time (our pre-release checklist covers what streaming platforms actually ask for).
Wrapping up
"Prepare your mix for mastering" was never bad advice — it was just written for someone else. When all you have is a finished file, preparation means: best source (Rule 1), no re-encoding (Rule 2), no pre-processing (Rule 3), upstream problems fixed upstream (Rule 4), untouched original kept (Rule 5). Run the 60-second check, then let the mastering pass do the one job it's actually for. All of it, start to finish, can happen on the Mac you already own.