Free checklist · Updated 4 November 2026
Pre-release master checklist
Before you upload a master, measure the exported file and confirm five things: integrated LUFS inside −14 ± 1.5, true peak at or below −1.0 dBTP, zero clipped samples, a sample rate of at least 44.1 kHz, and phase correlation of at least 0.2. Then check the tails, the export settings and the metadata below.
Download the PDFOne tidy file you can keep next to the bounce folder.
The five measured gates
| Gate | Pass | If it fails |
|---|---|---|
| Integrated LUFS | −14 ± 1.5 | Gain first, not another limiter |
| True peak | ≤ −1.0 dBTP | True-peak limiter last, re-measure |
| Clipped samples | 0 | Back to the unclipped mix |
| Sample rate | ≥ 44.1 kHz | Re-export at a legal rate |
| Phase correlation | ≥ 0.2 | Mix work, not mastering work |
01Loudness
The store measures the whole file, not the chorus. Measure the exported file the way a store would, then decide.
- Measure integrated LUFS on the exported file, not the session. Whole song, edges included.
- Aim inside −14 LUFS ± 1.5 LU. That is the lane CoProducer scores against. Spotify normalizes to −14 by default; Apple Music is commonly reported near −16, and YouTube, TIDAL and Amazon Music near −14.
- Too hot? Turn the mix down before the limiter, not after another squash. A louder master is turned down on playback, not played louder.
- Too quiet? Fine for dynamic music, but some services turn quiet files up, and the noise floor comes with them.
- Read short-term LUFS beside integrated to see what the chorus does. Integrated is the contract; short-term is the diagnostic.
- Releasing a set? Measure the album as a group so no song jumps out of the run.
- Match a reference as a local file, never a streamed copy that the store has already re-leveled.
Go deeper
02True peak
A WAV that never touches 0 dBFS can still clip after MP3 or AAC. True peak is the number that belongs on the release sheet.
- Final file at −1.0 dBTP or lower true peak. Spotify asks for below −1 dBTP, and below −2 dBTP when a master is louder than −14 LUFS. Amazon Music is often reported stricter.
- Zero clipped samples. Clipping baked into a fixed-point file cannot be undone.
- The true-peak limiter goes last, after any gain, fade or loudness change. Re-measure after every move.
- Turn on oversampling on saturators and clippers that sit before the limiter.
- Encode once to AAC or MP3, then listen and re-measure the encode. That is closer to what the store plays.
Go deeper
03Mono and phase
Phones, clubs and smart speakers collapse stereo. A wide chorus that vanishes in mono is a delivery risk, not a creative choice.
- Phase correlation stays at or above about 0.2 (CoProducer fails lower). Fully mono sits near 1.0; you do not need that.
- Fold to mono and listen: vocal, kick, bass and snare should all still be there, only narrower.
- Keep the lowest octaves mostly centered. Check the polarity and timing of layered kicks and 808s.
- Check left versus right balance, band by band, if the mix leans.
- Audition mono after every width move: wideners, copy-pasted doubles, mid/side tricks.
Go deeper
04Tails, fades and edges
Most last-minute export disasters are at the first and last second of the file.
- Let the tail ring. Export past the last note so reverb and delay decay to silence instead of getting chopped.
- Put the fade-out after the limiter and curve it, so the limiter does not pump the fade.
- Open and close cleanly: no click at the first sample, no stray seconds of silence. Trim the head, then measure again.
- Fade every edit. A cut mid-wave is a click.
- Releasing a gapless album? Encode it and check the joins, because MP3 and AAC can add gaps.
- Listen to the first three and last ten seconds of the exported file on speakers.
Go deeper
05Sample rate, bit depth and export
Boring, visible, non-negotiable. These are the settings a distributor checks before it ever hears the song.
- Export at your session sample rate, never below 44.1 kHz. Do not upsample a lower-rate file to look better.
- If the portal needs a different rate, convert once with a high-quality sample-rate converter on the final bounce.
- 24-bit WAV or FLAC unless the distributor asks for 16-bit. Going down to 16-bit? Dither once, at the very last step. No dither is needed at 24-bit.
- Normalize-on-export off, along with any export-time loudness or gain adjustment.
- A real master, not an MP3 renamed MASTER. One interleaved stereo file unless the portal asks for split mono.
- Name it so nobody has to guess:
Title_master_48k_24b.wav. - Re-open the exported file and measure that, not the session.
Go deeper
06Metadata and ISRC basics
Metadata is how a recording gets paid and found. Fix it before upload, because changing it afterwards is slow.
- ISRC is a 12-character code for one recording: country (2), registrant (3), year (2), then a 5-digit designation. Hyphens are for human display only. Your distributor can assign one if you do not have your own.
- One ISRC per recording. The same recording re-released or put on a compilation keeps its code. A whole-file level change, EQ, dither or sample-rate change does not need a new one. A remix or an edited (for example, clean) version does.
- UPC/EAN identifies the release (single, EP or album), not the track. Distributors normally issue it.
- Spell title, artist and featured artists identically everywhere. Tag versions consistently: (Clean), (Instrumental), (Radio Edit).
- Writers and splits agreed in writing before release. Set the explicit flag honestly.
- Cover art: square, high-resolution (3000 × 3000 px is a common request, so check your distributor), and free of URLs, prices or storefront logos.
- Keep a one-row release log: title, ISRC, UPC, file name, LUFS, true peak, date sent.
Go deeper
07The final proof
The last ten minutes before you press send.
- Measure the file you will upload, then save the report next to it. HTML, JSON or TXT: whichever you will actually find in a year.
- All six gates green: LUFS in lane, true peak ≤ −1.0, clips 0, sample rate ≥ 44.1 kHz, phase ≥ 0.2, and a score you trust.
- Listen once on a phone speaker, in earbuds, and in mono. If it still feels right, it is.
- Archive the session, stems and the exact file you sent, with the date.
- Upload with time to spare. A rejected file late on a Friday is a rough way to find a typo.
Go deeper
Sources: ITU-R BS.1770 (loudness and true-peak measurement); EBU R 128; Spotify for Artists, loudness normalization; AES TD1008; IFPI ISRC Handbook. Platform figures for Apple Music, YouTube, TIDAL, Amazon Music and Deezer are commonly reported values, not numbers those services publish, and services change them. Treat the lane as a safe default, not a law.
Spotted a number that has moved, or a step we missed? Write to NoDAW@nodawlabs.com. How we test, cite and correct.