Delivery: Formats, Dither, Metadata and Sequencing

The last stage of mastering is the least glamorous and the easiest to get wrong. A brilliant master delivered as the wrong file, with a missing code or a misspelt title, can delay a release or put the wrong version in front of listeners.

Which files to produce

For digital distribution (to a distributor that delivers to streaming stores), send a lossless master:

• WAV, usually 24-bit at the session sample rate (44.1 or 48 kHz). Many distributors also accept 16-bit/44.1 kHz; check what yours asks for and whether it accepts high-resolution files.
• The distributor and stores create their own lossy versions from it. Don't send them an MP3.

For other uses, keep a few extra versions:

• 16-bit/44.1 kHz WAV — the CD standard, and a safe universal file.
• MP3 (320 kbps) or AAC (256 kbps) — for sending to people, uploading to sites that don't accept WAV, or previews. Always create these from the high-resolution master, not from another lossy file.
• Instrumental and clean (radio edit) versions if they might be needed — easiest to make now while the session is open.

Dither: only when you reduce bit depth

When you convert from 24-bit or 32-bit float down to 16-bit, the extra detail has to be discarded. Simply truncating it creates quantisation distortion — an ugly, signal-dependent grit most audible in quiet passages and fade-outs.

Dither adds a tiny amount of noise before the conversion, which turns that distortion into a smooth, much less noticeable noise floor.

• Dither once, at the very last step, when you actually reduce bit depth.
• Don't dither a 24-bit or 32-bit float export that stays at that depth — there's nothing to fix.
• TPDF (triangular) dither is the neutral, safe choice. Noise-shaped dither moves the noise to frequencies where hearing is less sensitive; it's fine for a final 16-bit file but shouldn't be processed again afterwards.
• Lossy encoders (MP3/AAC) work from high-resolution input, so you don't need to dither before encoding — feed them the 24-bit file.

Sample-rate conversion

If you mastered at 48 kHz and need a 44.1 kHz file, do the sample-rate conversion before the final limiter where possible, or check the true peak again after conversion — resampling can create small new peaks.

Metadata and identifiers

• ISRC (International Standard Recording Code) is a 12-character code that identifies one specific recording. Each song gets its own, and a remix or new master gets a new one. Your distributor can usually issue ISRCs, or you can get them from your country's national ISRC agency. Keep the same ISRC if you re-release exactly the same recording.
• UPC/EAN identifies the release (single, EP or album) as a whole; distributors typically supply one.
• Track metadata: title, artist, featured artists, songwriter and producer credits, genre, release year and an explicit-content flag. Spell names exactly the same way on every release — stores match artist profiles on them.
• Embedded tags in WAV and MP3 files are useful for files you share directly, but stores take metadata from what you enter at the distributor, not from the file.

Album and EP sequencing

When songs are heard in order, mastering serves the whole record:

• Relative level. Don't make every song the same integrated LUFS. A ballad after a loud single should feel quieter, but not so much that the listener reaches for the volume. Set levels by listening to the transitions.
• Tonal consistency. Play the start of each song after the end of the previous one. A sudden jump in brightness or bass weight is jarring.
• Gaps and transitions. Decide how much silence sits between songs (often 1–3 seconds), or whether songs run into each other. Note that streaming services may play tracks individually or with crossfades, so make each file start and end cleanly on its own.
• Order. Sequence for energy across the whole release: a strong opener, contrast in the middle, an ending that feels like an ending.

Final checklist

1. Integrated loudness and max true peak (≤ −1 dBTP) measured on the final file.
2. 24-bit WAV master at the session sample rate; 16-bit version dithered once.
3. Lossy versions encoded from the high-resolution master.
4. Clean starts and ends, no clicks, fades intact.
5. ISRC per track, consistent titles and credits, explicit flag set correctly.
6. Whole release played through once, in order, from the delivered files.