A finished mastering job is a package, and a single loud WAV in your inbox means the job is half delivered. What I send out of my Burbank room: a 24-bit WAV at the native sample rate of the mix with true peaks held at -1 dBTP, a dithered 16-bit 44.1 kHz version for CD and DDP use, an instrumental pass of every master, and a text sheet with loudness, true peak, durations and ISRCs for each file. I mastered "Stefania" for Kalush Orchestra, with Sony's release metadata on Qobuz listing the credit alongside three others, and the package below is what that work actually ships as.
What is the primary master file?
One WAV per song, 24-bit, at the sample rate the mix arrived in. A 48 kHz mix becomes a 48 kHz master; converting it to 96 kHz manufactures data without adding information, and converting it down to 44.1 kHz throws information away for no reason, since every modern platform accepts 48. True peaks sit at -1 dBTP so lossy encoders have headroom to work without clipping, and the integrated loudness lands where the record wants it rather than at a magic number, because the platforms will normalize playback anyway. I went through that arithmetic in how platform normalization treats your master: a master crushed to win a loudness contest gets turned down to the same level as everyone else and keeps only its distortion.
This file is the reference. Every other deliverable in the folder is derived from it, through the same chain, in the same pass. If a store or a sync request ever asks for something exotic, this is the file it gets cut from.
Why does CD still get its own file?
Because CD audio is fixed at 44.1 kHz and 16-bit, and getting there from a 48 kHz, 24-bit master takes two deliberate steps: a good sample rate conversion, then dither. Dither is a controlled grain of noise added at the lowest bit, and it matters because simply truncating 24 bits to 16 deletes the quietest level of detail and replaces it with correlated distortion. The dithered file is rendered once, at the end, and never processed again afterward.
When physical discs are being pressed, the plant does not want a folder of WAVs. It wants a DDP image, a standard fileset that contains the audio, the track starts and gaps, the CD-Text, and the ISRC for each track, so nothing is retyped at the factory. Ask for two things with it: the PQ sheet, a human-readable listing of every track time and code that you approve before anything is pressed, and an audio proof of the full disc so the gaps and sequence get heard, once, by the person whose name is on the record.
Do you need an Apple Digital Masters file?
Mostly you already have one. Apple publishes the program's requirements, and the core of it is what a sane package contains anyway: a 24-bit source at the native sample rate, with peaks controlled so Apple's encoder does not clip when it makes the AAC. The badge on the store additionally requires the master to come through a provider on Apple's list and the distributor to flag the delivery. What matters for the sound is the file itself. Deliver the hi-res master and the encoded result is as good as that catalog entry can be, badge or no badge.
Which alternate versions belong in the folder?
The instrumental, always. A sync request moves on a deadline of days, and the question a music supervisor asks first is whether an instrumental exists; the answer has to be a file, not a promise to bounce one next week. Mine goes through the identical mastering chain in the same session as the vocal master, so the two sum-match everywhere except the voice. A clean version, if the lyric needs one for radio or playlists, same chain, same loudness. Shorter broadcast lengths are an editing job rather than a mastering one, and I covered that craft separately in cutting a song to a 30-second spot, but once an edit exists it gets mastered through the same chain too, so a :30 does not sound like a different record than the :60.
Every version is a separate recording with its own identity. Which brings in the codes.
Where do ISRCs and metadata live?
The ISRC is the one identifier that follows a recording for life: twelve characters that route plays to the right rightsholder, and the ISRC system's own documentation is short enough to read in ten minutes. One code per distinct recording: the main master gets one, the instrumental gets its own, the clean version its own. For digital releases the distributor usually assigns them at upload; for a CD they are embedded in the DDP so the pressed disc carries them in the subcode. What mastering adds is the paperwork that keeps it all straight: my delivery sheet lists, per file, the exact filename, duration, integrated loudness, true peak, sample rate, bit depth and ISRC if assigned. Filenames carry artist, title, version and format, so "final_master_v3_REALfinal.wav" never happens: KalushOrchestra_Stefania_Instrumental_48k24.wav answers every question before it is asked.
What do you check when the masters arrive?
Listen to every file, complete, top to tail, on the system you know best. Not the first chorus of each. The errors that survive to delivery live at the edges: a reverb tail clipped by an export region set short, a fade that ends abruptly, a wrong version in a right filename. Then check the numbers against the delivery sheet with a free loudness meter, confirm sample rate and bit depth in the file properties, and play the 16-bit CD version against the 24-bit master at matched volume to confirm nothing changed except the format. Ten focused minutes. A record I produced and mastered for Kalush Orchestra and The Rasmus shipped on Columbia, and the delivery check before a label handoff like that is exactly this list, every time.
What to do with this
Sending a mix out for mastering this month? Ask for the package by name: hi-res master, dithered 16-bit, instrumental through the same chain, DDP with a PQ sheet if discs are coming, and a delivery sheet with the numbers. The credits this practice comes from are documented on my credits page, and the longer background is on the about page.