Mastering delivery formats: what to deliver
The master sounds right, but the job isn't finished until the client has the correct files for where the music is actually going. Send the wrong sample rate or a lossy file and it either gets rejected by the distributor or, worse, released. Here's what a mastering engineer delivers, in what format, and why.
Two engineers can finish the same master and hand off completely differently. One sends a single lossy file and a "let me know if you need anything else." The other sends exactly the files the release needs, named clearly, at the right resolution, with the reference kept separate from the deliverable. The second one gets rehired. This is what goes in that delivery.
The core deliverable: a high-resolution WAV
The main master is a stereo 24-bit WAV at the sample rate the mix was made at. That's it for the format that matters most. Two rules do the heavy lifting:
- 24-bit, not 16-bit. The extra resolution preserves headroom and detail. You only drop to 16-bit at the very end, for CD, and only once.
- The session's sample rate, not a "better" one. Deliver a 44.1 kHz mix at 44.1 kHz, a 48 kHz mix at 48 kHz. Upsampling a 44.1 kHz mix to 96 kHz invents nothing and just makes a bigger file.
Keep the true peak under the streaming ceiling, around -1 dBTP (Amazon asks for -2). That one is not negotiable, because lossy encoding on playback pushes inter-sample peaks higher. If you want to see where a master sits, the free loudness meter reads true peak with oversampling, and this guide explains the difference between sample peak and true peak.
One master for streaming, a separate one for CD
You often deliver more than one file, but rarely for loudness reasons. Streaming platforms normalize playback, so a single well-made master translates across Spotify, Apple Music, YouTube and the rest; you don't cut a different loudness per platform. What genuinely differs is the destination format:
- Streaming and downloads. The 24-bit WAV at the session rate. This is what distributors ingest.
- CD. A 16-bit 44.1 kHz version, with a sample-rate conversion if the master was higher, and dither applied once at the bit-depth reduction. For replication, that usually means a DDP image rather than loose WAVs.
Loudness itself stays a creative judgment for the song and genre, not a number to hit. A denser, hotter master can be the right call; a quieter, more dynamic one often is too. The reference point most records land near is about -14 LUFS, but treat that as a reference, not a ceiling, and keep the true peak honest.
DDP for physical: the disc standard
If the release is going to physical CD, the plant wants a DDP (Disc Description Protocol) image, not a folder of WAVs. A DDP is a single verified package that carries the audio, the track order and gaps, the PQ codes, and the embedded ISRCs, all in the exact layout the disc will have. It removes ambiguity from replication, which is why it's the standard. For streaming and downloads you do not need one.
Metadata and ISRC codes
The audio is only part of the handoff. Each track needs its ISRC code (the unique recording identifier), plus titles and album information, embedded where the format allows: in the DDP for disc, and provided alongside the WAVs for the distributor to attach. ISRCs come from the label or the distributor, not from the mastering engineer, so confirm you have them before the final render rather than after.
References, instrumentals, and the extras
A dated MP3 reference is worth including for quick, casual listening, on a phone or in the car, clearly labeled so nobody mistakes it for the deliverable. The line matters: the client approves and releases the WAV, never the MP3. Anything else depends on the contract: instrumentals, clean or TV mixes, and stems are common paid extras. Agree on them up front so "can you also send the instrumental" isn't an unpaid afterthought.
Naming, headroom, and a quick checklist
Clear naming saves a round-trip. A convention like ArtistTitle_Master_24-48.wav tells the client what each file is at a glance. Before you send, run the list:
- 24-bit WAV at the session sample rate, true peak under the ceiling.
- A 16-bit 44.1 kHz version or a DDP image if there's a CD, dithered once.
- ISRC codes and titles confirmed and embedded where they belong.
- A labeled MP3 reference, and any contracted instrumentals or stems.
- Filenames that say what each file is, with no version ambiguity.
Deliver them so nothing gets lost or leaked
Getting the formats right is wasted if the handoff is a lossy MP3 in an email or an expiring transfer link. Deliver the exact approved version and its format variants together, password-protected so an unreleased master isn't sitting on an open URL, with the reference and any stems included as extra files. A recorded checksum means what the client downloads provably matches what you rendered.
The short version
Deliver a 24-bit WAV at the session sample rate as the main master, with true peak under -1 dBTP. Add a 16-bit 44.1 kHz version or a DDP image if there's a CD, embed the ISRCs and titles, and include a clearly labeled reference. Name everything so it's obvious, and hand it over as the exact files, protected, rather than a lossy attachment.
Frequently asked questions
What does a mastering engineer deliver?
At minimum, a high-resolution stereo master as a 24-bit WAV at the session's sample rate, with true peak left under the ceiling for streaming. Depending on the release, that's joined by a 16-bit 44.1 kHz version or a DDP image for CD, a dated MP3 reference for casual listening, and any contracted extras like instrumentals or stems. The metadata (track titles, ISRC codes) is embedded where the format allows.
What sample rate and bit depth should a master be delivered in?
Deliver the main master as a 24-bit WAV at the sample rate the mix was made at, commonly 44.1 kHz for music or 48 kHz for anything going to video. Don't upsample a 44.1 kHz mix to 96 kHz; it adds nothing. For CD, convert to 16-bit 44.1 kHz with dither applied once, at that final bit-depth reduction.
What is a DDP file, and do I need one?
A DDP (Disc Description Protocol) image is the standard master format for CD replication: it carries the audio, track order and gaps, PQ codes, and embedded ISRCs as one verified package. You need one if the release is going to physical CD; for streaming and downloads, high-resolution WAVs are what distributors want, not a DDP.
Do I need a separate master for streaming and CD?
Often yes, but not for loudness reasons. Streaming platforms normalize playback, so one master translates across them; the separate CD master exists because CD is a 16-bit 44.1 kHz format, usually a DDP image with PQ codes and ISRCs. Keep true peak under the ceiling either way (around -1 dBTP, -2 for Amazon), and treat loudness as a creative call for the song rather than a target to chase.
What file format should I send a mastered track in?
For the release master, send an uncompressed WAV (24-bit), not an MP3. An MP3 is fine as a labeled reference for quick listening, but it isn't the deliverable, because the client would be approving and releasing something lossy. Keep the WAV at the session's sample rate and deliver it in a way that preserves the exact file.
Deliver every format in one place
Soneam delivers the approved master and its variants together: password-protected, download-tracked, with a reference and stems as extra files and a SHA-256 checksum on record. Your client reviews losslessly and signs off, then gets exactly the files the release needs.