Mastering

Ideally, all files should be 24-bit (fixed) or 32-bit (float) at a sampling rate between 44.1 and 192 kHz. No dither. If you are using any processing on your master bus that is solely for the sake of loudness (e.g. compression and limiting), please send clearly named versions with it both enabled and disabled. Compression that is purely for vibe or color can be left enabled. If you have been mixing into a compressor and/or limiter and your track clips once you disable them it is easiest to send a 32-bit float bounce. Alternatively, you can pull down your master fader until you’ve eliminated any clipping.

The standard rate includes the following:

• Up to two revisions of each master transmitted via a private, lossless streaming page.

• Hi-res (24-bit, session sample rate) and CD quality (16-bit, 44.1Hz) .wav files of the approved final masters (check with your distributor or aggregator to see which they prefer)

• Reference MP3 and AAC files of the approved final masters

• A detailed QC log of any and all identified and/or corrected technical flaws

• A DDP file set can be provided upon request for an additional fee, but is typically only needed for CD production

Stem mastering is perhaps better described as a hybrid mixing and mastering approach in which you can send me a few sub-mixes (or stems) of different groups of instruments which I can then tweak individually before applying any master processing. Typical groups may be as minimal as drums, instruments, and vocals, or as granular as drums, bass, keyboards, synths, rhythm guitars, lead guitars, backing vocals, and lead vocals. Other combinations are of course possible, but the total number of stems should not exceed 8.

Whether or not stem mastering will benefit your project depends greatly on how confident you are with your final mix. If you’ve been working with another mixing engineer and/or producer you will certainly want to get their input as I would hate to step on their toes and change some balance they’ve spent hours working on. If you have any uncertainty as to whether or not stem mastering is the right approach, I will be happy to listen to your mix and give you my candid feedback.

DDP, or Disc Description Protocol, is a file set that that is used by many professional CD duplication plants. The file set is more robust than a CD-R master and includes several layers of error checking and correction. For this reason, some duplication plants prefer them as it will help them keep errors on a CD to a minimum, and ensure discs with longer lifespans that are compatible with the greatest variety of players. When in doubt, confer with your distributor.

Why AI Will Never Replace a Mastering Engineer

Let’s be honest.
AI mastering tools are impressive. Upload a track, wait a minute, download a “radio-ready” version. Platforms like LANDR and iZotope have made that process fast, affordable, and accessible to anyone with an internet connection.
But speed and accessibility are not the same thing as artistry.
And mastering is not just about hitting targets.

Mastering Is Judgment, Not Just Processing

An algorithm can measure LUFS.
It can analyze transients.
It can balance frequencies based on statistical averages.
What it can’t do is decide why something should feel a certain way.

Should the track breathe, or should it suffocate the listener in density?
Should the low end feel clean and controlled, or slightly unstable and dangerous?
Should the dynamics be preserved for emotional contrast, even if that means being less “competitive”?

Those decisions are not technical.

They’re aesthetic.

And aesthetics come from experience, taste, and context — not probability.

Music Is Culture, Not Data

AI works by learning patterns. It studies thousands of tracks and extracts trends. But music scenes aren’t trends. They’re movements.

A warehouse techno record doesn’t live in the same world as a glossy pop single. A raw underground trap beat doesn’t follow the same rules as a cinematic ambient piece.

A mastering engineer understands subculture, references, history, intent.

AI understands correlation.

That’s a fundamental difference.

Imperfection Is Often the Point

Algorithms are built to optimize, they correct harshness, they smooth resonances, they tighten dynamics, they aim for balance.
But sometimes that slight harshness is what makes a synth scream in the right way.

Sometimes the low end feels powerful because it’s almost too much.
Sometimes the emotional impact comes from tension, not polish.

Great mastering is knowing when not to fix something.

Optimization is not the same thing as taste.

There’s No Conversation

Mastering is collaborative, an artist might say:

“It feels too flat.”
“It needs more weight.”
“The drop doesn’t hit emotionally.”

Those aren’t measurable parameters, they’re feelings.
A human can interpret that language, ask questions, adjust, refine, push back when needed, support when necessary.

AI doesn’t have dialogue, it has presets and prediction.
That’s not the same thing.

Responsibility Matters

When a mastering engineer delivers a final version, they’re putting their name — even if invisible — behind that decision. AI generates an output, it doesn’t stand behind it.

And building a long-term sonic identity for an artist requires continuity, memory, and vision.
Those are human qualities.

AI Is a Tool — A Very Good One

To be clear: AI mastering has value.

It’s great for quick demos, it’s useful for fast references.

It can even get surprisingly close in some cases.
But “close” isn’t the same as definitive.

The final 5% — the part that separates functional from unforgettable — is rarely about numbers. It’s about intuition.

And intuition can’t be automated.

In Conclusion

As long as music is emotional, subjective, and tied to human culture, mastering will remain a human craft.

AI can assist.
It can accelerate.
It can approximate.

But it cannot care.

And in the final stage of a record — when everything is on the line — caring is not optional.