SAM WILLS.

From the Studio

LANDR vs Professional Mastering: An Honest Comparison

Sam Wills·Wired Masters, London·6 min read

AI mastering exists. It's not going away. And it's not all terrible. But it's not all good either, and the difference between what it can do and what a human engineer can do is bigger than most people realise.

What AI mastering actually does well

Credit where it's due. If you've got a really well-balanced mix where everything's sitting nicely and nothing's poking out, the AI tools can do a half-decent job. They'll get it louder, they'll apply some broad tonal shaping, and the result will be better than your raw mix.

Acoustic music, band stuff, anything with natural dynamics and space in the arrangement. That's where AI mastering handles itself best. The productions aren't as crowded, the frequency spectrum isn't as competitive, and there's less for the algorithm to get wrong.

For demos, rough mixes, tracks you want to play out at a set before they're properly finished. There's a place for it. Producers will stick a track through LANDR while they're still working on it just to get an idea of where it'll go. That makes sense. It's a tool.

Where it falls apart

Dance music. Anything kick-driven with a crowded arrangement and a narrow window for where the loudness needs to sit.

In my experience, AI mastering has trouble getting a track to sound loud and detailed and warm while retaining the bottom end. Something usually gives. If you've got a loud master, there's no bottom end. If you've got a nice warm bottom end, it's not a loud master. The algorithm can't navigate that trade-off the way a human can.

If a kick drum is absolutely smashing through, the AI won't handle it the way I would. It doesn't understand that the kick is the point of the track. It treats it like a problem to be tamed rather than the foundation to build around.

Hard techno with zero room and precision requirements? LANDR is not for you. A nice acoustic track that needs bringing up to production level? The AI isn't bad.

What a human hears that an algorithm doesn't

Feel.

When I'm mastering or mixing a track, I don't just look at level meters and I don't just listen. I can feel it. When my monitoring is set to the level I work at and I start pushing the bottom end, I feel the vibration in the floor and through my desk. The bass hitting me in the chest at a certain level tells me something that no meter can.

Having been in this room for a long time, I know that when the bass is kicking me in the chest at a specific point, it's in the right place. When the highs are turned up loud but they're not harsh, they're in the right place. It's a full sensory experience. Not just audio. Physical.

AI mastering can't physically feel a track. And it makes a bigger difference than people think.

Will AI replace mastering engineers?

Maybe. I've definitely thought about it. AI is getting incredibly good across the board. Creativity is more difficult for it, but I won't dismiss the fact that it could get there quicker than most people think. The more it listens to, the more it learns. The more major labels get on board so it can learn from their catalogues, the better and quicker it will get.

It will never do the whole "feel it in your chest and through the floor" thing. There will always be nuance it finds difficult. But that doesn't mean it won't get there eventually.

I think there will always be people who don't want to use it. But if you're already using LANDR and it gets really good, you probably won't have a reason to stop.

What I'd say to "LANDR sounds fine to me"

That's fine. It's your music. It's how you want it to sound.

I'd argue that if you gave me the same track, mine would be better. But if you're happy with what you've got, then you're happy. I'm in a service industry. I do what I think is best for the track. If someone tells me to change it, I change it. I'm not precious about what I do. I want to make sure the client is always happy.

And if they're happy with a LANDR master, then they're happy. No judgement from me.

The real comparison

It's not about AI being bad. It's about what you're trying to achieve and how much the details matter to you.

If you're demoing tracks, testing ideas, or putting out music casually, AI mastering is a perfectly reasonable tool at a fraction of the cost.

If you're releasing on a label, competing for playlist spots, or you care about your music sounding as good as it possibly can, a human engineer working in a proper room, in your genre, who can feel the track as well as hear it, is going to get you further.

You're only as good as your last master. Make sure yours counts.

At a glance

LANDR / AI MasteringSam Wills — Professional
PriceSubscription (£/month)From £100 per track
Genre knowledgeNone13 years, dance specialist
Dance musicConsistently strugglesCore specialism
Stem masteringNot availableFull stem sessions available
Feel and physical responseNoYes — felt as well as heard
RevisionsRe-render with different presetUntil you are happy
Best forDemos, testing, casual releasesLabel releases, serious projects

Work with Sam

Want to hear the difference a human engineer makes? Send your track and Sam will master it properly. Then decide for yourself.

Try a Real Master