AI Private Coach
Home · Use cases · AI in music production
For musicians and producers · Implementation

AI in music production — without it taking your signature

I spent eight years building audio AI before I started talking about it.

In short: AI has long arrived in music production, just in different places than the headlines suggest: stem separation, cleaning up recordings, finding material in your own library, mastering. Those are the places where it actually saves time. The composition process is not one of them — and whoever approaches it the other way round, loses.

How it works today

  • Hunting for that one 2019 recording you know is somewhere — 40 min
  • Comparing mastering references until your ears are tired
  • Cleaning stems, removing noises one by one — 2 hrs
  • Press texts, funding applications, descriptions: dreaded and postponed

How it works afterwards

  1. Your own library searchable by sound, not just by filename
  2. Cleanup work that used to take hours, done in minutes
  3. Your head stays free for the decisions that are actually musical

What we actually build

1. Split your workflow honestly

We walk through your production and separate two kinds of work: the kind where you make decisions, and the kind where you execute. AI touches only the second kind. This separation takes one session and is the most important part of the whole thing.

2. Automate the cleanup work

Stem separation, noise removal, timing correction, dialogue cleanup for video scoring. The tools here have become good enough that the question is no longer whether to use them, but which one suits your kind of material.

3. Unlock your archive

Most producers sit on ten years of material they can't find again. We build a way to make that archive searchable by sonic criteria — which is exactly the task I worked on for eight years at my own company.

4. Everything around the music

Press texts, funding applications, sync descriptions, label communication, announcements. The part nobody wants to do and that decides visibility. The same voice-profile approach applies here as in content production.

Tools — and where it doesn't help

For stem separation and restoration there are now several good options with different strengths depending on the material — what works for drums isn't automatically the best choice for vocals. On mastering, my position is more reserved than the market's: automatic mastering is a usable starting point and a bad final destination.

Where it doesn't help: writing music. Generative models can produce material that sounds like something — but the decision why this chord and not the other one is the profession. Whoever hands it over, hands over the profession. That's not a romantic position but a practical one: that decision is exactly what clients pay for.

In music, it's less about data protection and more about rights

The decisive questions here are different ones: Which rights do you keep in material that has passed through an AI tool? What do the provider's terms say about using your uploads? And what applies when you produce for a client who knows nothing about it? We go through these points before the first session — they're industry-specific and overlooked almost everywhere.

→ How we bring law and AI together

What this means in time

On cleanup and restoration work, 60 to 80 percent is realistic — that's the area where the tools are genuinely superior. On the rest of the production, considerably less, and that's fine. Anyone promising you 80 percent time savings on an entire production has never made one.

Frequently asked questions

What qualifies you on this topic?

I built an audio-AI company starting in 2013 and sold it after eight years — working with the Fraunhofer Institute, Elobyte and Ursa DSP. Before that, I performed internationally as a musician. I'm not selling courses about tools I know from YouTube.

Is AI taking musicians' work away?

In some areas, yes — production music, background beds and simple jingles are under pressure, and naming that honestly beats talking it away. In other areas, demand grows because cheaper production makes more projects possible at all. Those who know their actual strength find their answer faster.

Can I use AI-generated material commercially?

It depends on the tool, and the answers differ substantially. Some providers grant full commercial rights, others only on certain tiers, others retain usage rights to your output. Sync and advertising jobs often add the client's own requirements. This belongs before the project, not after.

Is automatic mastering good enough?

For demos, references and quick approvals: yes, and much better than three years ago. For a release where the sonic signature is part of the work: no. It hears no context — it knows neither the album nor the intention.

Step 3 of our approach · Implementation

Bring a track you're currently working on

In the free intro call we go through where the time in your production actually sits — and where it doesn't.

1:1 coaching from CHF 320Flat hourly rate CHF 180/h · online or on-site in Basel
Book a free call — 20 min All use cases

Related use cases