Outsourcing Your Sound: The Hidden Cost of AI in Music Production
Don't let AI dilute your sound. There are risks to outsourcing your music production - make sure you preserve your sonic fingerprint.
AI tools are revolutionary. They are changing the way work gets done, including what’s possible in creative fields like songwriting and music production, at a staggering pace.
From a listener’s perspective, it’s clear that the audience is already starting to divide. Some people want absolutely nothing to do with music made by AI, while others likely won’t care as long as the song creates a vibe or gives them a good beat to run to.
As musicians, producers, and songwriters, we have to look at this from a completely different angle. When you start handing off various parts of your process to AI tools, there is a very real risk that you are letting the core ingredients of your artistry slip away.
The Slippery Slope of Convenience
The purpose of this article isn’t to claim that using these tools is inherently bad. What’s wrong with using a prompt-based AI plugin, such as Waves Illugen, to generate some samples and atmospheric background textures when you were the one who wrote the prompt? Maybe nothing. Producers have been using sample packs and loops for a long time.
How about feeding your acoustic demo into Suno to hear a few different genre versions and arrangements for fresh production inspiration? Perhaps you aren’t planning to use the audio it generates; you’re just using it as a sounding board. After all, you wrote the core demo composition. What’s wrong with that? Again, maybe nothing.
What about getting ChatGPT to help you brainstorm a few lyric concepts when you’re stuck on the second verse? People have been using co-writers for decades, not to mention tools like RhymeZone and a trusty thesaurus, so what’s the difference?
You might decide that all of this is perfectly fine. I’m not here to dictate what’s right and wrong for your workflow. However, what I am concerned about regarding my own work - and what I would suggest is worth thinking deeply about - is that each small task you offload risks diluting your “sound”. That mysterious, hard-to-define quality that people recognise and love about the music you make is fragile. If you’re not careful, it will slowly fade out of your productions, and you won’t even notice until it’s too late.
Two Questions Every Artist Should Ask Themselves
Before mindlessly integrating AI into your creative workflow, hit pause. I think every artist should seriously consider two fundamental questions:
Which parts of the music-making process give you the most personal satisfaction and fulfilment?
What is the unique ‘sonic fingerprint’ that makes your music unmistakably yours?
(Note: If you’re a part of a band or group, this is a conversation you should be having together.)
Let’s dig into why each of these is a helpful question to ask.
Question 1: Where Do You Find Your Fulfilment?
This comes back to the whole point of doing creative work. Why do you do it? If you like making music, there must be some parts of the process that you particularly enjoy and draw deep fulfilment from. What are those parts?
On the surface, it seems obvious that you’d want to keep doing those specific tasks yourself. If someone (or something) else did them for you, you’d just be giving away the best bits, leaving yourself with the less interesting tasks you don’t get much satisfaction from. Why would anyone do that?
It’s not always that simple, though. In reality, we can sometimes be tempted to skip over even the parts of the process we value the most in exchange for convenience and speed. It’s easy to get sucked into the addictive vortex of external gratification fueled by past successes. The “output” part of the creative process can be extremely rewarding: having people interact positively with your releases, getting radio play, racking up streams, or feeling the buzz of a live crowd singing along to your songs.
The risk here, of course, is that you stop focusing on the actual process of making the work in pursuit of the output. If your workflow becomes a mechanical production line rather than a creative zone because you’ve outsourced the best bits, a rift can form, and you can start losing the meaning you once found in the craft itself. That is not sustainable.
Figure out what parts of the process you genuinely draw meaning and satisfaction from. The work that feels great to have done, even if it was hard along the way. Hold on to those parts; don’t be tempted to farm them out.
Question 2: What is Your Sonic Fingerprint?
Where exactly does your sound live? Does it live in your lyrics? Is there something slightly off-kilter about the arrangement choices that people instinctively recognise as you? Is it the nuance in the particular way you play your instruments, or the chord progressions you always seem to gravitate toward?
Or, as I suspect is the case with my solo project n1ghtmar3cat, does it include the specific way you use recording technology? I think my sound comes from my particular style of songwriting and compositional decisions, combined with the specific, sometimes unorthodox production choices I make.
This is a crucial question to ask, but it’s not an easy one to answer. The reality is that your musical identity is likely a complex web of many different things that give your tracks a recognisable flavour. This includes external influences (the cultures and communities you’ve been exposed to, your artistic heroes, the gear you have access to) and internal factors (your values, subconscious tendencies, and preferences).
You might not even be able to fully recognise your own sonic fingerprint until years after you’ve made a record. Your sound emerges and becomes more distinct as you continue to build your discography, often without you fully realising it. The countless micro-decisions compound over time.
Many prominent bands’ first albums sound like a mix of fairly obvious references. The 1975’s early music strongly resembles Michael Jackson and INXS, but their more recent albums have a sound people recognise as “The 1975”, and they’ve become influential in their own right. Both Radiohead and Silverchair had strong musical ties to the Seattle grunge scene in their early records, but they grew in different directions, and neither much resembled grunge in their later work.
Pointing out these strong early links isn’t a dig; it’s fun to hear artists figure out where they sit in the musical landscape. After the second or third album, the music begins to sound more like them. They’ve done the reps, written a ton of songs, and worked together as a unit long enough that their sound is identifiable as their own, rather than a collection of influences.
Looping back to the topic at hand: I would suggest you don’t let AI anywhere near anything you recognise as contributing to your sonic fingerprint. Don’t outsource the parts of the process that you think are the reasons people love your music, because an AI won’t approach them the way you do.
If you’re unsure whether a particular part of your process might be contributing to your unique sound, think twice before using AI tools there, too. Often, listeners recognise your characteristic quirks more easily than you do, but they’re not in the room to tell you what they are when you’re making your next record. These ambiguous elements might be the ones artists are most likely to outsource, thinking they aren’t important, not realising what they might be losing. If you start outsourcing too many of those micro-decisions along the way, your sound may never get the chance to fully flourish.
The Safe Zone: What to Actually Automate
With all this in mind, my advice is to use AI tools only for parts of your process that you are almost certain do not contribute to your sonic fingerprint.
The most obvious examples include administrative, organisational, and productivity tasks. AI excels here, and utilising these tools can save you a whole lot of time, effort, and stress by removing friction that was previously necessary but came at the expense of time spent actually being creative.
The Premium on Authenticity
As music becomes easier and faster to generate, the market will inevitably flood with perfectly competent, algorithmically polished tracks. When that happens, the messy, human, recognisable “sound” that you’ve fought to protect will become your greatest asset.
People need community, and they inherently crave real, authentic connection. AI can generate a song, but it can’t generate the human experience behind it. By fiercely protecting the parts of the process that make you you, you’re giving listeners something real to hold onto.
Do your best to figure out what those parts are, and keep them in your own hands.
ICYMI: My Latest Release
Copycat EP 01 is a collection of covers dedicated to artists who have influenced my musical identity. It was a gratifying creative challenge to transform the songs into something new with the sonic fingerprint of n1ghtmar3cat while retaining what made them special to me in the first place.
Yeah it’s interesting this. In my professional career I’ve sat on both sides of the outsourcing fence. Customer and supplier. Business process outsourcing and plain old IT outsourcing.
When outsourcing first came along, donkeys years ago, those proposing buying would often cite “it’ll be cheaper” and/or “we’ll be able to tap into all the extra skills the outsourcer has, that we don’t, or we won’t need to hire people when we can’t really justify a full time position”. I don’t recall the supplier trying to dissuade the buyer 😉.
What became apparent over the years was that costs were rarely saved, however you did get a predictable spend. And you rarely needed the sexy skills, which the outsourcer may have, but usually only in very small supply, at all.
What outsourcers did well (if they were good) was run a contract with a clear defined scope of works at or above the service level required. The thing with AI may be different, my gut is saying that won’t be eventually, but we are so early in reality in the cycle / adoption that it’s hard to say. Let’s not forget that behind every AI is a person or two! (For now anyway).
All that said, the things you outsourced were things you didn’t need the skills in and/or things that someone else could do at least as well as you, and vitally, you didn’t outsource things that were your IPR. Your intellectual property. The thing that made you different. The thing that made your customers and employees like working with you.
I don’t believe it’s any different with AI in music. If you give it things that are your IPR to do, why would that ever make any sense.
I think that’s what you’re saying.
Though I concede it’s early on in the cycle and it “may” look cheaper now. My bet is that will change.
I “saved” this one. Thanks, Dave, very helpful.