Ai in Music
There is no doubt that little can compare the vitriol generated in the discussion of AI in music. There are haters and lovers of Suno and Udio and all the other players.
The thing is that whilst these are innovative, AI has been 'hidden' in music production for some years
AI is widely used in recording studios and professional music production today (as of 2025–2026), primarily as assistive tools for technical tasks rather than full generative replacement of human work. Adoption among professionals is high—typically 70–80%+ report using some form of AI—though intensity varies, and generative AI (full song creation) remains less central than tools for cleanup, separation, mixing assistance, and efficiency.
Key Adoption Figures - Surveys of music professionals, producers, engineers, and musicians consistently show broad uptake:~77% of music professionals use some form of AI (Production Expert poll of ~570 responses). Most of this is assistive only (~51%), with smaller shares combining assistive + generative occasionally (~17%) or daily (~7%). Pure generative use is rare (~1.6%). Post-production sees even higher rates (~89%).
78% of professional musicians use AI for music-related work (vs. ~60% of hobbyists), per a Water & Music × Moises survey of 1,525 musicians. Pros are also more willing to pay for tools.
~70% experiment with or use AI at least occasionally, with about 1 in 5 as regular users (Sonarworks surveys of 1,100+ producers/engineers/songwriters).
Other figures include LANDR surveys showing ~87% of music makers incorporating AI somewhere in their workflow, and Muse Group data indicating ~70% of surveyed US musicians already using it.
These numbers reflect professionals more than pure hobbyists, and “use” often means targeted plugins or features rather than relying on AI for entire tracks.
The dominant applications are practical and technical (the “unglamorous stuff”), not end-to-end generation: Stem separation / vocal isolation — Frequently the top use case (around 71% among AI users in the Moises survey). Tools extract vocals, drums, bass, etc., from mixed tracks for remixing, restoration, or cleanup—something that was nearly impossible a few years ago.
Noise reduction, audio restoration, and cleanup are very common (e.g., iZotope RX, Waves Clarity Vx). These help with recorded material in real sessions. Mixing and mastering assistance — AI-powered assistants in tools like iZotope Neutron/Ozone, LANDR, or similar analyze tracks and suggest EQ, compression, levels, or produce reference masters. Used as starting points that humans refine.
Other technical aids — Automatic leveling, session organization, timing fixes, microphone/amp modeling (e.g., AI-based virtual systems), and speech-to-text for dialogue/lyrics.
Creative/ideation support — Generating ideas, melodies, chord progressions, accompaniments, or samples are less dominant; full song generation from text prompts ranks lower, e.g., ~24%). Tools like Suno or Udio appear in demos, songwriting rooms (especially noted in Nashville/country contexts), and sample creation (e.g., AI funk/soul samples in hip-hop).
DAWs and plugins increasingly embed AI features (Ableton, Pro Tools, etc.) for labeling, suggestions, or real-time processing.
In professional studios, AI functions more like an advanced assistant engineer—handling repetitive or time-consuming work so humans focus on creative decisions, performance, and final polish. High-end commercial records still typically involve significant human oversight; fully AI-generated commercial releases exist but are more common in independent, demo, or high-volume content spaces.
Assistive tools (cleanup, separation, mixing aids) dominate professional workflows and face less resistance.
Generative tools (Suno, Udio, etc.) are growing fast for ideation and demos but are less routinely used for final commercial products due to quality, legal/ethical concerns, originality issues, and industry pushback.
Variations by role/genre — Higher in post-production, independent/project studios, and certain genres (e.g., hip-hop sampling, country demos).
Traditional high-budget sessions may use it more selectively. Many view it as efficiency-boosting and non-replaceable for human creativity/taste. Concerns exist around originality, job impacts on session musicians/assistants, training data consent, and detection of AI content. Some surveys note secrecy about use or refining of expectations after initial hype. Broader market signals include rapid growth in AI music tools, high daily AI track uploads on platforms (though they capture only a small share of streams), and integration into major workflows.
In short, AI is now a normalized part of many recording studios and production chains for technical acceleration and assistance, with adoption in the majority range among pros. It is not yet (and may never fully be) a wholesale replacement for skilled human engineers, producers, or performers in professional contexts. Usage continues to evolve quickly with new tools and features.
Generative AI use with structured inputs (lyrics + chords/structure/melody) is increasingly common among songwriters, producers, and in studio workflows, especially for demos, ideation, and rapid production. It goes well beyond simple text prompts like “make a sad country song.” Tools such as Suno and Udio support custom lyrics (with section tags), style/genre descriptions, audio uploads (e.g., guitar + vocal memos or chord progressions), and—more recently—stronger MIDI and structural controls. Adoption is real but still secondary to purely assistive AI, and often kept somewhat private in professional circles.
So where does this leave us? Ai has been involved in music production for at least a few years but generative AI like Suno and similar is what is producing the most noise (and legal issues) at the moment.
As a writer, of ancient years :), I won't diss people who have no musical theory knowledge, can't play an instrument or produce harmony and an arrangement but have a genuine emotional thing to say.
This is giving them a voice. It may not be an original voice and bits may have been nicked through the AI training but it's still their voice, albeit 'enhanced' by the work of others.
Am I pissed that original work has been used to train these AI systems - Hell yes. Do I think original artists work used to train these AI systems should be compensated? Of course!
But at the end of the day, should we stop progress? God no!
Do I approve of cretins that type mindless prompts into AI to produce hundreds of songs a month to upload to distrokid and make money? Of course not. They need to be removed.
Should the industry be more honest about existing AI use? Hell yes.
A while back there was a rumour circulating that all of my music was AI generated with no human input. That was clearly nonsensical - Some of the songs here were written in the 1970's and the majority were penned prior to generative AI being available in 2022.
Am I anti AI in music? Absolutely not. What I was angry about is that someone or some thing (probably an AI generated chat bot) had said publicly that ALL my work was AI, which it is not.
Would I use AI in the future? Well it's already used in the studio process in several different ways as indicated above. However, I think generative AI like Suno and Udio may have a place for writers who write their own lyrics, harmony and melody in order to produce Demo's and for people like me with limited singing ability. I have no problem with people without the ability to play an instrument or lacking musical knowledge using it for their own amusement but I don't believe that they should be able to flood streaming services or to make money from it.