Guide · Original Composition
Making music with AI: practical tools to try today
AI music tools you can try today: Moises for separating tracks and rehearsing clean, Suno for generating a full song from scratch, and why neither one turns you into a musician without the groundwork you already had.
AI doesn’t compose for you. It amplifies what you already know how to do, and if there’s no real musical grounding behind it, the result shows. That’s the starting point for everything that follows —and it’s exactly what I already laid out from the technical, professional side in «AI as a tool, not as the final product», on labs.tever.es. Here I go straight to the practical ground: what concrete tools you can open today, and what you’ll actually be able to do with them in an afternoon.
Separating tracks to rehearse clean: Moises
Before generating anything, there’s one AI tool every musician should know about even if generative composition doesn’t interest them at all: Moises. What it does is simple and genuinely useful: it takes any song and separates its tracks —vocals, drums, bass, everything else— so you can isolate or mute whatever you want.
For anyone who rehearses, this changes the day-to-day. If you want to play a song’s bass line, you mute the original bass and play over the rest, without “stepping on” the actual musician from the recording. If you want to sing, you mute the vocal and use the rest as a backing track. There’s a free plan with a limited number of uploads per month; unlimited use and the full feature set (chord detection, tempo adjustment, lyric transcription) require a subscription.
Generating a song from scratch: Suno
The central tool of this post is Suno, today one of the most widely used for generating complete songs —lyrics, music and sung vocals— from a text description. Xataka’s tutorial on Suno explains the basic mechanics well, and here I’ll summarize it in concrete steps:
- Go to suno.com and hit “Make a song”. You can sign in with Google, Microsoft, Discord or your own account.
- Choose simple mode if it’s your first time: write a description of the song you want (genre, theme, mood) and hit generate.
- Wait a few seconds. Suno generates the full song: lyrics, music and a synthetic voice performing it, plus cover art.
- If you want more control, use custom mode: there you write the title, the full lyrics and the exact genre yourself, instead of letting the system invent them.
The free plan gives you around 50 daily credits (roughly 10 songs), works better in English, and doesn’t allow commercial use of the result —for that you need the paid plans. If you want to go deeper than the basics, suno.hk’s Suno handbook has more extensive guides on custom mode.
Other tools worth knowing
Moises and Suno aren’t the only pieces of this ecosystem. To get a sense of the full landscape, this Xataka roundup of 14 free AI music tools is a good starting point:
| Tool | What it does |
|---|---|
| Udio | Generates songs by splitting creation into three steps: style, lyrics and feel |
| AIVA | Assisted composition, with more than 250 styles and track editing |
| LALAL.AI | Track separation, an alternative to Moises |
| ElevenLabs | AI voice synthesis and cloning — doesn’t generate full songs, but it’s the reference if what you need is a quality synthetic voice for a demo or a voiceover |
| Lyria | Google’s music generation model (text to music), built into the Gemini ecosystem |
What’s happening in the industry meanwhile
This isn’t an isolated hobbyist experiment: it’s reshaping the music business in real time. Spotify is already dealing with a real flood of AI-generated songs —to the point that in April 2026 an AI-generated artist, Ruby Black, made it to the top of Spain’s 50 most viral chart— and it’s had to respond by setting clearer limits on how that content gets labelled and paid. In parallel, on May 22, 2026, Spotify and Universal Music Group signed a licensing deal that lets Premium users generate AI remixes and covers from UMG’s catalogue, under three explicit conditions: prior consent from artists and songwriters (an opt-in model), attribution, and financial compensation for every version created.
What I learned teaching this to adults
This 2025/26 school year I worked as a substitute Digital Skills teacher for the Principality of Asturias, and among the tools I taught adults to use —some with little prior experience with computers in general— was generative AI. I wasn’t doing it as a gaita player or from the angle of traditional composition: it was general digital teaching, but the lesson I took away applies just as well to any creative trade. What surprised me most wasn’t how powerful Suno is: it was watching a student put together a full song, with her own lyrics, without me ever having explained that specific tool to her. All it took was exploring the interface for an afternoon.
That confirms something worth being clear about before touching any of these tools: anyone, in an afternoon, can put together a reasonably decent song. The technical barrier to entry has practically disappeared.
What doesn’t change
And here’s where I come back to the starting point. A tool letting you customize a song doesn’t make you a musician. There’s a lot of music theory and an enormous amount of accumulated practice that these tools don’t teach you and can’t replace —you can’t really compose without knowing them, and without the real hours of practice it takes to play an instrument.
For the industry, this is a deeper problem: millions of songs are produced every day, a pace that’s flooding the market, and barely any of them last more than a month in the conversation. As a complementary tool in the creative process of a musician who already knows what they want to say, AI adds real value —it hands you plenty of ideas fast and spares you a lot of time staring at a blank page. But that’s exactly what it is: a tool. Not an end in itself.
Bibliography
- Xataka. «Cómo crear música con IA: 14 herramientas gratis imprescindibles». xataka.com. Overview of AI music generation tools.
- Xataka. «Suno AI: qué es y cómo usarlo para crear una canción». xataka.com. Reference tutorial on how Suno works.
- Moises. Product documentation. moises.ai. Track separation features and available plans.
- wwwhatsnew.com. «Spotify y Universal Music acuerdan que los fans puedan hacer versiones y remixes con IA de sus canciones favoritas». wwwhatsnew.com. Details of the May 22, 2026 Spotify-UMG licensing deal.
Frequently asked questions
-
How does copyright affect using AI to make music?
It’s still a field under construction, with two clear fronts:
- Personal vs. commercial use: on tools like Suno, the free plan doesn’t authorize commercially exploiting what you generate (uploading it to for-profit streaming platforms, for example); that requires the paid plans.
- Rights over another artist’s style: the industry is already regulating this explicitly. On May 21, 2026, Spotify and Universal Music Group signed a licensing agreement so fans can create AI covers and remixes of UMG artists, with direct compensation to the original authors for every use — a direct response to the underlying problem: until then, anyone could generate a piece “in the style of” an artist without that artist seeing anything in return.
The legal framework keeps moving fast and varies by country and platform. The practical rule, today: if you’re going to use the result beyond private experimentation, check the specific tool’s terms before publishing or monetizing it.
-
What is AI in the context of music?
In the musical field, generative AI covers two very different fronts:
- Track separation (Moises, LALAL.AI): starts from an existing recording and isolates its components —vocals, drums, bass, the rest of the instruments— so you can rehearse over it, remix it or study it.
- Generation from scratch (Suno, Udio, AIVA): starts from a text description and produces a complete new piece, with lyrics, melody and —in Suno’s case— a synthetic voice performing it.
In both cases the tool amplifies what the musician already knows how to do: it doesn’t replace music theory, instrument practice or the judgment needed to tell whether the result is any good.
-
Why is Suno one of the most used AI music tools?
Suno has become a reference point through a simple combination: complete result, minimal entry barrier. In a single step it generates lyrics, music and a synthetic voice singing the piece, and it does it in seconds from a written description —no need to know music theory, play an instrument or use production software.
The free plan offers around 50 daily credits (roughly 10 songs), which lets you genuinely experiment before considering paying. And it has two modes: a simple one (you describe what you want and the system decides the rest) and a custom one (you control the title, the full lyrics and the exact genre). That combination —instant result, free to try, two levels of control— explains why it’s spread so widely, even among people with no prior experience with digital tools.
-
What is AI (artificial intelligence)?
AI (artificial intelligence), in the sense used by tools like Suno or ElevenLabs, is a system trained on huge amounts of data —songs, voices, text— that learns statistical patterns from that data and uses them to generate new content from an instruction (a prompt).
It doesn’t “think” or “create” independently: it recombines what it has learned based on how it’s directed. The quality of the result depends directly on whoever is steering the tool —their judgment, their prior knowledge and how well they know how to ask for what they want— not on the machine acting on its own.