About Suno

I’ve spent enough time using this technology seriously to know what I would consider important before paying someone to teach it. That doesn’t make me some certified authority on AI music. I’m not teaching Suno, though I probably could, and I’m not trying to discourage anyone from signing up for the sudden wave of SkillsFuture AI music courses either.

Suno is giving people an opportunity that, for most of our lives, simply did not exist. That earworm inside your head, that emotion looking for an outlet, can finally become something tangible that you can hear and share with other people. For most people, music-making has always felt divided between the listener and the musician: those who consumed music, and those blessed with the ability, training or tools to create it.

Not anymore.

Credentials, accreditation and industry history are impressive certifications and records. But they do not guarantee music that will connect. Likewise, someone rich, attractive, famous, well-connected and impressive on paper does not guarantee they will be a good partner. I think music works the same way.

A music-production résumé tells me someone understands music. A teaching résumé tells me they understand education. If I were specifically paying to learn Suno, I would still want to hear what they had made with Suno specifically. That isn’t anti-credential. It is simply asking the evidence to match the particular thing I am paying to learn.

I was reminded of this reading a recent MustShareNews report on reactions to the 2026 NDP songs. What struck me was that the criticism being discussed was about the finished listening experience: lyrics, melodies, autotune, whether some of the songs felt right for National Day, and whether listeners connected with them. I haven’t actually spent enough time with those songs to make that judgment myself. What interested me was that the conversation still came back to the songs themselves, rather than the résumés behind them.

Music has an unusually ruthless final auditor: the listener. If I were a Suno student paying for training, I would want to hear what the trainer had actually published on Suno. I think of it a little like choosing an interior designer. Qualifications and software skills would matter to me, but I would still want to see the portfolio. The eyes seek the same assurance that the ears do in music creation.

Everyone can generate something immediately because prompts are concrete. Useful? Absolutely. But there is an old distinction between giving someone a fish and teaching them how to fish. In Suno terms, prompts are the feed that gets you the fish. Judgment and discernment are the actual fishing. If I were paying to learn, I would want to come away better able to tell AI slop from something worth developing: where and how to find resources for lyrics and instrumentation, how to count syllables and choose lexicons, recognise hooks worth keeping, test song structure and density, design vocals and shape prosody.

I caught myself wondering: if I were the trainer, would I really be willing to share all the little hacks and workarounds accumulated through hundreds or thousands of runs? Probably not every one of them. And perhaps a good course doesn’t need to. What I would want is enough of the underlying judgment that eventually I no longer need the trainer’s recipe every time I want to make something.

The same caution applies to follower counts and engagement. It is nice to get Likes and have people follow you. After all, we do want people to listen to what we have created. But those numbers are also affected by discoverability, consistency, promotion, platform behaviour and the audience someone has already accumulated. They are signals. They are not verdicts.

Recently I discovered another kind of signal on Suno that I find more interesting: somebody choosing to cover or remix your song. That still does not certify that a song is good. But another creator has done more than tap Like. They heard something in the work and decided it was worth spending their own time and credits rebuilding, extending or reinterpreting. Maybe it was the melody. Maybe the lyrics. Maybe they simply heard somewhere else the song could go.

I have done it from the other side too. I once went through more than 60 generations trying to make new words sit properly inside somebody else’s melody. So when another creator spends their own credits working with something you made, I don’t read that quite the same way as another Like. It is voluntary creative response. Not certification, and not proof of mastery. Just another piece of evidence.

Which leaves me with a hierarchy I find increasingly useful when looking at people teaching creative AI:

1. 01Credentials → context
2. 02Follower count → reach
3. 03Engagement → response
4. 04Portfolio → capability
5. 05Repeated strong work → consistency
6. 06Covers and remixes → voluntary creative response
7. 07Your own ears → final judgment

That last one is the part I keep coming back to. My dad was playing me a slow Chinese AI jazz track recently, and almost immediately I could hear the clash. The instrumentation was definitely jazz. On the surface, it sounded convincing enough. The right kinds of instruments were there, so you could easily say, yes, this is jazz.

But the singing wasn’t behaving like jazz at all. It was rushing through the song. There was no restraint, no pain, no sense of hanging on a word. No vocal improvisation, no spaciousness, no relaxed, lagging-behind-the-beat delivery. The instruments were carrying the genre, but the voice wasn’t inhabiting it. The emotional resonance never really arrived.

That is the kind of distinction I mean when I talk about judgment. Being able to generate a genre is not the same as understanding what makes it convincing. And that is the kind of thing I would want a course to help me hear for myself.