How to Make a Suno Album: My Complete Workflow After 18 Releases
I have released eighteen records and a hundred and forty-eight tracks in about twenty months, all of them made with AI, all of them distributed to Spotify, Apple Music and the rest. This is how I actually build a Suno album, start to finish, including the boring parts nobody talks about.
It is not a magic-prompt article. There is no prompt at the end that writes a good record for you.
Every Suno album starts with a concept, not a prompt
Every Suno album I have made that turned out well started as an idea, and every one that went nowhere started as a genre. “Synthwave” is not a concept. “A city at night, something ending, something about to start” is a concept, and it tells me what every track has to serve.
My first album came out of a very specific situation, and the theme wrote itself before I noticed it was happening — I wrote about that here. Since then I have done it deliberately. A record about reconstruction. A record about a wedding. A record about a cat. The concept decides the tracklist, and the tracklist decides the prompts.
Write prompts like you are briefing a session musician
This is the part people get wrong. They type a genre and a mood and hope. Instead, tell Suno what you would tell a human in a studio:
- Genre and sub-genre, specifically. Not “electronic” but “synthwave with analogue bass and gated reverb drums”.
- Tempo and energy, and where it changes.
- Instrumentation, including what you do not want. Excluding things works.
- The vocal — gender, register, delivery, whether it sits forward or buried.
- The emotional arc across the track. Where it lifts, where it drops out.
Write your own lyrics. This is the single biggest difference between a track that sounds generic and one that sounds like yours. The model will happily write words for you and they will be fine, and fine is the problem.
Generate a lot, keep almost none
For every track that ends up on a record I generate many that do not. That ratio is not a failure, it is the job. The generation is the fast part; choosing is the slow part, and choosing is where the album actually gets made.
What I listen for when I decide: does the vocal sit right, does the arrangement go somewhere or just loop, is there one moment I would play to somebody. If there is no moment, it does not matter how clean it sounds.
Sequencing a Suno album
A Suno album is not a playlist. Eleven good tracks in a random order is eleven tracks. In the right order it is an album, and people listen to the whole thing. I spend real time on this: what opens, what follows the opener, where the slow one goes, what you are left with.
Two practical rules that have held up across eighteen releases. The opener has to earn the second track in under thirty seconds. And never put your two most similar tracks next to each other, even if they are both good.
Export properly, or the rest was wasted
Here is the part almost nobody covers, and it is the part that gets releases rejected or shipped sounding wrong.
Export at the highest quality the platform gives you, then convert once, to FLAC at 48 kHz / 16-bit for distribution and MP3 320 for everything else. Never convert from a lossy file and call it lossless — distributors check, and listeners with good headphones notice.
Then strip the metadata and write it yourself. AI music tools stamp their own tags into the files: generator name, version, sometimes a URL back to the generation. Those tags travel with your file into every store and every library. Clear them, then write correct title, artist, album, year, track number and genre, and embed the cover art in every single file. I automated this step because doing it by hand across a hundred and forty-eight tracks is how mistakes happen.
Distribution and what to expect
I distribute through Symphonic, which puts a release on Spotify, Apple Music, YouTube Music, Tidal, Amazon Music and a long tail of smaller platforms. You need a UPC per release and an ISRC per track; your distributor issues both.
Be realistic about what a Suno album earns. Streaming pays a fraction of a cent per play, so at the beginning it pays nothing, and anyone telling you otherwise is selling a course. What it does give you is a catalogue that exists, that people can find, and that keeps earning quietly while you make the next one. Selling the files directly pays roughly a hundred times better per listener, which is why I also put the albums in the shop in lossless.
The honest part
Does it count as making music? I get asked a lot and I am not going to argue about it. The model generates the audio, so calling it a Suno album is accurate. I choose the concept, write the words, direct every track, throw away most of what comes back, and decide the order. If somebody else did all of that, it would be their record, not mine.
What I will say is that it removed the gap between having an idea and hearing it. That gap used to be years of learning an instrument I never learned. Now it is an evening. Whether that is cheating or just a new instrument is a question I stopped losing sleep over somewhere around release six.
If you want to hear where all this ended up, everything is in the discography — eighteen releases across synthwave, deep house, rock, meditation and Latin. Start with Synthwave Dreams, because it was first and because it is still the one I am most attached to.