“A song written and sung in a few minutes, for the price of a coffee.” Put like that, it sounds either too good or a bit lukewarm: you picture a one-size-fits-all jingle with a first name dropped into it, the kind of gift that earns a polite smile for three seconds.
What actually happens is more interesting. Here, step by step, is everything that sits between the moment you type your memories into a form and the moment someone hears themselves being sung about: what the machine does very well, what it cannot do, what it costs and how long it takes.
What AI changed, and what it did not
Commissioning a bespoke song used to mean finding a songwriter, briefing them, waiting for the demo and then the recording. Budget €150 to €600 and one to three weeks. What AI removed is the cost and the wait. Not the nature of the gift.
What it did not change is where the emotion comes from. A song lands because it names true things: the nickname nobody uses in public, the time she missed the train to Lisbon, the sentence your father has been repeating for twenty years. No engine invents those: you supply them. The machine handles the rest: the rhyme, the melody, the voice, the mix.
Step by step, what actually happens
1. You tell the story, ten minutes flat
A form, not an agency brief. The first name, the occasion, how you know each other, the musical style (pop, rock, rap, piano ballad) and above all three or four concrete details. Write the way you would talk to a friend who has never met her: “she sings out of tune and owns it” beats “she is cheerful” ten times over.
2. The engine writes the lyrics, then composes
Your text first becomes structured lyrics: verses, chorus, bridge. Your details are spread out so the chorus carries the heart of it and the verses do the storytelling. The music comes second: our guided music AI lays down the melody, the instruments and the sung voice in the style you asked for.
3. You listen to two versions and pick one
This is the step nobody expects: the song does not arrive alone. You get two interpretations of the same lyrics, different melodies, different voices. You keep the one that sounds like the person. On the Platinum plan you can even ask for two different styles: a guitar-and-voice version for the living room, and one that holds up at a party.
4. Karaoke, cover art, MP3: what you actually get
Not just an audio file. Depending on the plan: a page you send as a single link, the MP3 to download and keep, a full-screen karaoke where the lyrics scroll in time with the voice, illustrated cover art. The honest way to judge is to listen: a full demo song is online, karaoke included, with no account and no card.
AI does not know your story. It is waiting for you to tell it.
What AI does well, and what it does not
What it does remarkably well
Rhyming without forcing it, holding a structure, singing in tune, delivering a clean mix first time. It is also unbeatable on awkward constraints: fitting “physiotherapist” into a chorus, switching style because rock would suit her better after all.
What it does not do
It guesses nothing. Give it “great mum, very kind” and you will get a correct, interchangeable song: technically flawless, emotionally flat. And its voice is beautiful but belongs to nobody in particular; if you want a recognisable timbre, that is a different craft, compared in detail in AI or musician, which one to choose.
The real work, the part that makes the difference, fits into three questions: what does this person do that nobody else does? Which moment always comes back to you when you think of them? What have you never said to their face? Three honest answers beat a page of adjectives.
On the day, the staging matters as much as the song: play it at dessert, not in the middle of the present pile. Bring tissues. The karaoke earns its keep on the second play: the first time she cries, the second time the whole table sings the chorus.
The mistakes that spoil the result
- Staying vague. “She is wonderful” gives you nothing; “she shouts hello on the phone as if the line were bad” gives you a chorus.
- Trying to say everything. A song carries four to six ideas, not fifteen years of biography: whatever overflows weakens the rest.
- Choosing the style you like rather than hers. The gift is her, even if rap does nothing for you.
- Discovering the song at the same time as everyone else. Listen to it alone once beforehand: you will know the exact second to turn the volume up.
Price and turnaround, no dodging
Three plans, no subscription: The Single at €5 (about 2 min 30, two versions, MP3), the Gold Record at €8 (about 3 min 30, plus full-screen karaoke and illustrated cover art, the most popular one) and the Platinum at €13 (about 5 min, two styles, photo slideshow, shareable video). Delivered in minutes, at any hour.
For comparison: an online songwriter starts around €100 to €150 with a one to two week wait, and a local studio goes past €400 quickly. The song is not cheaper because it is worth less, but because the expensive part, studio time, is gone. The full ranges are broken down in how much a custom song costs, and the criteria for sorting offers in how to compare services.
Can you tell it was made by AI?
In the lyrics, no: they are about your story, which no generic text can do. On the voice, most people cannot tell the difference, especially on the first listen to a gift.
How long does it really take?
A few minutes between confirming and the finished song, two versions included. The only real delay is you: set aside a quiet quarter of an hour to tell the story properly. An order placed at 11 pm is playing at ten past.
Do I need to be good at writing?
No: you are not writing the lyrics, you are telling a story. Jot your memories down in any order; rhyme and structure are not your job.
What if I like neither of the two versions?
It happens, usually because the original story was too short. Add two precise memories, change the style and run it again: at this price, one more attempt costs less than a bouquet.
Key takeaway
An AI song is your story set to music in a few minutes, for €5 to €13, with two versions to listen to before you choose. The technology handles the rhyme, the melody, the voice and the mix; you handle the only part that matters, the memories. Listen to a demo first, then start yours.