Melodica AI
A song written for one person only
Tell us who it is for and what makes them them. Melodica AI writes the lyrics, sings them in your language and draws the cover — in minutes, inside Telegram.
Create a songHear nine of them
Each one made from a two-line description, in a different language. Nothing here — words, voice or artwork — was made by a human.
Strong models, gentle price
The words and the music come from some of the strongest models that exist today. We keep the price as low as we can, so that a song stays a gift rather than a purchase.
✍️ Lyrics — Claude Opus 5
Anthropic's flagship model, running at its highest reasoning setting. It doesn't drop a name into a template — it writes about one particular person, from the details you gave.
🎧 Music and voice — Suno v5.5
The newest generation of the engine: a real singing voice in twelve languages, not a track stitched together from ready-made pieces.
💛 What it costs
The first song is free. If it came out wrong, one remake is free too. After that it stays deliberately cheap — we would rather you sent a second song than hesitated over the first.
Where it stands today
Real figures, not a promise. This is a young service and these are the honest numbers behind it.
songs made so far
languages it sings in
minutes from idea to song
Made a song already? Tell us how it went using the form below — the good and the bad both help.
How it works
Two ways: say it in one sentence, or choose every detail yourself.
Describe them
Who the song is for, the occasion, and the small details only you know.
Approve the words
Read the lyrics, edit any line, or bring your own text instead.
Hear a preview
Thirty seconds with the cover art, before you commit to the full song.
Send the gift
Full track and cover art, ready to send.
As much control as you want
Quick mode decides everything for you. Advanced mode hands it all over.
🌍 Twelve languages
The song's language is separate from yours — order a Spanish song from a Russian chat.
🎻 Instruments
Piano, violin, guitars, saxophone, strings — pick any combination, or take a ready-made style.
🎤 The voice
Male, female or a duet; soft, deep, young or powerful — and an accent for Spanish and English.
⏱ Length
Set by the words rather than a setting — the song runs as long as what was written for it, usually around three minutes.
🧩 Shape
A bridge or none, an instrumental break, a short intro, and how the last line ends.
🔁 Another like it
Loved the sound? Keep the music and the voice, write about someone else.
Any occasion
Questions
How long does it take?
A couple of minutes. The lyrics arrive in seconds; the music takes the longest, and you get a preview before the full track.
Is the song really unique?
Yes. Both the words and the music are generated for your description alone — nothing is picked from a library, and no two songs come out the same.
Can I use my own lyrics?
Yes. There is a mode for exactly that: send your text and it is sung word for word, with nothing rewritten.
What if I don't like it?
Regenerate the lyrics as many times as you like before the music is made, and edit any line by hand.
Can I share it?
The song is yours. Send it, post it, keep it — there are no limits on what you do with it.
Where to find it
Today Melodica lives inside Telegram — nothing to install. The phone apps are being built.
We’ll announce both launches in the bot. Until then, everything already works there.
Say something to us
A question, a problem, a song that came out wrong — it reaches a person, not a queue.
Someone is about to get a song about themselves
It takes one message to start.
Open in Telegram