Updated
MiniMax Music: covers of your own track, and vocals in any language
MiniMax Music is kept here for one thing nothing else can do: it covers a recording you already have — it takes your track, transcribes the words and re-sings them in another genre. It also sings in the language of whatever lyric you write, including the ones Lyria does not cover.
Below is the model itself. Describe the song you need: the answer arrives on this page, and the first attempt needs no sign-up.
What you are trying here
This is not a demo mode: the request goes to the same model the main chat uses. Music generation is slower than a text answer, which is normal; no need to close the page.
Covers: the words first, the money second
A cover here is not one button but two steps, and that is for your benefit.
- Analysis. You supply the original recording and the model transcribes the words from it and tags them by section. Generation has not started yet, and the result of the analysis lives for a day.
- The cover. You correct the transcribed text — where the transcription misheard, where you want different words — and only then does the model re-sing it in the genre you asked for.
Read the transcript before generating
This is the one place in the chapter where real money is saved. Transcription does not catch everything: one misheard line and the cover is sung wrong, and fixing it costs a full new generation. A cover costs exactly the same as a song from scratch.
The analysis step itself does not cost a generation, and its result is valid for twenty-four hours — so you can proofread the words calmly and not in one sitting.
Vocals in the language you write
The singing language follows the language of the lyrics: write the words in English and you get an English song, in Russian and you get a Russian one. Lyria 3.5 does the same for about half the price; MiniMax is the pick when you want its voice and delivery, or a second take of the same song.
You can supply your own words or ask for them to be written. A separate tool composes lyrics with a title and section tags — and it does not spend a generation, so the text can be proofread and corrected while the track is still unpaid for.
About pronunciation — honestly
Pronunciation quality is inconsistent. Sometimes stress lands wrong, sometimes an accent shows, and there is no predicting in advance which line it will happen on. There is no voice picker on this model either: you cannot ask for “a different vocalist”, only regenerate and get a new one.
The practical consequence: short, phonetically simple lines come out more reliably than long, complicated ones. If the lyric contains a proper name that has to be pronounced cleanly, check that name in the very first take.
Tagging the lyrics by section
The words are sung verbatim, so structure is set with tags on their own lines: [Intro], [Verse], [Chorus], [Bridge], [Outro].
Stage directions inside the lyrics will be sung
Anything the model does not recognise as a section tag is treated as words of the song and sung. A note like “quiet and dreamy here”, left in the middle of the lyrics, will end up in the vocal verbatim. Put arrangement notes in the track description, not in the words.
MiniMax or Lyria — the short fork
| Job | Which model |
|---|---|
| A cover of your own track | MiniMax |
| An instrumental, a bed for video | Lyria |
| Trying ideas cheaply | Lyria Clip |
| A sung song, in any language | Either — Lyria 3.5 costs about half, MiniMax brings a different voice |
What MiniMax Music will not do
- It will not let you choose a voice. Not by name and not from a sample — the vocalist is whoever turns up.
- It will not edit a finished track. Any change is a new generation at full price.
- It will not guarantee pronunciation of long, phonetically awkward lines.
- It will not be cheap. For trying options, Lyria Clip costs far less; bring MiniMax in once the lyric and genre are settled.
Frequently asked questions
How do I make a cover of my own song?
Can I choose the vocalist’s voice?
Why is the stress or the accent wrong in the song?
Can I give the model my own lyrics?
Which is cheaper, MiniMax or Lyria?
Do I need to select MiniMax in the chat?
Examples and prices are on the music generation page. What we connected and when is in the release post. Other models are in the book’s table of contents; music in general is in the music generation book.
Try it yourself
Everything in this guide runs inside Twelver
One chat for text, images, video, music and voice — no separate services or subscriptions.
Open Twelver chat