The lead voice
The original vocal is separated and replaced with the voice you pick. On this route that is a My Voice model trained on recordings of your own singing.
Your songs, your voice
The AI cover generator in MusicCreator puts a different voice on a song while the melody and backing stay, and this route uses the voice that belongs to you. Train a My Voice model on your own singing, pick a song you generated from My Music, and listen to that track come back with you in the lead.
It is a way to cover my own AI song without a studio session: everything runs in the browser, and processing needs a login.

A song generated in MusicCreator arrives with an AI vocal. The AI cover generator works on that finished track, so it helps to know which parts move and which stay before you run your first cover.
The original vocal is separated and replaced with the voice you pick. On this route that is a My Voice model trained on recordings of your own singing.
MusicCreator describes its covers as keeping the song's melody and instrumental intact, so the arrangement you liked when you generated the song carries over.
Song to Cover takes a track From My Music, listed as Select from workspace, or a file through Upload Music in MP3, WAV, OGG, M4A, AAC, FLAC or WMA.
Main vocals volume, Instrumental volume and Reverb level set the starting mix, and Advanced holds Song pitch, Breathiness preservation and Voice similarity.
Every setting on the cover panel has a range and a starting value. Writing down what you changed from these values makes it easy to return to a take you liked, or to repeat the same balance on the next song from My Music.
How each setting sounds on a given song is something to judge by ear, one change at a time, with the first take left at these starting values.
Three stages take a song you generated to an AI cover in my own voice, the result this route is built around. Training comes first; once your voice model exists, it can cover any song, so each new song only needs the last two steps.

A cover in your own voice depends on two things you control: the recordings behind your voice model and the song you hand it. Treat each cover of your own AI song as a small experiment, with one song, one change and a full listen before the next run. These short habits keep that routine steady.
Train on singing, not speechStart from a generated song you already likeKeep the first take at the starting valuesChange one setting per takeNote the values that workedCover songs you have the rights to
Most people on this route already make songs in MusicCreator and want the lead vocal to be theirs. A cover tool that runs in the browser fits that habit, since the songs and the voice model live in the same account.
You generate songs in MusicCreator and want listeners to hear your own voice on them, with the arrangement you already chose kept in place.
Generated drafts give you a melody to work with, and a cover in your own voice lets you hear how it sits for you before the next draft.
Songs you made for fun turn into a playlist sung in your own voice, one cover at a time.
Answers about songs, voice training, settings, credits and rights for covers in your own voice.
Songs, voice and settings
Song to Cover takes a track From My Music, which is where songs you generated in MusicCreator wait, or a file through Upload Music in MP3, WAV, OGG, M4A, AAC, FLAC or WMA.
This route centres on songs you made. For an uploaded file, use a recording you own or have permission to cover; a commercial track by another artist stays theirs after a cover.
Training takes 10–30 minutes of clean vocal-only audio, spread across up to 5 files. Training Audio comes from Upload Music or Record, and training also takes a Voice Name and Gender Of this Voice. Record captures your singing in the browser.
Training isolates the vocals from your recordings. Bring singing to the training, since the model is meant to sing songs back to you.
Advanced holds three settings: Song pitch runs from −6 to +6 and starts at 0, Breathiness preservation runs from 0 to 100% and starts at 72%, and Voice similarity runs from 0 to 100% and starts at 50%.
Leave them at their starting values for a first take. When you change one, change only that one, so the next listen tells you what it did to your song.
Credits and rights
The songs you cover start as generations in MusicCreator. The free plan includes 30 credits a year, which is 15 generations. Each song uses 2 credits, and Number of Songs to Generate defaults to 2.
Unused credits do not roll over; they reset on your renewal date. Your credits, subscription status and next refresh are listed in your profile, and processing a cover needs a login.
Monthly and yearly plans include commercial rights within the scope of the Commercial License. Songs made before you upgrade can get the license later with credits, within the scope of the Commercial License.
Within the scope of the Commercial License, the certificate costs 5 credits per song on monthly plans and with Pay Once, and 0 credits on yearly plans. Cover songs you made or hold the rights to, and keep the voice on them your own.
Train My Voice once, pick a song you generated from My Music, and listen to it with your own voice in the lead.
Processing a cover needs a login.