If you have made more than one song on Floniks, you have probably uploaded the same voice sample twice. Maybe three times. Every new track meant finding that recording again, dragging it in, and waiting while the AI re-learned a voice it had already heard. Saved voices fix that: record once, name it, and reuse it for every song afterwards.
Quick answer: On AI Cover, add a voice sample as usual, then tick "Save this voice so you can reuse it next time" and give it a name. From then on, that voice appears as a one-tap option above the upload area — in both the cover and the new-song tab. No re-uploading, and for new songs the AI skips voice cloning entirely, which takes roughly 45 seconds off the wait.
Why this needed fixing
The original flow treated every generation as if it were your first. That made sense for a single experiment, and stopped making sense the moment anyone came back for a second song.
Two things went wrong in practice. The obvious one: you had to keep the voice file around and find it again each time. The less obvious one: the backend re-ran voice cloning on every single request, even when the input was byte-for-byte the same recording it had processed an hour earlier. That is time you wait for and cost we pay, both spent re-deriving something we already had.
How saved voices work
There are two paths, and they are genuinely different under the hood — worth knowing, because the benefit differs:
- Sing a new song — the AI needs a voice model (a persona) to perform with. When you pick a saved voice, that persona already exists, so the whole cloning stage is skipped. This is where the time saving comes from: roughly 45 seconds and one fewer processing stage per song.
- Cover a song — the converter works directly on the raw audio of your sample, with no persona involved. Picking a saved voice here means we reuse the stored recording instead of asking you to upload it again. The convenience is the same; the mechanism is simpler.
In both cases the original recording is kept, not just the derived model. That is deliberate: if a voice model ever needs rebuilding, we can do it from the source audio instead of asking you to record again.
Saving a voice
Add a voice sample the way you always have — upload a file or hit Record now and speak into your mic. Once the sample is in, a checkbox appears right below it:
Save this voice so you can reuse it next time
Tick it, type a name ("My voice", "Studio mic", "Lower register" — whatever helps you tell them apart), and generate. The voice is saved alongside the song you just made.
It is off by default. Saving a recording of your own voice is not something a tool should do quietly on your behalf, so it only happens when you ask for it.
Using a saved voice
Next time you open AI Cover, your saved voices sit directly above the upload area under Use a saved voice. Tap one and it is selected — the upload prompt stays available underneath if you would rather add something new.
Both tabs read from the same list. A voice saved while making a new song is immediately available for covers, and the other way round.
What it is good for
Iterating on one song. Trying six sets of lyrics with the same voice used to mean six uploads. Now it means six taps.
Keeping several voices apart. Your speaking voice and your singing voice produce noticeably different results. Save both, name them clearly, and pick per track instead of guessing which file was which.
Coming back weeks later. The recording lives in your assets, not in your downloads folder. Nothing to dig up.
Questions
Where are saved voices stored?
In your own asset library, under your account. They appear in My Stuff alongside your other assets, and only you can use them.
Can I delete a saved voice?
Yes — from your assets. Deleting it does not affect songs you already made with it.
Does saving a voice cost credits?
No. Saving is free; you are only charged for the generation itself, at the same rate as before.
Does reusing a voice make the result different from a fresh upload?
No. For new songs it reuses the same voice model the AI would have rebuilt anyway. For covers it reuses the same audio file. The output is equivalent — you just skip the waiting.
I don't see the option. Why?
The save checkbox only appears after you have added a voice sample, and only when you are signed in — a saved voice has to belong to an account. If you have no saved voices yet, the section above the upload area tells you so and explains how to save your first one.
Can I use someone else's voice?
Only with their permission. Saved voices are private to your account, and the usual rule applies: do not clone a voice you have no right to use.
What makes a good sample to save?
The same things that make a good sample generally — 10 to 30 seconds, clear, no background music, no echo. Since you will be reusing it, it is worth recording one clean take rather than settling for whatever was nearest.
Try it
Open AI Cover, add a voice, and tick the save box. If this is your first time here, the AI Cover walkthrough covers the rest of the flow.