One of the most common notes I get with a mastering order is some version of "the vocal has static, can you clean it up?" Sometimes it's a fizz on the S sounds. Sometimes it's a crackle that rides along with sustained notes, or a gritty shimmer on the top of the voice that gets worse in the chorus.
The short answer: I can make it less noticeable. I can't make it go away. Here's why, and what you can do before you send the file.
Where the static comes from
In AI music, vocal noise is usually a generation artifact, not a recording problem. The generator builds the voice as part of one piece of audio, and when it struggles, the struggle shows up as texture: fizz, ringing, a grainy edge. My note on ring or fuzz goes through how to tell the different kinds apart by ear.
Because the artifact was generated together with the voice, it isn't sitting on top of the vocal like tape hiss. It's woven into it, in the same frequency range as the voice itself, usually somewhere between about 4 and 10 kHz where consonants and breath live.
Why mastering can't pull it out
Mastering works on the whole stereo mix. Any tool I use on the vocal's frequency range also hits the hi-hats, the guitars, the reverb and everything else living there.
Noise-reduction plugins work by learning what the noise sounds like and subtracting it. On a full mix with a generation artifact, there's no clean noise to learn from. Push denoising hard and you get a smeared, underwater sound on the whole song. That's usually worse than the fizz.
What I can do is targeted and gentle: a de-esser or a dynamic EQ that dips the harsh band only when it flares up, so the worst moments sit back. On a mild case, that's enough that most listeners won't notice. On a heavy case, it helps, and the static is still there.
Send the rawest version you have
The biggest thing you control is which file reaches me. Every step between the generator and the mastering order can make vocal noise harder to deal with.
Send the straight download, before you ran it through any "enhance," "clean up," denoise or AI mastering step. Those tools often trade the static for a different artifact, and I can't undo theirs either. Send a WAV rather than an MP3, because lossy encoding adds its own swirl on exactly the frequencies where the fizz lives. And if you separated the song into stems and put it back together to fix the vocal, send the original too, because that round trip adds artifacts of its own.
If you have several takes and one has less static but a slightly weaker performance, send both and tell me. That's a choice worth making with your ears, not mine.
When the fix is upstream
If the static is on every line and it bothers you, the real fix is a new generation of the vocal, or a different take of the section where it's worst. Repairing the vocal on its own, as a separate track, is possible, but that's mixing work on the stem, not mastering.
The fizz you hear on your laptop is the fizz you'll hear after mastering, just a little quieter. Decide now whether you can live with it.
Josh, Founder, JG BeatsLab
Want this handled on your track? I master AI-generated music by hand, one song at a time, and every master comes with a Mastering Dossier that shows what I did and why. Book mastering or take the 3-minute scorecard to see where your workflow stands.
