Harsh S sounds (sibilance) live in the 5-8kHz range, and Suno sometimes renders them sharp — especially on bright female vocals. Two fixes: prompt-side and edit-side.
1. Prompt-side: soften the vocal description. warm smooth vocal, soft consonants, gentle beats bright crisp powerful vocal for sibilance. 'Crisp' and 'bright' are literally sibilance instructions.
2. The guaranteed fix: de-essing. This is a 2-minute edit:
- Get Stems, isolate the vocal stem
- Any free de-esser (TDR Nova free, or Audacity's EQ): dip 2-4dB around 6kHz only when S sounds hit
- Or manually: find the harsh S in the waveform, select just that syllable, drop it 3dB
3. Rewrite the worst offenders. If one word ('sweetest') stings every time, swap it for a synonym without S sounds. Lyric surgery beats processing.
4. Tilt the whole vocal darker: in your DAW, a gentle high-shelf cut above 8kHz on the vocal stem smooths everything without killing clarity.
Sibilance is the most fixable vocal problem there is — don't regen over it, de-ess it.
Good luck! Hank founder of SonorixPG