Welcome to kits.ai! This guide will teach you everything you need to know about voice transformation.

If you’re having support issues please contact us here, if you have a quick question try our FAQS.

Quick Tips

Our AI voice conversion technology is powerful but there are a few quick tips you can do to ensure you get the best results possible.

Keep your input audio clean and dry (no reverb, delay, chorus, excessive compression).

✅ Good Conversion ✅

Before

no reverb, no stacking, mono, no noise

when_OG.wav

After

clean, high quality output

when2.wav

No harmonies, layers, doubletracking, stereo effects.

❌ Bad Conversion ❌

Before

reverb, stereo, harmonies

waydownstereo2.wav

After

cursed output

waydown_bad.wav

Keep your input audio free from background noise, instrumentals, or any non-vocal audio.

❌ Bad Conversion ❌

Before

background noise

waydownBGNoise.wav

After

weak, raspy output

waydownBGNoise_2.wav

Match the style of the original singer

Every artist model is trained on a specific singing style. For a realistic conversion, try to match the style of the original artist.

✅ Style Match ✅

Before

pop female vox

shake_it_off.wav

After

classic Sara Phillips output

shake_it_off_sphil_v1_epoch_1850_0.wav

❌ Style Mismatch ❌

Before

low-pitch male rapping

bad_and_boujee.wav

After

Sara Phillips model attempting to recreate

bad_and_boujee_sphil_v1_epoch_1850_0.wav