Clone your voice · use it to sing any song

If it doesn’t sound like you

Record in a quiet room and stay close to the mic.
Use one clear, steady voice for about 10–30 seconds — avoid noise, echo, and overlapping voices.
Not happy with the result? Create another voice with a clearer take or a different read-aloud script, then compare.
Hearing two singers in one song — one like you, one like someone else? When generating, pick Solo Singer or Lead Vocal under Voices.
Also add [solo singer] at the start of your style prompt so the model sticks to a single lead voice.

1. Your Voice Recording

Clear single-person voice · no background music · WAV or MP3 · aim for 30s+ (min 10s, max 2 min)

Recording Tips — for best results

✓ Speak clearly, at a natural pace
✓ Record in a quiet room with no background noise
✓ Duration: aim for 30s+ · minimum 10s · up to 2 min
✓ Stay close to the microphone, single speaker only
✗ No background music, echo, or multiple voices

You can read this aloud:

Read aloud below (~15–20s). Tap “Another sample” for another sample; keep recording to 30s+ if you can.

Hello, my name is [your name]. Today I am recording a clear sample of my natural voice. One, two, three, four, five, six, seven, eight, nine, ten. I love music because it makes mornings brighter. The quick brown fox jumps over the lazy dog. Please keep a steady pace and speak clearly.

Microphone

Multiple mics detected on some machines — enable access to pick the right input before recording.

Generating once costs 30 credits

Credits are only charged when your voice is created successfully. Failed attempts are not charged.