Voice cloning models reproduce the whole recording, not just the voice. A fan, a fridge hum, room echo or music in your sample will appear in every sentence your clone speaks. Five minutes spent on a clean recording pays off forever.
1. Choose the right room
- Small rooms with soft furnishings (bedrooms, walk-in closets) are best.
- Avoid kitchens, bathrooms and empty rooms: hard surfaces cause echo.
- Switch off fans, air conditioning and notifications.
2. Position the microphone
- Phone or headset: about 15–20 cm (a hand-width) from your mouth.
- USB microphone: 10–15 cm away, slightly off-axis so your breath doesn't hit it.
- Keep the same distance for the whole recording.
3. Read 10–20 seconds naturally
Read a short paragraph at your normal pace, in the voice you want the clone to use. The Studio suggests a paragraph you can use. Avoid whispering, shouting or long pauses. Don't start with "um" or a cough: the Studio trims silence, but not noises.
4. Check before you clone
Play the recording back with headphones. Listen for hiss, hum, echo or distortion. If you hear any, fix the cause and record again. It only takes a minute.
Hear the difference
Both clones below were made from the same voice. The first reference was clean; the second had background hum and hiss added. Notice how the noise carries over into the cloned speech.
Quick checklist
- Quiet, soft room
- Steady microphone distance
- 10–20 seconds of continuous, natural speech
- One speaker, no music or effects
- Checked with headphones