The voice encoder could not load on this device (not enough memory). Voices that were already set up keep working; new recordings are saved but cloning them may be unavailable here.
No voices yet. Record 10 seconds of your voice and use it for any text.
New voice
How do you want to add a voice?
Record with a script
Read a short text aloud — the transcript fills itself in. Takes about 30 seconds.
Recommended
Upload audio
Use an existing clip of the voice. You type the transcript yourself.
Pick a script to read. It becomes the transcript automatically.
0.0sReady when you are — aim for 5-12 seconds
Tap to record — starts after a 3-2-1 countdown
Drop an audio file here
or click to browse — MP3, WAV, M4A, OGG · 5-12 seconds of one person speaking works best
0.0s
Clip trimmed to the first 15 seconds — longer audio does not improve cloning.
Analyzing voice… (one-time setup)
✓
Voice saved
Generation Library
0 generations
No generations yet. Create something in the Studio!
Settings
Storage Usage
Cached modelscalculating...
Generation historycalculating...
Saved voicescalculating...
Totalcalculating...
All data is stored locally in your browser using IndexedDB. Nothing is sent to any server. Clearing cached models means ~3 GB will need to be re-downloaded on next use.