Record 15–25 seconds of your voice in English, upload it to the homelab, and test the clone right away.
Speak naturally, no music or other voices, and keep 2–3 cm from the microphone. Shorter than 30 s — the server rejects longer samples.
The first time, the browser will ask for microphone permission. Recording stops automatically at 25 seconds.
Default is http://localhost:8004. Change it only if you open this page from another machine.