Clone any voice with a 5-second reference recording and generate natural speech locally using F5-TTS.