Hi, does anyone know why when the model generates a bold text, it's not sent to the voice input to be said ?