From a practical standpoint, both models behave similarly to other text-to-speech models ... and output it as an audio file, from the provided sample recordings and prompt. Using a 24-second ...