“Users can create a Personal Voice by reading along with a randomized set of text prompts to record 15 minutes of audio on iPhone or iPad.”
Fifteen minutes of reading gets you a synthetic copy of your own voice, pitched at people with ALS who are losing theirs. Apple stresses that the machine learning runs on the device. It still means every new iPhone ships with a voice cloner that needs a quarter of an hour of audio. Apple launched it under the accessibility banner, where nobody asks hard questions.