Comment by corysama

Can anyone confirm/deny my theory that AI audio generation has been lagging behind progress in image generation because it’s way easier to get a billion labeled images than a billion labeled audio clips?