zlacker

Voice cloning could be as simple as a few seconds of audio in the context window since GPT-4o is a speech to speech transformer. They wouldn't need to claim anything, just switch samples. They haven't launched the new voice mode yet, just demos.