I am not impressed. We already have better models for text to peach and voice synthetization. What we see here is integration with a LLM. One can do it at home

I am not impressed. We already have better models for text to peach and voice synthetization. What we see here is integration with a LLM. One can do it at home by combining Llama3 with text to speech and voice synth. What would amaze me would be for GPT 4 to have better reasoning capabilities and less hallucinations.