“Google's AI models are natively multimodal, which will enable zero-friction conversion of content between formats like text and audio.”