OpenAI has introduced advanced audio models named 'gpt-4o-mini-tts' and 'gpt-4o-transcribe', enhancing real-time speech synthesis and transcription capabilities. The 'gpt-4o-transcribe' model supports over 100 languages and boasts an English word error rate of just 2.46%. Meanwhile, the 'gpt-4o-mini-tts' model enables customization of emotional expressions, marking a notable advancement in voice AI technology. Additionally, OpenAI has updated its ChatGPT voice mode, which now interrupts less frequently, allowing for natural pauses. This update is available to all users, while paid users will experience an improved model personality that is described as engaging, direct, and concise. The updates reflect OpenAI's commitment to enhancing user interaction through more natural and effective voice communication.