Recent advancements in artificial intelligence have led to significant developments in voice technology. Kyutai Labs has launched MoshiVis, an open-source real-time speech model capable of integrating visual inputs, which adds 206 million parameters via lightweight cross-attention modules. This model builds on their previous work with Moshi, enhancing dialogue capabilities with visual interaction. Concurrently, OpenAI has introduced its new GPT-4o voice models, which allow applications to communicate in real-time with human-like emotions and responses. These models, including 'gpt-4o-mini-tts' and 'gpt-4o-transcribe', enhance speech synthesis and transcription capabilities, marking a move towards more interactive AI systems. However, the rise of voice cloning technology has raised concerns over potential scams, underscoring the need for safety measures in the industry. The only safety measure noted by Sesame's CSM-1B, which also offers hyperrealistic voice mimicry, is a warning against scamming. The rapid evolution of these technologies highlights both the opportunities and ethical challenges that come with advanced AI voice capabilities.