OpenAI Introduces ChatGPT Voice for Work and Codex on Desktop

OpenAI is enhancing the functionality of ChatGPT Voice by integrating it into its Work and Codex features within the desktop application for macOS and Windows users. This new capability allows eligible users to engage with ChatGPT using voice commands to initiate tasks, monitor progress, and manage collaboration across various projects and agents. The integration aims to streamline workflows by enabling users to communicate with ChatGPT in a more natural and efficient manner, moving beyond simple text-based interactions.

The voice feature is accessible through the ChatGPT desktop application, with additional support for paired iPhone access, allowing users to oversee and direct their ongoing projects remotely. However, it is important to note that the standalone Voice functionality in Work and Codex is not yet available through web or mobile applications. This new development is part of OpenAI's ongoing efforts to enhance user experience and productivity by leveraging voice technology in professional settings. Users can utilize the voice feature to not only dictate prompts but also to engage in more complex interactions, such as initiating tasks, checking on the status of agents, or providing new instructions while work is in progress.

The underlying technology driving this enhanced voice experience is OpenAI's GPT-Live, which utilizes a full-duplex architecture. This allows ChatGPT to listen and respond simultaneously, facilitating a more dynamic conversational flow. The model can acknowledge users with brief affirmations and maintain the dialogue while processing more intricate queries in the background. Visual responses, such as weather updates or sports scores, can be displayed alongside spoken replies, enriching the interaction. Additionally, GPT-Live is capable of utilizing web searches and memory functions where applicable, further enhancing its utility. On compatible macOS systems, the AppShot feature can capture the current application window and provide context to Codex, allowing it to better understand the user's environment. However, it is distinct from continuous screen sharing, as live video capabilities are not initially supported. Access to the voice features in Work and Codex is available to users with eligible paid accounts, though it may vary based on the user's subscription plan, geographic location, application version, and workspace configurations. Administrators for enterprise and educational accounts may need to activate advanced voice functionalities during the rollout phase, ensuring that users can fully leverage the new capabilities offered by OpenAI.