OpenAI’s Advanced Voice Mode: focus on new functions and security

OpenAI continues to drive the development of Artificial Intelligence and has announced that it will make the Advanced Voice Mode of its ChatGPT widely available. This extension promises to significantly improve the interactivity and naturalness of voice conversations.


The new features of the Advanced Voice Mode include the ability to interrupt AI responses mid-sentence, recognizing and responding to emotions based on the user’s tone of voice, and a personalized voice mode that can retain specific information about the user. This mode also offers improved speech capabilities in multiple languages and introduces five new voices called Arbor, Maple, Sol, Spruce and Vale, which were created with the help of professional voice actors.

This extension is initially available for ChatGPT Plus users ($20 per month) and Team users ($30 per month with a higher message limit). A phased rollout for enterprise customers and educational institutions is planned starting next week, with all Plus users gaining access by the end of the fall. However, the feature remains unavailable for users in the EU, UK and some other countries due to geographical restrictions.

Particular attention is being paid to the comprehensive security measures: OpenAI has had the Advanced Voice Mode tested by external experts in 45 different languages and from 29 geographical regions. The GPT-4o system contains specific mechanisms to prevent problematic content such as violent or erotic language, as well as to ensure the lawful use of voice recordings and protection against copyright infringements.

This move demonstrates OpenAI’s commitment to user safety while maintaining a high level of innovation, putting great pressure on the industry to keep pace. The ability to now respond to emotion-based interactions and communicate in over 50 languages could significantly expand the scope of voice and voice assistance technologies and open up new market potential.

Summary

  • Sophisticated functions: Voice interruptions, Emotion-based interactions, Personalized voice modes.
  • Availability: For ChatGPT Plus and Team users, phased rollout for enterprise customers and educational institutions, geographic restrictions in EU and other countries.
  • Security measures: Review by external experts, mechanisms to avoid problematic content.
  • Multilingual: Support for over 50 languages.
  • User experience: Fast and realistic interactions, high usability.

Sources: OpenAI @ X