OpenAI has announced the rollout of Advanced Voice Mode to a select group of paid ChatGPT users, enabling them to experience more natural, real-time conversations.
This new feature allows ChatGPT to provide instantaneous responses that can be interrupted and to detect and respond to humor and sarcasm. Unlike the current ChatGPT voice, Advanced Voice Mode doesn't need to convert speech to text and back, resulting in lower latency interactions.
The Advanced Voice Mode was first demonstrated in May, featuring an AI voice called Sky, which resembled actress Scarlett Johansson. Johansson publicly expressed her shock and displeasure as she had declined offers from OpenAI CEO Sam Altman to be the voice of ChatGPT. Following legal counsel, the Sky voice was removed, with OpenAI stating it was not intended to mimic Johansson’s voice.
Since then, OpenAI has focused on improving the safety and quality of voice interactions. The Advanced Voice Mode now includes four preset voices and blocks outputs that differ from these voices to prevent celebrity mimicry. Additionally, guardrails have been implemented to block requests for violent or copyrighted content. The early access phase will help refine the feature before a broader release.
Users selected for Advanced Voice Mode will receive email instructions, with more users being added gradually. OpenAI plans to grant all Plus subscribers access to this feature by fall.
News Source: MacRumors
