ChatGPT’s Advanced Voice Mode enables it to perceive and react to comedy, sarcasm, and other elements while providing real-time, interruptable responses. Lower latency interactions result from the new model’s lack of the requirement to translate speech to text and back again, which is what the existing ChatGPT voice model performs.
In May, OpenAI debuted Advanced Voice Mode, showcasing an artificial intelligence voice named Sky that bore a striking resemblance to Scarlett Johansson. Without Johansson’s consent, the voice was produced and utilized, and she eventually released a statement about the incident. Sam Altman, the CEO of OpenAI, reportedly made several proposals to Johansson, but she declined them all because she didn’t want to represent ChatGPT. She described herself as “shocked, angered, and in disbelief” that Altman had produced a voice that “eerily similar” to her own. Although OpenAI stated that Sky’s voice was not meant to sound like Johansson’s, it was taken down when she retained legal counsel.
OpenAI claims to have been trying to enhance the security and caliber of voice communications ever since it demonstrated Advanced Voice Mode. In order to avoid copying the voices of celebrities, Advanced Voice Mode speaks in four preset voices and is designed to prohibit outputs that deviate from those voices. In addition, OpenAI has “implemented guardrails” to prevent requests for violent or copyrighted content. The service will be improved in early testing and then made available to a wider audience.
Users will get instructions via email after being approved for Advanced Voice Mode access; OpenAI intends to add users on a rolling basis. Advanced Voice Mode will be available to all Plus users in the fall.