OpenAI finally launches human-like Advanced Voice Mode for ChatGPT
OpenAI has announced the launch of its Advanced Voice Mode (AVM) for a broad range of ChatGPT's paid users. The audio feature, designed to make conversations with ChatGPT more human-like, is initially available to Plus and Teams tier customers. Enterprise users are set to gain access next week. The AVM feature is now represented by a blue animated sphere, replacing the black dots used in OpenAI's technology demonstration in May.
AVM adds 5 new voices to ChatGPT
The AVM update introduces five new voices to ChatGPT: Arbor, Maple, Sol, Spruce, and Vale. These additions bring the total number of available voices to nine. The aim of these nature-inspired names is to enhance the natural feel of using ChatGPT. The AVM feature improves speech patterns, tone, pitch and adds emphasis and prosody for more engaging and lifelike dialogues.
OpenAI removes Sky voice
The voice named Sky, showcased by OpenAI during its spring update, is not included in this rollout. This exclusion follows a legal threat from actress Scarlett Johansson, who claimed that Sky's voice was too similar to hers. In response, OpenAI removed Sky's voice and stated it never intended to mimic Johansson's voice, despite several staff members referencing the movie Her in tweets at the time.
Video and screen sharing feature not included
The AVM update does not include ChatGPT's video and screen sharing facility, which was introduced by OpenAI during its spring update four months back. This feature is designed to allow GPT-4o to process visual as well as audible information simultaneously. However, OpenAI has not provided a timeline for when these multimodal capabilities will be launched.
AVM enhanced since its limited alpha test
Since the limited alpha test of AVM, OpenAI has made several improvements. The company claims that ChatGPT's voice feature is now better at understanding accents, and its conversations are smoother and faster. Additionally, some of ChatGPT's customization features have been extended to AVM, including Custom Instructions and Memory. However, an OpenAI spokesperson confirmed that AVM is not yet available in several regions including the EU, UK, Switzerland, Iceland, Norway and Liechtenstein.
