news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

original title: chatgpt advanced voice mode is now available! supports over 50 languages. in the demo video, when saying "i'm sorry" in mandarin, scarlett's voice is gone.

on september 25, openai officially announced that chatgpt's advanced voice mode will be officially launched to chatgpt plus individual users and small business team (teams) users this week. this feature will be launched first in the us market.

in addition, openai said that this feature will be open to openai enterprise edition and education edition users next week. but it is worth noting that the new voice function is applicable to openai's gpt-4o model, not the recently released preview model o1.

image source: x social platform

this update means thatstarting this week, chatgpt plus users and small business team users can interact with chatbots through "voice" rather than traditional text input.。

two highlights of the advanced voice mode are particularly eye-catching:support users to set "custom commands" for voice assistants, to achieve personalized operation; second,possesses a "memory" function that can remember the user's preferred interaction method, a similar feature to the one launched earlier this year for the text version of chatgpt.

image source: x social platform

in the official video, charlotte cole, technical project manager at openai, and mike, research engineer at openai, said,users can not only customize the speed of the conversation, but also let the model communicate with the user's name or preferred name, making communication more cordial and natural.

also,users can also preset their personal name and address information in the systemwhen initiating a new round of conversation, such as asking "the weather is great this weekend, can you recommend any fun outdoor activities?" the advanced voice assistant will call up the address information previously entered by the user, proactively recommend nearby places to visit, and even thoughtfully plan travel routes.

image source: x social platform

in order to meet the preferences of different users,advanced voice mode adds five new unique voices: arbor, maple, sol, spruce, and vale join the original four voices of breeze, juniper, cove, and ember to create nine voice lines. the names of these voices are inspired by natural elements and are designed to provide diverse tones and characteristics.

it is worth noting thatopenai removes sky voice after being accused of imitating actress scarlett johanssonpreviously, scarlett accused openai's chatgpt of illegally using her voice and demanded that the voice be removed from the shelves.

in addition, openai said they alsothe conversation capabilities of some foreign languages ​​have been optimized, which not only improves the conversation speed and fluency, but also makes detailed adjustments to the accent., striving to communicate closer to nature.

drew, a model designer at openai, also shared his experience of using it. he said that in daily use, users can put the advanced voice assistant aside, and it will wait in silence without disturbing the user. when users have any questions or needs, they can start a conversation with it at any time. it will quickly capture the changes in the tone of the conversation and flexibly play various roles, just like talking to a real friend naturally and smoothly.

image source: x social platform

chatgpt advanced voice mode now supports more than 50 languages, expanding the user's communication scope. what's particularly interesting is that in the official demonstration video, the user asked the voice assistant to apologize to his grandmother for keeping her waiting for a long time.the advanced voice assistant first fluently summarized it in english, and after the user said "grandma only speaks mandarin", it expressed it again in standard mandarin "sorry, i'm late", as if openai was apologizing to users for repeatedly delaying the release of advanced speech models.

it should be noted thatthe advanced voice mode is not yet available in the eu, uk, switzerland, iceland, norway and liechtenstein.openai has not yet announced when these regions will be open.

image source: x social platform

openai ceo sam altman could not hide his excitement on the social platform, saying "i hope you think the wait is worth it", and added an expression of grievance and a heart shape.

image source: x social platform

greg brockman, president of openai, who is still on vacation, also enthusiastically participated in the promotion. he said: "the introduction of advanced voice functions allows you to easily have a smooth and unimpeded conversation with chatgpt. at that moment, you may realize how unnatural it was to communicate by typing laboriously on a computer in the past."