← 回總覽

OpenAI 发布 GPT‑Live 全双工语音模型

📅 2026-07-09 08:00 Berryxia.AI 人工智能 4 分鐘 3756 字 評分: 89
GPT‑Live 全双工 语音模型 OpenAI 实时翻译
📌 一句话摘要 OpenAI 推出 GPT‑Live,采用全双工架构实现更自然的实时人机对话,具备实时翻译和对话人格增强。 📝 详细摘要 该推文引用了 OpenAI 官方公告,介绍了 GPT‑Live 作为新一代语音模型的核心升级:全双工架构使模型能够同时听取和输出语音,无需等待用户说完再响应;同时具备更好的时间感、实时翻译能力以及更鲜明的对话人格。模型还能在后台将复杂问题委托给前沿模型处理,再把结果自然地带回对话中。演示显示对话流畅度和打断处理较之前有明显提升,目标是让语音交互接近人与人之间的自然交流,尤其在移动端和日常使用场景中具有重要意义。 📊 文章信息 AI 初评:89 来源:B
Skip to main contentAudio 2 ![Image 1: LogoBest Blogs](https://www.bestblogs.dev/ "BestBlogs.dev")

Search Ctrl+K

Change language Switch ThemeSign In

Curated Daily BriefWeekly PicksTopicsWorld Cup Special SettingsHelp CenterCollapse

Narrow Mode

OpenAI Launches GPT-Live Full-Duplex Voice Model

OpenAI Launches GPT-Live Full-Duplex Voice Model

![Image 2: Berryxia.AI](https://www.bestblogs.dev/en/tweets?sourceId=SOURCE_4287449f) Berryxia.AI

@berryxia

OpenAI推出了GPT-Live,一个新的语音模型世代,主打更自然的实时人机对话。

核心升级是全双工架构(full-duplex),模型可以一边听一边说,不用等用户说完再响应。

同时具备更好的时间感、实时翻译能力,以及更鲜明的对话人格。

它还能在后台把复杂问题委托给前沿模型处理,再把结果自然地带回对话里。

从演示来看,对话的流畅度和打断处理比之前有明显进步,目标是让语音交互真正接近人与人之间的自然交流。

这其实是把语音从“辅助输入方式”进一步推向“主要交互界面”的尝试,尤其在移动端和日常使用场景里。Show More

!Image 3: OpenAI

#### OpenAI

@OpenAI · 13h ago

Introducing GPT-Live, a new generation of voice models for natural human-AI interaction.

Rolling out in ChatGPT starting today.

You’ll want to turn the sound on for this one.Show More

!Image 4: Video thumbnail

03:34

971

1,700

19.2K

3.9M

Jul 9, 2026, 12:00 AM View on X

1 Replies

0 Retweets

1 Likes

1,225 Views ![Image 5: Berryxia.AI](https://www.bestblogs.dev/en/tweets?sourceid=4287449f) Berryxia.AI @berryxia

Follow

One Sentence Summary

OpenAI introduces GPT-Live with a full-duplex architecture for more natural real-time human-computer conversation, featuring real-time translation and enhanced conversational persona.

Summary

This tweet references OpenAI's official announcement, introducing GPT-Live as a new generation of voice model with core upgrades: full-duplex architecture enables simultaneous listening and output without waiting for the user to finish; better sense of time; real-time translation capabilities; and a more distinct conversational persona. The model can also delegate complex tasks to frontier models in the background and bring the results naturally back into the conversation. Demos show noticeably improved dialogue fluency and interruption handling compared to before, aiming to make voice interaction truly approach natural human conversation, especially on mobile and in daily use scenarios.

AI Screening

89

Influence Score 1

Published Today

Language

Chinese

Tags

GPT-Live

Full-Duplex

Voice Model

OpenAI

Real-time Translation

Make your daily reading actually fit you.A daily brief built from the sources you follow. Get started free HomeDiscoverWorld CupSettings

查看原文 → 發佈: 2026-07-09 08:00:57 收錄: 2026-07-09 12:00:40

🤖 問 AI

針對這篇文章提問,AI 會根據文章內容回答。按 Ctrl+Enter 送出。