Google DeepMind 推出 Gemini 3.1 Flash Live：讓語音 AI 更自然、更可靠
Original: Gemini 3.1 Flash Live: Making audio AI more natural and reliable
Google DeepMind has officially unveiled its latest voice model, "Gemini 3.1 Flash Live." This model is positioned to deliver lower-latency…
Google DeepMind 發表最新語音模型 Gemini 3.1 Flash Live。該模型專為即時語音互動設計，透過顯著降低延遲與提升精準度，讓 AI 語音對話變得更加流暢與自然。這項更新將大幅優化語音助理與即時互動應用的使用者體驗，並為開發者提供更強大的即時語音 API 支持。
Google DeepMind has officially unveiled its latest voice model, "Gemini 3.1 Flash Live." This model is positioned to deliver lower-latency, higher-precision voice interaction experiences. As AI voice technology becomes more widespread, users have increasingly high expectations for immediacy and naturalness. Gemini 3.1 Flash Live is designed to address the lag and semantic misunderstandings that have commonly plagued voice AI, enabling human-AI voice conversations to flow as smoothly as talking to another person.
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on Google DeepMind Blog →Summaries are AI-generated; the original article is authoritative.