Philipp Schmid(@_philschmid)

Today, we’re launching Gemini 3.5 Transcribe, our new speech-to-text model with sub-second streaming...

8.5内容质量

TL;DR · AI 摘要

Google推出Gemini 3.5 Transcribe语音识别模型，实现亚秒级实时流处理，非流式WER低至2.6%，支持85种语言并减少70%处理时间。

核心要点

  • Gemini 3.5 Transcribe非流式WER达2.6%，流式为4.0%
  • 支持85+语言及字母数字标记处理
  • 相比Chirp 3模型处理时间缩短70%

结构提纲

按章节快速跳转。

  1. §模型发布

    宣布推出Gemini 3.5 Transcribe语音转文本模型

  2. ·核心特性

    支持亚秒级实时流处理和智能后处理

  3. ·性能指标

    非流式WER 2.6%，流式WER 4.0%

  4. ›语言支持

    覆盖85+语言及字母数字标记处理

  5. ·应用优化

    减少70%最终转录时间，支持多说话人识别

  6. ›用户反馈

    提升语音输入使用率，智能修正指令语境

思维导图

用一张图看清主题之间的关系。

查看大纲文本（无障碍 / 无 JS 友好）
  • Gemini 3.5 Transcribe
    • 核心特性
      • 亚秒级实时流处理
      • 智能后处理
    • 性能指标
      • 2.6%非流式WER
      • 4.0%流式WER
    • 应用优化
      • 85+语言支持
      • 70%时间缩减

金句 / Highlights

值得收藏与分享的关键句。

#语音识别#Gemini#AI模型#实时处理
打开原文

Philipp Schmid on X: "Today, we’re launching Gemini 3.5 Transcribe, our new speech-to-text model with sub-second streaming and intelligent post-processing for agent interfaces. 2.6% WER on non-streaming and 4.0% on streaming. Supports 85+ languages, cleans up conversational disfluencies ("um", "ah… / X

Philipp Schmid

@_philschmid

Today, we’re launching Gemini 3.5 Transcribe, our new speech-to-text model with sub-second streaming and intelligent post-processing for agent interfaces. 2.6% WER on non-streaming and 4.0% on streaming. Supports 85+ languages, cleans up conversational disfluencies ("um", "ah", and mid-sentence self-corrections), handles alphanumeric tokens (postal codes or IDs), 70% reduction time for the final transcription compared to Chirp 3. - gemini-3.5-transcribe-live: Sub-second, bidirectional streaming via the Gemini Live API - gemini-3.5-transcribe: Recorded processing via the Interactions API with word-level timestamps and speaker attribution (up to 3 speakers). Having access to Transcribe has made me use voice input way more than I ever did before. The native post-processing is incredible. It knows you mean .json and not a person named "Jason". I sometimes speak 5 minutes into and it perfectly refactors my instruction to my context. Available today in public preview in Gemini API,

@

GoogleAIStudio

and already in the

GeminiApp

and

antigravity

. Try here:

ai.studio/apps/bundled/g…

$

00:00

/$

5:05 PM · Aug 26, 2026

14.5K

Views

22

8

127

27