Microsoft AI Releases MAI-Transcribe-2-Streaming: #1 Real-Time Speech-to-Text Model on Artificial Analysis
https://www.marktechpost.com/2026/10/02/microsoft-ai-releases-mai-transcribe-2-streaming-1-real-time-speech-to-text-model-on-artificial-analysis/📌 【Microsoft AI】語音辨識刷新第一，100 毫秒出字
TL;DR：Microsoft 推出串流語音辨識模型 MAI-Transcribe-2-Streaming,在 Artificial Analysis 排名第一。
對語音 agent 來說,慢半秒不是小事,是整段對話體驗的差異。Microsoft AI 這次直接把「等使用者說完才開始辨識」這個假設拿掉。
🤔 即時辨識,邊聽邊出字
Microsoft AI 於 2026 年 10 月 1 日發布 MAI-Transcribe-2-Streaming,這是該公司第一款串流語音辨識(speech-to-text)模型,同時一併推出兩款文字轉語音模型 MAI-Voice-2.1 與 MAI-Voice-2.1-Flash。Artificial Analysis 在 38 款模型的排行中,把它列為最終逐字稿與第一版部分逐字稿(partial transcript)準確率雙項第一。這款模型的目標場景是語音 agent、即時字幕與語音輸入,也就是延遲直接決定使用體驗的場合。
🧩 先猜,再修正,最後定稿
MAI-Transcribe-2-Streaming 是 9 月發布的批次版 MAI-Transcribe-2 的即時版本,支援 60 種語言並具備自動、連續的語言偵測能力。音訊持續串流進來,文字也在講者還在說話時同步串流輸出。模型在收到音訊後,僅約 100 毫秒就會給出第一版猜測(稱為 partial),之後隨著上下文增加持續修正,最終收斂成一份穩定的最終逐字稿。這代表 agent 可以在使用者話還沒說完時,就開始推理或呼叫工具。Microsoft 團隊表示,內部測試顯示文字出現的速度是最接近競品的兩倍。
📊 怎麼測出來的
AA-WER Streaming 這項指標使用約 8 小時的音訊,組成是 AA-AgentTalk(50%)、VoxPopuli(25%)與 Earnings22(25%);延遲則是從語音結束那一刻開始計算,偵測方式採用 SileroVAD。測試結果顯示,第一版 partial 的準確率已經和最終逐字稿相當,這一點對那些需要在講者說完前就採取行動的 agent 特別關鍵。Microsoft 也把這款模型標示在準確率與延遲的 Pareto frontier 上。
💡 定價與整合方式
MAI-Transcribe-2-Streaming 的費用是每小時音訊 0.54 美元,這是到 2026 年底為止的推出期優惠價,Artificial Analysis 將其換算為每千分鐘 9.00 美元。相比之下,批次版 MAI-Transcribe-2 每小時只要 0.10 美元。在串流這個項目上,Microsoft 的定價高於 xAI 與 Meta,大致與 Google 的估算費率相當。整合方式有兩條路徑:一是相容於 OpenAI Realtime WebSocket 的 Realtime API,適合既有架構已採用該協定的應用;二是 Azure Speech SDK,負責處理連線管理、重試與音訊串流,兩者都會回傳中間結果與最終結果。模型也已上架 MAI Playground,並可透過 Vercel 與 Azure Voice Live 取用,LiveKit 支援則標示為即將推出。若要組成完整的語音迴圈,Microsoft 建議搭配 MAI-Voice-2.1-Flash:45 秒音訊在 150 毫秒端到端延遲下生成,費用為每百萬字元 15 美元;MAI-Voice-2.1 則涵蓋 23 種語言、26 個地區,費用為每百萬字元 22 美元。
🎯 對工程師的意義
如果你在打造語音 agent 或即時字幕系統,第一版 partial 的準確率是否足以直接驅動後續動作,是比單純看最終逐字稿 WER 更值得關注的指標。串流版的定價明顯高於批次版,評估導入前,值得先確認自己的場景是否真的需要「邊聽邊反應」的即時性。
🔗 來源
- 標題：Microsoft AI Releases MAI-Transcribe-2-Streaming: #1 Real-Time Speech-to-Text Model on Artificial Analysis
- 作者／機構：Michal Sutter, MarkTechPost
- 連結：https://www.marktechpost.com/2026/10/02/microsoft-ai-releases-mai-transcribe-2-streaming-1-real-time-speech-to-text-model-on-artificial-analysis/
#MicrosoftAI #SpeechToText #VoiceAI #RealTimeAI #AzureSpeech #VoiceAgents #AIInfrastructure #SpeechRecognition #MachineLearning #ArtificialAnalysis
原始資料 MarkTechPost · 收集於 2026-10-03
摘要原文
Microsoft AI has released MAI-Transcribe-2-Streaming , its first streaming speech-to-text (STT) model. It launched on October 1, 2026, alongside 2 text-to-speech models, MAI-Voice-2.1 and MAI-Voice-2.1-Flash. Artificial Analysis ranks it #1 of 38 models for final and first partial transcript accuracy. The model targets voice agents, live captions and dictation, where latency decides the experience. MAI-Transcribe-2-Streaming is the real-time sibling of the batch MAI-Transcribe-2, released in September. It transcribes 60 languages with automatic, continuous language detection. Audio streams in continuously, and text streams back while the speaker is still talking. The model emits its first hypotheses, called partials, just over 100ms after receiving audio. It revises those partials as context arrives, then commits a stable final transcript. An agent can therefore start reasoning or calling tools mid-sentence. Microsoft team states its internal tests show words appearing 2x faster than its closest competitor. The AA-WER Streaming index uses about 8 hours of audio. The mix is AA-AgentTalk (50%), VoxPopuli (25%) and Earnings22 (25%). Latency is timed from the end of speech, as detected by SileroVAD. The first partial is as accurate as the final transcript. That matters for agents that act before the speaker finishes. Microsoft also places the model on the accuracy versus latency Pareto frontier. MAI-Transcribe-2-Streaming costs $0.54 per hour of audio. This is an introductory price through the end of 2026. Artificial Analysis normalizes it to $9.00 per 1,000 minutes. Batch MAI-Transcribe-2 costs $0.10 per hour. On streaming, Microsoft charges more than xAI and Meta, and roughly matches Google’s estimated rate. Microsoft documents 2 integration paths. The Realtime API suits apps already using an OpenAI Realtime-compatible WebSocket. The Azure Speech SDK handles connection management, retries and audio streaming. Both return intermediate and final results. The model is also available in the MAI Playground , through Vercel and Azure Voice Live. LiveKit support is listed as coming soon. Microsoft pairs it with MAI-Voice-2.1-Flash for full voice loops. Flash generates 45s of audio at 150ms end-to-end latency for $15 per 1M characters. MAI-Voice-2.1 covers 23 languages and 26 locales at $22 per 1M characters. Sources: Artificial Analysis , 9to5Mac (Muse pricing) , DataCamp (Grok pricing) , The Batch (Gemini pricing) . Verified October 2, 2026. Check out the Technical details . All credit goes to the researcher of this project. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter . Wait! are you on telegram? now you can join us on telegram as well. Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us The post Microsoft AI Releases MAI-Transcribe-2-Streaming: #1 Real-Time Speech-to-Text Model on Artificial Analysis appeared first on MarkTechPost .
由 claude-code/sonnet 自動生成