Microsoft launches MAI-Transcribe-2-Streaming – tops voice AI benchmarks
Microsoft has launched the voice model MAI-Transcribe-2-Streaming, which has reached the top spot in a global benchmark for streaming speech-to-text with a word error rate of only 2.5 per cent.
What happened?
On 1 October 2026, Microsoft launched the voice model MAI-Transcribe-2-Streaming, a WebSocket-based service for streaming speech-to-text. The model secured first place for both final and partial transcription accuracy among 38 models in the AA-WER Streaming benchmark by Artificial Analysis, with a word error rate (WER) of 2.5 per cent. The launch was reported by media on 3 October 2026.
Key facts
| Lanseringsdatum | 1 oktober 2026 |
|---|---|
| Publiceringsdatum | 3 oktober 2026 |
| Ordfelexponent (WER) | 2,5 % |
| Antal testade modeller | 38 stycken |
| Latens för slutgiltigt svar | 0,13 sekunder |
Why it matters
The model is being launched alongside the text-to-speech model MAI-Voice-2.1-Flash and the orchestration layer Azure Voice Live, which is now generally available. This allows Microsoft to offer a complete infrastructure for voice AI with ultra-low latency without the need to send audio data to third-party services.
Who is affected?
The launch is primarily aimed at developers and companies building voice assistants, customer service bots, and real-time interpretation tools. With these tools, developers can run the entire voice agent cycle directly on Azure instead of relying on external providers such as Deepgram or AssemblyAI.
Impact on the EU
The services are distributed globally via Microsoft Azure, meaning that Swedish and European customers are covered by Azure's existing data protection and GDPR frameworks. The models are available in European cloud regions.
What else you should know
The system provides final transcriptions 0.13 seconds after the completion of speech and begins partial hypotheses approximately 100 milliseconds after audio input begins. This compares to competitors such as Deepgram Nova-3, which has an error rate of 5–7 per cent in production environments.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Vilka bolag berörs i Sverige?
The link opens in a new window and leads to the publisher's own site.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.
AI-verktyg i artikeln
Topics
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "Microsoft launches MAI-Transcribe-2-Streaming – tops voice A"