Multilingual Error Rate Drops 50%: Microsoft Upgrades Doctor-Patient Conversation AI Transcription Engine, Now Covering 1/3 of US Physicians
IT之家 November 18 news: at the Ignite 2026 event held yesterday (November 17), Microsoft announced that its healthcare AI product Dragon Copilot Transcribe has officially adopted its self-developed speech recognition model MAI-Transcribe-2.

Microsoft says the model is currently the fastest and most accurate multilingual speech transcription model, and now handles 100% of doctor-patient conversation transcription tasks in Dragon Copilot.
In terms of scale application, Dragon Copilot Transcribe processes about 10 million patient visit scenarios per month, and achieves about a 50% reduction in error rate in multilingual workflows, with about one-third of physicians in the United States already using it. Microsoft says the model can recognize multiple languages and common mixed-language scenarios such as Chinese-English and Hindi-English, and can maintain high recognition quality in noisy environments.
In real-world scenarios, this means consultation recordings can be converted into structured clinical records faster and more accurately, reducing the time spent manually organizing medical records, verifying terminology, and translating information. For multilingual patients, more stable transcription capabilities also help reduce the risk of medical information distortion caused by language recognition errors.
MAI-Transcribe-2 supports speaker diarization, word-level timestamps, keyword biasing, and configurable transcription styles. It can also automatically detect language and supports two output modes, "verbatim" and "concise," suitable for different needs such as compliance archiving, analysis, and medical record writing.
Microsoft says MAI-Transcribe-2 achieves leading performance across 60 languages on the FLEURS benchmark, with an average word error rate of 5.2%; it is also in the leading range in Artificial Analysis's combined accuracy and latency evaluation. The company also says the model can process long audio up to 10 times faster than some competitors, and is offering a limited-time price of $0.10 per hour of audio (IT之家 note: approximately 0.67 RMB at current exchange rates).
Related Articles
- Microsoft's most cost-effective coding AI model: MAI Code 2 achieves about 96% of Claude Opus 5.5's performance at 5% of the cost
- Why are some Microsoft products always a bit short on polish? Azure engineering lead says large customer demands take priority
- No subsidies, no passing costs to others: Microsoft promises to pay for its AI expansion electricity bills itself
- Microsoft officially releases .NET 11: development saves more time, AI tools more complete
- AI boom collides with resource disputes, Microsoft's under-construction data center in the Netherlands faces protests
- Borrowing money to do AI: sources say Microsoft plans to raise at least $25 billion through US bond issuance