Clinical-grade speech models reduce word error rates by up to 93% versus generalist speech models and APIs, powering greater accuracy for the agentic era of healthcare.

Corti, the frontier lab for clinical-grade AI, launched Symphony for Speech-to-Text: a new generation of clinical-grade speech-to-text models for real-time dictation, conversational transcription, and batch audio file processing. The latest models sharpen the accuracy of today’s voice-powered healthcare applications while giving the next generation of clinical agents a more reliable speech layer to reason from.

Highlights from the research include:

• Up to 93% lower word error rate versus leading speech models tested on English, German, and French medical terminology, achieving 1.4% WER in English compared with 17.7% for OpenAI, 18.1% for ElevenLabs, 17.4% for Whisper, and 18.9% for Parakeet.
• Clinically usable formatting, reaching 98.3% recall on formatted entities such as dosages, measurements, and dates, compared with 44.3% for the strongest baseline.
• Dragon Medical One-level clinical dictation performance, achieving 4.6% WER on real-world English medical dictation versus Dragon’s 5.7% (improvement of 19%), with higher medical term recall than Dragon (93.5% vs 92.9%).
• Consistent multilingual gains, with 2.4% WER in German versus 13.0% for the next-best system, and 3.9% WER in French versus 10.6%.

Marketing Technology News: From Data to Impact: How AI is Transforming Interactive CTV Ads

Why it matters

Healthcare’s speech recognition options have until now largely split into two categories: general-purpose cloud APIs built for broad-domain transcription and challenged by clinical terminology, or dedicated medical dictation products optimised for clinician dictation but not built as infrastructure for the ambient AI, agents, and real-time clinical support tools now emerging across healthcare. Symphony for Speech-to-Text is Corti’s latest answer: a system that produces structured, clinically usable output through a production-grade API designed for a wide range of clinical workflows – helping downstream AI reason over cleaner facts, not just cleaner transcripts.

“Speech has always been one of healthcare’s most important inputs,” said Andreas Cleve, co-founder and CEO of Corti. “What is changing is what happens after the words are captured. In the agentic era, speech recognition requires more than simply producing a transcript – we need to give AI systems accurate clinical facts to reason from. If a model mishears a medication, dosage, or symptom, every downstream step becomes less reliable. Symphony for Speech-to-Text gives healthcare builders a speech layer accurate enough to thrive in clinical reality.”

Marketing Technology News: MarTech Interview with Haley Trost, Group Product Marketing Manager @ Braze

Early adopters are already building on Symphony for Speech-to-Text in some of the most linguistically demanding clinical environments. Switzerland for example – where care is delivered across multiple languages, often within a single institution – is a stringent proving ground for multilingual medical speech:

“In a clinical conversation, every word matters – a missed medication name, a misheard dosage, or a mistranscribed symptom can change the meaning of an encounter. Symphony’s accuracy on clinical terminology gives us the foundation to bring more trusted AI capabilities into clinical workflows with our Voicepoint Xenon^® platform,” said Pierre Corboz, Head of Solutions & Business Development, Voicepoint. “When Corti improves the speech layer, the workflows we build together become sharper, safer, and more useful for clinicians in Switzerland.”