Google has launched MedASR, an open-weight medical speech-to-text model designed specifically for healthcare and life sciences applications, as part of its Health AI Developer Foundations program. The release comes alongside continued development of MedGemma 1.5, Google’s latest medical image and text interpretation model, positioning both tools as a unified foundation for next-generation clinical AI workflows.
MedASR has been trained on approximately 5,000 hours of de-identified medical audio, including physician dictations and clinical conversations spanning specialties such as radiology, internal medicine, and family medicine. Unlike general-purpose speech recognition systems, MedASR is optimized for medical vocabulary, clinical workflows, and complex terminology, allowing it to more accurately convert spoken medical language into structured text.
The model is designed to support medical dictation, clinical documentation, and physician-patient transcription, and is intended to serve as a core building block for voice-enabled healthcare applications. Google has positioned MedASR as a foundational model that developers can use to build healthcare-focused voice systems, from clinical reporting tools to conversational medical assistants.
The launch is aligned with Google’s broader rollout of MedGemma 1.5, an upgraded version of its open medical AI model built for interpreting medical images and clinical text. MedGemma 1.5 extends support to high-resolution medical imaging, including CT scans, MRI volumes, and digital pathology slides, while also improving performance on medical reasoning and classification tasks.
Together, MedASR and MedGemma 1.5 enable systems that can take spoken clinical input, convert it into text, and analyze it alongside medical images for advanced healthcare applications.
Both models are released under Google’s Health AI Developer Foundations initiative and are available to researchers, startups, and healthcare technology teams through platforms such as Hugging Face and Google Cloud’s Vertex AI, allowing them to be integrated into commercial and research-grade healthcare solutions.
With the introduction of MedASR, Google is expanding its portfolio of open medical AI models aimed at improving clinical documentation, diagnostic workflows, and voice-based healthcare systems, while MedGemma 1.5 provides the imaging and reasoning backbone needed for comprehensive medical AI applications.
Also Read