Eka Care Unveils Clinical Speech Model Parrotlet for Medical Transcription

The model includes safeguards aimed at reducing hallucinations, or instances in which AI systems generate information not present in the source audio.

Healthtech startup Eka Care has announced the launch of Parrotlet-a v2, a clinical-grade automatic speech recognition (ASR) model developed for healthcare documentation in India.

The model is designed to transcribe doctor–patient conversations in Hindi and Indian English into structured clinical notes in near real time and is designed to operate in multilingual, acoustically variable clinical environments.

The model has been trained to process code-mixed Hindi and Indian English, regional accents, overlapping speech, and India-specific medical terminology and includes safeguards aimed at reducing hallucinations, or instances in which AI systems generate information not present in the source audio.

Commenting on the new launch, Vikalp Sahni, Founder & CEO, Eka Care. “India’s clinical reality is multilingual, acoustically noisy, and filled with hyper-local medical terminology that global AI systems are not trained to handle. Parrotlet-a v2 is tuned for Indian healthcare. In benchmark evaluations against leading global and India-focused models, including Gemini 3 Pro, Gemini 3 Flash, and Saaras V3, it demonstrated leading Semantic Word Error Rate performance while delivering sub-second inference speeds, enabling documentation to happen seamlessly during consultations.”

The model is integrated into EkaScribe, the digital medical scribe platform developed by Eka Care, which is currently being used by over 3,000 doctors.

The system is based on a 5-billion-parameter architecture, intended to balance computational efficiency with performance in the clinical transcription task.

The internal evaluations of the model suggest that it performs at par with larger systems in Indian healthcare settings and is expected to support wider deployment.

The company further stated that the model delivers transcription outputs in under a second and is expected to support near-real-time note generation in consultation settings within a limited time frame.

Highlighting the model’s significance, Deepak Tuli, Co-founder and COO, Eka Care, said, “Healthcare digitization at scale requires AI that is not only accurate, but fast, reliable, and economically sustainable. “Our focus was to build a specialized AI engine that truly understands Indian clinical practice so providers can deploy it confidently across networks without compromising workflow speed, affordability, or patient safety.”

Currently, the model is commercially available through Eka Care’s platform and APIs.

Founded in 2020, Eka Care operates as a connected healthcare platform offering electronic medical record systems, patient health records, services aligned with India’s Ayushman Bharat Digital Mission (ABDM), and AI-enabled clinical tools.

Going forward, the startup plans to continue research focused on India-specific health AI models and expand integrations across hospital, insurance, and public health systems.

Stay tuned for more such updates on Digital Health News