Enhancing ASR Performance in the Medical Domain for Dravidian Languages

Sri Charan Devarakonda 1 , Ravi Sastry Kolluru 1 , Manjula Sri Rayudu 2 , Rashmi Kapoor 2 , Madhu G 2 , Anil Kumar Vuppala 1 · 2026

Read Enhancing ASR Performance in the Medical Domain for Dravidian Languages by Sri Charan Devarakonda 1 , Ravi Sastry Kolluru 1 , Manjula Sri Rayudu 2 , Rashmi Kapoor 2 , Madhu G 2 , Anil Kumar Vuppala 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-19797.txt.

Description

Devarakonda and five colleagues propose a confidence-aware training framework for speech recognition in Telugu and Kannada medical settings, where annotated data is scarce and words are complex. The method blends real and synthetic speech through a hybrid confidence mechanism mixing static perceptual and acoustic similarity scores with dynamic model entropy. Unlike direct fine-tuning, it uses both fixed-weight and learnable-weight confidence aggregation to guide sample weighting during training. The framework is tested on Telugu and Kannada medical datasets, making varied data sources usable despite limited annotation.

Catalog record

CREATORSri Charan Devarakonda 1 , Ravi Sastry Kolluru 1 , Manjula Sri Rayudu 2 , Rashmi Kapoor 2 , Madhu G 2 , Anil Kumar Vuppala 1
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24906
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.19797

Browse the full library · About the Tower of Babel