# SamaVaani: Auditing and Debiasing Multilingual Clinical ASR for Indian Languages

arXiv: https://arxiv.org/abs/2606.26901 · HTML: https://mlsanity.com/paper/2606.26901/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 46 references that are in the map.
- First posted on arXiv 25 Jun 2026.

## Builds on (most cited first, 11 of 11)

- [A Simple Framework for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2002.05709/) (2020)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [A Survey of Code-switched Speech and Language Processing](https://mlsanity.com/paper/1904.00784/) (2019)
- [Cross-modal Contrastive Learning for Speech Translation](https://mlsanity.com/paper/2205.02444/) (2022)
- [Contrastive Learning for Fair Representations](https://mlsanity.com/paper/2109.10645/) (2021)
- [Svarah: Evaluating English ASR Systems on Indian Accents](https://mlsanity.com/paper/2305.15760/) (2023)
- [A Contrastive Learning Approach to Mitigate Bias in Speech Models](https://mlsanity.com/paper/2406.14686/) (2024)
- [Aligning Large Language Models for Enhancing Psychiatric Interviews Through Symptom Delineation and Summarization: Pilot Study](https://mlsanity.com/paper/2403.17428/) (2024)
- [A Deep Dive into the Disparity of Word Error Rates Across Thousands of NPTEL MOOC Videos](https://mlsanity.com/paper/2307.10587/) (2023)
- [The NT-Xent loss upper bound](https://mlsanity.com/paper/2205.03169/) (2022)
- [DASHA: Diarization-Aware Speech-LLM ASR for Healthcare in Code-Switched Indic Doctor-Patient Conversations](https://mlsanity.com/paper/2603.06373/) (2026)

## Abstract (arXiv)

Automatic Speech Recognition (ASR) is increasingly used to document clinical encounters, yet its reliability in multilingual and demographically diverse Indian healthcare context remains largely unknown. In this study, we first conduct the systematic audit of ASR performance on real-world psychiatric interview data spanning Kannada, Hindi and Indian English, comparing eight state-of-the-art models including IndicWhisper, WhisperLargeV3, Sarvam, GoogleS2T, Gemma3n, OmniLingual, Vaani, and Gemini. Our results reveal substantial variability across models and languages, with some systems performing competitively in Indian English but failing in regional speech. We further fine-tune two of the best performing opensource models, i.e., Gemma3n and OmniLingual, using various methods. With this, we uncover systematic performance gaps tied to speaker role and gender, raising concerns about equitable deployment in clinical settings, which are further mitigated by fairness-aware fine-tuning. To this end, we propose SamaVaani, a unified debiasing technique that simultaneously improves ASR performance and improves fairness across demographic groups.