Skip to content

Understanding Multilingual Medical ASR Adaptation Through Layer-Wise Analysis

Aug 2026 · 0 citations · 25 references
Computer Science

TL;DR

This paper investigates how multilingual medical adaptation reshapes the internal representations of Whisper models through layer-wise encoder analysis, and shows that English medical fine-tuning produces the dominant encoder shift, whereas multilingual continuation largely preserves the adapted representation space.

Abstract

Medical automatic speech recognition (MedASR) requires adaptation to specialised terminology, limited annotated clinical data, and multilingual use cases. Although large-scale pretrained ASR models such as Whisper achieve strong generalisation, their behaviour after medical and multilingual adaptation remains insufficiently understood beyond word error rate (WER). This paper investigates how multilingual medical adaptation reshapes the internal representations of Whisper models through layer-wise encoder analysis. We compare zero-shot decoding, English-only fine-tuning, German-only diagnostic fine-tuning, two-stage EN->EN+DE continuation, and direct EN+DE fine-tuning across Whisper model sizes. Fine-tuning substantially improves MedASR performance, but the best model depends on the adaptation setting: Whisper-Medium gives the lowest English WER (7.72%) and the lowest combined EN+DE WER under direct EN+DE training (26.30%); German-only Whisper-Large-v3 gives the lowest German WER (44.96%), but as a within-corpus diagnostic on 86 single-speaker training utterances rather than robust generalisation. Layer-wise analysis of the two-stage Whisper-Small trajectory shows that English medical fine-tuning produces the dominant encoder shift, whereas multilingual continuation largely preserves the adapted representation space. Domain and language information remain highly recoverable across layers, while linearly recoverable error-predictive cues weaken as WER improves.

View source

Similar papers

Preprint Aug 2026

myMediWhisper: Construction of Burmese Medical Speech Corpus and Whisper Fine-Tuning for Clinical Dialogue ASR

This work presents a Burmese medical speech recognition framework built on a high-quality 28-hour corpus recorded and validated by native speakers, and fine-tune Whisper models using full fine-tuning (FFT) and parameter-efficient fine-tuning (PEFT) with LoRA.

Ye Kyaw Thu, Y. Lin, Thura Aung et al. · 0 citations
Conference Jul 2026

RG-RCT: Real-Governed Representation-Calibrated Training for Low-Resource Medical ASR

Developing Automatic Speech Recognition (ASR) systems for specialized medical domains is challenging in lowresource settings due to the limited availability of annotated speech data. Synthetic speech generated using text to speech (TTS) systems is often used to augment training data, but directly mixing synthetic and real speech can introduce distribution mismatch that degrades encoder representations. This work introduces Real-Governed Representation-Calibrated Training (RG-RCT), a training strategy that regulates the influence of synthetic speech at the representation level. The proposed framework combines reliability-based confidence estimation, representation alignment, and layer-wise representation governance to stabilize the encoder feature space while preserving acoustic diversity from multi-speaker synthetic data. Experiments on medical-domain speech datasets using Wav2Vec2 and Whisper demonstrate that RG-RCT consistently outperforms conventional strategies such as direct mixing and confidence-aware training. The proposed method achieves the lowest Word Error Rate (WER), reaching 17.1% and 16.6% for Telugu and 23.3% and 25.6% for Kannada. These results indicate that regulating encoder representation geometry enables more effective utilization of synthetic speech for low-resource medical ASR.

S. Devarakonda, Ravi Sastry Kolluru, Srihari Bandarupalli et al. · 0 citations
Open access Jul 2026

Benchmarking Speech Recognition Models for Medical Consultations in Latin American Spanish: A Comparative Evaluation with Fine-Tuning

Whisper Large v3 and its fine-tuned variant are the best open-source STT models for transcribing medical conversations in LatAm Spanish and provide an evidence base for developing AI medical scribes tailored to Spanish-speaking LatAm.

R. M. Carrillo, A. Carbajal Serrano, P. S. Condori Pinedo · 0 citations
Open access Aug 2026

Application of Conformer Architecture in Clinical Speech Input and Intelligent Medical Record Generation

This study proposes a multi-stage intelligent documentation framework that integrates a 12-layer Conformer architecture, BERT-BiLSTM-CRF semantic modeling, and BART-based structured text generation that illustrates the effectiveness of deep time-frequency feature extraction and contextual sequence modeling for complex noisy signals.

X. Zou, L. Wang, J. Sun et al. · 0 citations
Preprint Jul 2026

Bridging the English-Arabic Medical Knowledge Gap: Targeted Low-Rank Adaptation via Causal Layer Selection

It is shown that mechanistic diagnosis can serve as a practical guide for targeted adaptation in underrepresented-language medical LLMs, and Targeted Low-Rank Adaptation (TLoRA) is proposed, restricted to the layer window where cross-lingual representations diverge, upstream of the output layers where the failure manifests.

Chaimae Abouzahir, Musa Khan, Hala Ali-Hassan et al. · 0 citations
Conference Jul 2026

Domain-Adapted Language Models for English-Turkish Medical Text Translation: An Evaluation of Zero-Shot, Few-Shot, and Fine-Tuning Strategies

This study investigates the English-to-Turkish clinical translation performance of three large language models (Aya-23, Seed-X-Instruct, Gemma-3) using zero/few-shot prompting and QLoRA fine-tuning. Our three-stage setup evaluates: (1) general translation after FLORES-devtest training, (2) medical domain adaptation on i2b2 via a mixed dataset, and (3) generalization capacity on the independent FLORES-test set after mixed-data fine-tuning, using SacreBLEU, chrF, and COMET. Results show that Seed-X-Instruct achieves the strongest fine-tuned performance, reaching 32.70 BLEU, 64.34 chrF, and 90.65 COMET on the general FLORES test set, and 22.40 BLEU, 54.61 chrF, and 79.70 COMET on the medical i2b2 set. After mixed-data training, Seed-X improves from 11.48 to 32.39 BLEU on the unseen FLORES test set, whereas Aya-23 drops from 19.23 to 10.83 and Gemma-3 largely preserves its strong zero-shot baseline. These findings indicate that QLoRA effects are model-dependent and that domain adaptation does not yield the same generalization behavior across models.

Muhammetalp Erdem, Ş. Bora, Oguz Dikenelli · 0 citations

Related blog posts