High-precision medical speech recognition through synthetic data and semantic correction: UNITED-MEDASR
Not on the map. Why?
Abstract (arXiv)
Automatic Speech Recognition (ASR) systems in the clinical domain face significant challenges, notably the need to recognise specialised medical vocabulary accurately and meet stringent precision requirements. We introduce United-MedASR, a novel architecture that addresses these challenges by integrating synthetic data generation, precision ASR fine-tuning, and advanced semantic enhancement techniques. United-MedASR constructs a specialised medical vocabulary by synthesising data from authoritative sources such as ICD-10 (International Classification of Diseases, 10th Revision), MIMS (Monthly Index of Medical Specialties), and FDA databases. This enriched vocabulary helps finetune the Whisper ASR model to better cater to clinical needs. To enhance processing speed, we incorporate Faster Whisper, ensuring streamlined and high-speed ASR performance. Additionally, we employ a customised BART-based semantic enhancer to handle intricate medical terminology, thereby increasing accuracy efficiently. Our layered approach establishes new benchmarks in ASR performance, achieving a Word Error Rate (WER) of 0.985% on LibriSpeech test-clean, 0.26% on Europarl-ASR EN Guest-test, and demonstrating robust performance on Tedlium (0.29% WER) and FLEURS (0.336% WER). Furthermore, we present an adaptable architecture that can be replicated across different domains, making it a versatile solution for domain-specific ASR systems.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 23 of its 45 references that are in the map. First posted on arXiv 24 Nov 2024.
Builds on
23 of 45 references.
References · 23
- 2019BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehensioncited by 5,893
- 2022Robust Speech Recognition via Large-Scale Weak Supervisioncited by 4,057
- 2022FLEURS: Few-shot Learning Evaluation of Universal Representations of Speechcited by 469
- 2021W2v-BERT: Combining Contrastive Learning and Masked Language Modeling for Self-Supervised Speech Pre-Trainingcited by 347
- 2020Pushing the Limits of Semi-Supervised Learning for Automatic Speech Recognitioncited by 187
- 2023StyleTTS 2: Towards Human-Level Text-to-Speech through Style Diffusion and Adversarial Training with Large Speech Language Modelscited by 152
- 2023SeamlessM4T: Massively Multilingual & Multimodal Machine Translationcited by 125
- 2023Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognitioncited by 124
- 2023Machine Learning for Synthetic Data Generation: A Reviewcited by 120
- 2020Self-training and Pre-training are Complementary for Speech Recognitioncited by 99
- 2021SpeechStew: Simply Mix All Available Speech Recognition Data to Train One Large Neural Networkcited by 83
- 2019Generating Synthetic Audio Data for Attention-Based Speech Recognition Systemscited by 47
11 more
- 2023HyPoradise: An Open Baseline for Generative Speech Recognition with Large Language Modelscited by 42
- 2021SynthASR: Unlocking Synthetic Data for Speech Recognitioncited by 34
- 2024Generative AI for Synthetic Data Across Multiple Medical Modalities: A Systematic Review of Recent Developments and Challengescited by 24
- 2023ArchiSound: Audio Generation with Diffusioncited by 20
- 2024FAdam: Adam is a natural gradient optimizer using diagonal empirical Fisher informationcited by 18
- 2018Whisper: Fast Flooding for Low-Power Wireless Networkscited by 14
- 2024Mamba for Streaming ASR Combined with Unimodal Aggregationcited by 8
- 2023Deep Generative Models, Synthetic Tabular Data, and Differential Privacy: An Overview and Synthesiscited by 7
- 2024Voice EHR: Introducing Multimodal Audio Data for Healthcited by 2
- 2024Assessing Latency in ASR Systems: A Methodological Perspective for Real-Time Usecited by 2
- 2023Exploring the Viability of Synthetic Audio Data for Audio-Based Dialogue State Trackingcited by 1
Built on by
Citing papers per year: 2024: 1; 2025: 1.
2 papers.