Summary of Vietmed: a Dataset and Benchmark For Automatic Speech Recognition Of Vietnamese in the Medical Domain, by Khai Le-duc
VietMed: A Dataset and Benchmark for Automatic Speech Recognition of Vietnamese in the Medical Domain
by Khai Le-Duc
First submitted to arxiv on: 8 Apr 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents VietMed, a large-scale Vietnamese speech recognition dataset in the medical domain. The dataset comprises 16 hours of labeled medical speech, 1000 hours of unlabeled medical speech, and 1200 hours of unlabeled general-domain speech. VietMed is notable for being the world’s largest public medical speech recognition dataset in seven aspects: total duration, number of speakers, diseases, recording conditions, speaker roles, unique medical terms, and accents. The dataset also covers all ICD-10 disease groups and all accents within a country. To facilitate research, the authors release pre-trained models for Vietnamese ASR, including w2v2-Viet and XLSR-53-Viet, as well as fine-tuned models for medical ASR. Interestingly, even without using any medical data during unsupervised pre-training, the best pre-trained model outperforms state-of-the-art models on the medical domain task by reducing word error rate (WER) from 51.8% to 29.6%. The authors make all code, data, and models publicly available. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper creates a huge dataset of Vietnamese speech in the medical field. This is important because most medical datasets are private, so researchers can’t use them. The dataset has lots of different types of medical speech, including conversations between doctors and patients. It’s also really big – 16 hours of labeled speech and 2000 hours of unlabeled speech! The authors want to help other researchers by giving away the code, data, and models they used. |
Keywords
» Artificial intelligence » Unsupervised
