Automatic Speech Recognition (ASR) performance is heavily dependent on the availability of large-scale, high-quality datasets. For low-resource languages, existing open-source ASR datasets often suffer from insufficient quality and inconsistent annotation, hindering the development of robust models. To address these challenges, we propose a novel and generalizable data aggregation and preprocessing pipeline designed to construct high-quality ASR datasets from diverse, potentially noisy, open-source sources. Our pipeline incorporates rigorous processing steps to ensure data diversity, balance, and the inclusion of crucial features like word-level timestamps. We demonstrate the effectiveness of our methodology by applying it to Vietnamese, resulting in a unified, high-quality 500-hour dataset that provides a foundation for training and evaluating state-of-the-art Vietnamese ASR systems. Our project page is available at https://github.com/qualcomm-ai-research/PhoASR.
ViSpeechFormer: A Phonemic Approach for Vietnamese Automatic Speech Recognition
Faculty of Information Science and Engineering(信息科学与工程学院);University of Information Technology(信息技术大学);Vietnam National University, Ho Chi Minh city(胡志明市越南国家大学)
ViSpeechFormer 是首个基于音素的越南语 ASR 框架,通过建模音素表示实现了更强的性能和更好的泛化能力。
Phonetic Modeling of Dialectal Variation in Vietnamese Speech
Faculty of Computer Science(计算机科学系);Faculty of Information Science and Engineering(信息科学与工程系);University of Information Technology(信息技术大学);Vietnam National University, Ho Chi Minh city(越南国家大学,胡志明市)
Beyond Phones: Structured Phonemic Modeling for Vietnamese Automatic Speech Recognition
Faculty of Information Science and Engineering(信息科学与工程学院);Faculty of Computer Science(计算机科学学院);University of Information Technology(信息技术大学);Vietnam National University, Ho Chi Minh city(越南国家大学,胡志明市)
VietASR: Achieving Industry-level Vietnamese ASR with 50-hour labeled data and Large-Scale Speech Pretraining
School of Computer Science, MoE Key Lab of Artificial Intelligence, Shanghai Jiao Tong University, shanghai innovation instituteChina Tencent AI LabUSA(计算机科学学院、人工智能国家重点实验室、上海交通大学、上海创新研究院中国腾讯AI实验室美国)
Dialogue Act Segmentation for Vietnamese Human-Human Conversational Texts
Thainguyen University (TNU)(太原大学(TNU));University of Information and Communication Technology(信息与通信技术学院);Vietnam National University, Hanoi (VNU)(河内越南国家大学(VNU));University of Engineering and Technology(工程技术学院)
ViToSA: Audio-Based Toxic Spans Detection on Vietnamese Speech Utterances
University of Information Technology(信息技术大学);Vietnam National University Ho Chi Minh City(越南国家大学胡志明市分校);Faculty of Computer Science(计算机科学学院);Faculty of Information Science and Engineering(信息科学与工程学院)
ViMedCSS: A Vietnamese Medical Code-Switching Speech Dataset & Benchmark
College of Engineering and Computer Science(工程与计算机科学学院);Center for AI Research(人工智能研究中心);College of Health Sciences(健康科学学院);University of Technology Sydney(悉尼科技大学)
University of Information Technology(信息技术大学);Vietnam National University, Ho Chi Minh City(胡志明市越南国家大学)
针对其他用户的仇恨和攻击性语言的增加是社交网络平台使用增加的不利副作用之一。这可能会使人类版主难以审查分类系统过滤的标记评论。为了帮助解决这个问题,我们提出了ViHOS(Vietnamese Hate and Offensive Spans)数据集,这是第一个人类注释的语料库,包含11 k条评论上的26 k个跨度。我们还提供了越南语评论中的仇恨和攻击性跨度的定义以及详细的注释指南。此外,我们使用各种最先进的模型进行实验。具体而言,XLM-R $_{Large}$在单跨度检测和所有跨度检测中获得最佳F1分数,而PhoBERT $_{Large}$在多跨度检测中获得最高分数。最后,我们的误差分析表明,在我们的数据中检测特定类型的跨度的困难,为今后的研究。免责声明:本文包含可能被认为是亵渎,冒犯或辱骂的真实评论。
ViHateT5: Enhancing Hate Speech Detection in Vietnamese With A Unified Text-to-Text Transformer Model
Faculty of Information Science and Engineering, University of Information Technology(信息科学与工程学院,信息科技大学);Vietnam National University, Ho Chi Minh City(越南国家大学胡志明市分校)
Human-Guided Reasoning with Large Language Models for Vietnamese Speech Emotion Recognition
University of Information Technology, Ho Chi Minh City, Vietnam(越南胡志明市信息技术大学);Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam(越南胡志明市国家大学)
Whisper based Cross-Lingual Phoneme Recognition between Vietnamese and English
The University of Danang, Vietnam - Korea University of Information and Communication Technology(丹江大学-韩国信息通信技术大学);Communication Technology Industrial University of Ho Chi Minh City(胡志明市通信技术工业大学)
Revisiting End-to-End Speech-to-Text Translation From Scratch
University of Edinburgh(爱丁堡大学);University of Zurich(苏黎世大学)
端到端(E2 E)语音到文本翻译(ST)通常依赖于通过语音识别或文本翻译任务使用源转录来预训练其编码器和/或解码器,没有这些任务,翻译性能会大幅下降。然而,成绩单并不总是可用的,以及如何显着这样的预训练是E2 E ST很少在文献中研究。在本文中,我们重新审视这个问题,并探讨在何种程度上的语音翻译对单独训练的E2 E ST的质量可以提高。我们重新审视了之前被证明对ST有益的几种技术,并提供了一组最佳实践,使基于Transformer的E2 E ST系统偏向于从头开始训练。此外,我们提出了参数化的距离惩罚,以促进局部性的语音自注意模型的建模。在涵盖23种语言的四个基准上,我们的实验表明,在不使用任何成绩单或预训练的情况下,所提出的系统达到甚至超过了以前采用预训练的研究,尽管在(极)低资源环境中仍然存在差距。最后,我们讨论了神经声学特征建模,其中神经模型被设计为直接从原始语音信号中提取声学特征,其目标是简化归纳偏差并在描述语音时增加模型的自由度。我们第一次证明了它的可行性,并在ST任务上显示出令人鼓舞的结果。
Residual Convolutional CTC Networks for Automatic Speech Recognition
Tencent AI Lab, Tencent(腾讯AI实验室,腾讯);Tsinghua University(清华大学)
深度学习方法已广泛应用于自动语音识别(ASR)中,并取得了显着的准确性提高。特别是,卷积神经网络(CNN)最近在ASR中被重新审视。然而,现有工作中使用的大多数CNN都少于10层,这可能不够深,无法捕获所有人类语音信号信息。在本文中,我们提出了一种新型的深度和广度的CNN架构,表示为RCNN-CTC,它具有剩余连接和连接主义时间分类(CTC)损失函数。RCNN-CTC是一个能够同时利用语音信号的时域和频谱结构的端到端系统。此外,我们介绍了一种基于CTC的系统组合,这是不同于传统的逐帧的senone为基础的。组合中采用的基本子系统是不同类型的,因此相互补充。实验结果表明,与ASR中广泛使用的几种神经网络系统相比,我们提出的单系统RCNN-CTC在WSJ和Tencent Chat数据集上可以实现最低的字错误率(WER)。此外,所提出的系统组合可以在这两个数据集上提供进一步的错误减少,从而分别在WSJ dev 93和Tencent Chat数据集上导致$14.91\%$和$6.52\%$的相对WER减少。
MultiMed: Multilingual Medical Speech Recognition via Attention Encoder Decoder
University of Toronto(多伦多大学);University Health Network(大学健康网络);Knovel Engineering Lab(Knovel工程实验室);FPT University(佛邦大学);Florida Institute of Technology(佛罗里达技术学院);KU Leuven(鲁文大学);VNU University of Engineering and Technology(越南工程与技术大学);Johns Hopkins University(约翰霍普金斯大学);University of Alabama at Birmingham(阿拉巴马大学伯明翰分校)
HSD Shared Task in VLSP Campaign 2019:Hate Speech Detection for Social Good
Umeå University(于默奥大学);The Australian E-Health Research Centre, CSIRO(澳大利亚电子健康研究中心,联邦科学与工业研究组织);University of Engineering and Technology (UET), Vietnam National University, Hanoi(河内国家大学工程技术大学);InfoRE Technology Company(InfoRE技术公司);Hanoi University of Science (HUS), Vietnam National University, Hanoi(河内国家大学自然科学大学)