arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2306.06384cs.CL
收藏

面向低资源不流利修正的对抗训练

Adversarial Training For Low-Resource Disfluency Correction

Indian Institute of Technology Bombay(印度理工学院孟买分校)

Vineet Bhat, Preethi Jyothi, Pushpak Bhattacharyya

AI总结:

针对低资源场景下的不流利修正问题,提出结合少量标注真实数据、大量无标注数据与合成不流利数据的对抗训练序列标注模型,在三种印度语言及英语口吃修正任务上取得显著提升,并建立新基准。

摘要:

不流利现象在会话语音中普遍存在。带有不流利内容的语音会产生含噪声的自动语音识别(ASR)文本,进而影响机器翻译等下游任务。本文提出一种用于不流利修正(DC)的对抗训练序列标注模型,该模型结合少量标注的真实不流利数据与大量无标注数据进行训练。我们通过在三种印度语言——孟加拉语、印地语和马拉地语(均属于印度-雅利安语系)上开展DC任务评估,验证了所提技术的有效性,该技术的核心依赖于合成生成的不流利数据。我们的技术在去除由言语障碍导致的ASR文本中的口吃不流利现象方面也表现良好。在上述三种语言上,我们的方法相较于强基线模型的F1值平均提升了6.15个点。据我们所知,我们是首个将对抗训练用于DC任务,并将其应用于修正英语口吃不流利现象的研究,为该任务建立了新的基准。

英文摘要:

Disfluencies commonly occur in conversational speech. Speech with disfluencies can result in noisy Automatic Speech Recognition (ASR) transcripts, which affects downstream tasks like machine translation. In this paper, we propose an adversarially-trained sequence-tagging model for Disfluency Correction (DC) that utilizes a small amount of labeled real disfluent data in conjunction with a large amount of unlabeled data. We show the benefit of our proposed technique, which crucially depends on synthetically generated disfluent data, by evaluating it for DC in three Indian languages- Bengali, Hindi, and Marathi (all from the Indo-Aryan family). Our technique also performs well in removing stuttering disfluencies in ASR transcripts introduced by speech impairments. We achieve an average 6.15 points improvement in F1-score over competitive baselines across all three languages mentioned. To the best of our knowledge, we are the first to utilize adversarial training for DC and use it to correct stuttering disfluencies in English, establishing a new benchmark for this task.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑