arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2307.03322cs.CL
收藏

BiPhone:文本中跨语言语音影响的建模

BiPhone: Modeling Inter Language Phonetic Influences in Text

Google Research(谷歌研究院);IIT Madras(印度理工学院马德拉斯分校)

Abhirut Gupta, Ananya B. Sai, Richard Sproat, Yuri Vasilevski, James S. Ren, Ambarish Jash, Sukhdeep S. Sodhi, Aravindan Raghuveer

AI总结:

本文提出BiPhone模型,通过挖掘母语与第二语言间的音素混淆来合成带损坏的L2文本,并构建FunGLUE基准,揭示现有语言理解模型在语音噪声下的性能下降,同时提出音素预测预训练任务以提升模型鲁棒性。

摘要:

由于技术不对称,大量人群被迫使用他们识字水平较低的语言上网。这些用户的第二语言(L2)书面文本中往往包含大量受其母语(L1)影响的错误。我们提出一种方法,针对L1和L2语言对挖掘音素混淆(即L1说话者很可能混淆的L2语音)。这些混淆随后被输入到一个生成模型(Bi-Phone)中,用于合成生成带有损坏的L2文本。通过人工评估,我们表明Bi-Phone能够生成合理的损坏文本,这些损坏在不同L1之间存在差异,并且在Web上具有广泛的覆盖性。我们还利用该技术对流行的语言理解基准SuperGLUE进行损坏处理(FunGLUE,即语音噪声化GLUE),并表明当前最先进的(SoTA)语言理解模型表现不佳。我们还引入了一个新的音素预测预训练任务,该任务帮助字节级模型恢复到接近SuperGLUE的性能。最后,我们还发布了FunGLUE基准,以促进语音鲁棒语言模型的进一步研究。据我们所知,FunGLUE是第一个在文本中引入L1-L2交互的基准。

英文摘要:

A large number of people are forced to use the Web in a language they have low literacy in due to technology asymmetries. Written text in the second language (L2) from such users often contains a large number of errors that are influenced by their native language (L1). We propose a method to mine phoneme confusions (sounds in L2 that an L1 speaker is likely to conflate) for pairs of L1 and L2. These confusions are then plugged into a generative model (Bi-Phone) for synthetically producing corrupted L2 text. Through human evaluations, we show that Bi-Phone generates plausible corruptions that differ across L1s and also have widespread coverage on the Web. We also corrupt the popular language understanding benchmark SuperGLUE with our technique (FunGLUE for Phonetically Noised GLUE) and show that SoTA language understating models perform poorly. We also introduce a new phoneme prediction pre-training task which helps byte models to recover performance close to SuperGLUE. Finally, we also release the FunGLUE benchmark to promote further research in phonetically robust language models. To the best of our knowledge, FunGLUE is the first benchmark to introduce L1-L2 interactions in text.

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。

↑