arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2304.05974eess.AS
收藏

面向语音应用的对比预测编码正则化

Regularizing Contrastive Predictive Coding for Speech Applications

Saurabhchand Bhati, Jesús Villalba, Piotr Żelasko, Laureano Moro-Velazquez, Najim Dehak

AI总结:

针对CPC语音表征学习,本文提出自表达约束与Left-or-Right正则化两种慢变特征约束,使100小时无标注训练达到基线360小时效果,并在多场景ABX任务中稳定优于基线。

摘要:

对比预测编码(Contrastive predictive Coding,CPC)等自监督方法已大幅提升了无监督表征的质量。这些表征显著减少了下游任务(如自动语音识别)达到所需性能时需要的标注数据量。CPC通过在给定当前帧的情况下学习预测未来帧来学习表征。基于声学信息(例如音素)的变化慢于CPC中特征提取速率这一观察,我们提出了对特征施加慢变约束的正则化技术。本文提出两种正则化技术:自表达约束和Left-or-Right正则化。我们在ABX、线性音素分类、声学单元发现以及自动语音识别任务上评估了所提出的模型。在100小时无标注数据上训练的正则化CPC,可与在360小时无标注数据上训练的基线CPC性能相当。我们还表明,所提正则化技术与数据增强互补,并可进一步提升系统性能。在单语、跨语或多语设置下,无论是否使用数据增强、无论训练数据量多少,我们的正则化模型在ABX任务上均优于基线CPC模型。

英文摘要:

Self-supervised methods such as Contrastive predictive Coding (CPC) have greatly improved the quality of the unsupervised representations. These representations significantly reduce the amount of labeled data needed for downstream task performance, such as automatic speech recognition. CPC learns representations by learning to predict future frames given current frames. Based on the observation that the acoustic information, e.g., phones, changes slower than the feature extraction rate in CPC, we propose regularization techniques that impose slowness constraints on the features. Here we propose two regularization techniques: Self-expressing constraint and Left-or-Right regularization. We evaluate the proposed model on ABX and linear phone classification tasks, acoustic unit discovery, and automatic speech recognition. The regularized CPC trained on 100 hours of unlabeled data matches the performance of the baseline CPC trained on 360 hours of unlabeled data. We also show that our regularization techniques are complementary to data augmentation and can further boost the system's performance. In monolingual, cross-lingual, or multilingual settings, with/without data augmentation, regardless of the amount of data used for training, our regularized models outperformed the baseline CPC models on the ABX 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。

↑