arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2201.10207eess.AScs.CLcs.LGcs.SD
收藏

SPIRAL:用于语音预训练的自监督扰动不变表示学习

SPIRAL: Self-supervised Perturbation-Invariant Representation Learning for Speech Pre-Training

Huawei Noah’s Ark Lab(华为诺亚方舟实验室)

Wenyong Huang, Zhenhe Zhang, Yu Ting Yeung, Xin Jiang, Qun Liu

摘要:

我们介绍了一种名为SPIRAL的语音预训练新方法,该方法通过在教师-学生框架中学习扰动数据的去噪表示来工作。具体来说,给定一个语音话语,我们首先将话语馈送到教师网络以获得相应的表示。然后,相同的话语被扰动并馈送到学生网络。学生网络被训练成输出类似于教师的表示。同时,教师网络被更新为学生权重在训练步骤上的移动平均值。为了防止表征崩溃,我们将话语中的对比损失作为预训练目标,并对教师的输入施加位置随机化。与最先进的语音预训练方法wav 2 vec 2.0相比,SPIRAL实现了具有竞争力或更好的结果,并显着降低了训练成本(BASE模型为80%,LARGE模型为65%)。此外,我们解决了噪声鲁棒性的问题,这是现实世界的语音应用程序的关键。我们提出了多条件预训练通过扰动学生的输入与各种类型的加性噪声。我们证明了多条件预训练的SPIRAL模型对嘈杂的语音更鲁棒(在真实嘈杂的测试数据上相对字错误率降低9.0% - 13.3%),相比于仅在微调阶段应用多条件训练。源代码可在https://github.com/huawei-noah/Speech-Backbones/tree/main/SPIRAL上获得。

英文摘要:

We introduce a new approach for speech pre-training named SPIRAL which works by learning denoising representation of perturbed data in a teacher-student framework. Specifically, given a speech utterance, we first feed the utterance to a teacher network to obtain corresponding representation. Then the same utterance is perturbed and fed to a student network. The student network is trained to output representation resembling that of the teacher. At the same time, the teacher network is updated as moving average of student's weights over training steps. In order to prevent representation collapse, we apply an in-utterance contrastive loss as pre-training objective and impose position randomization on the input to the teacher. SPIRAL achieves competitive or better results compared to state-of-the-art speech pre-training method wav2vec 2.0, with significant reduction of training cost (80% for BASE model, 65% for LARGE model). Furthermore, we address the problem of noise-robustness that is critical to real-world speech applications. We propose multi-condition pre-training by perturbing the student's input with various types of additive noise. We demonstrate that multi-condition pre-trained SPIRAL models are more robust to noisy speech (9.0% - 13.3% relative word error rate reduction on real noisy test data), compared to applying multi-condition training solely in the fine-tuning stage. Source code is available at https://github.com/huawei-noah/Speech-Backbones/tree/main/SPIRAL.

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。

↑