arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2303.13072cs.SDcs.CLeess.AS
收藏

超越 Universal Transformer:在用于自动语音识别的 Transformer 中使用适配器进行块重用

Beyond Universal Transformer: block reusing with adaptor in Transformer for automatic speech recognition

NTNU(挪威科技大学);KU Leuven(鲁汶大学);National Institute of Informatics(国立信息学研究所);Zhejiang University(浙江大学)

Haoyu Tang, Zhaoyi Liu, Chang Zeng, Xinfeng Li

AI总结:

针对边缘设备ASR模型参数量大的问题,提出块重用策略BRST与适配器模块ADM,在AISHELL-1上仅用7.6M/8.3M参数即实现9.3%/6.63%的CER,有效平衡了资源限制与识别准确率。

摘要:

基于 Transformer 的模型最近在端到端(E2E)自动语音识别(ASR)的应用中取得了显著成就。借助基于 Transformer 的模型,将 E2E ASR 系统部署在智能设备上成为可能。然而这些模型仍然存在需要大量模型参数的缺点。为了克服 Universal Transformer 模型在边缘设备上应用 ASR 的缺陷,我们提出了一种可以在 Transformer 模型中重用块的解决方案,适用于小占用空间的 ASR 系统,这满足了在不损害识别准确率的情况下适应资源限制的目标。具体来说,我们为语音 Transformer(BRST)设计了一种新颖的块重用策略以提高参数的有效性,并提出了一种适配器模块(ADM),该模块在每次重用块时仅伴随少量额外的可训练参数,即可生成紧凑且适应性强的模型。我们在公开的 AISHELL-1 语料库上使用所提出的方法进行了实验,结果表明,该方法在无 ADM 和有 ADM 的情况下,分别仅使用 7.6M/8.3M 参数,就实现了 9.3%/6.63% 的字符错误率(CER)。此外,我们还进行了更深入的分析,以展示 ADM 在通用块重用方法中的作用。

英文摘要:

Transformer-based models have recently made significant achievements in the application of end-to-end (E2E) automatic speech recognition (ASR). It is possible to deploy the E2E ASR system on smart devices with the help of Transformer-based models. While these models still have the disadvantage of requiring a large number of model parameters. To overcome the drawback of universal Transformer models for the application of ASR on edge devices, we propose a solution that can reuse the block in Transformer models for the occasion of the small footprint ASR system, which meets the objective of accommodating resource limitations without compromising recognition accuracy. Specifically, we design a novel block-reusing strategy for speech Transformer (BRST) to enhance the effectiveness of parameters and propose an adapter module (ADM) that can produce a compact and adaptable model with only a few additional trainable parameters accompanying each reusing block. We conducted an experiment with the proposed method on the public AISHELL-1 corpus, and the results show that the proposed approach achieves the character error rate (CER) of 9.3%/6.63% with only 7.6M/8.3M parameters without and with the ADM, respectively. In addition, we also make a deeper analysis to show the effect of ADM in the general block-reusing method.

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。

↑