arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.13581eess.AScs.AI
收藏

EfficientTDNN:一种高效的说话人识别搜索结构

EfficientTDNN: Efficient Architecture Search for Speaker Recognition

Tongji University(同济大学);Durham University(杜伦大学);Zhejiang University(浙江大学);Zhejiang University of Technology(浙江工业大学)

Rui Wang, Zhihua Wei, Haoran Duan, Shouling Ji, Yang Long, Zhen Hong

摘要:

卷积神经网络(CNN),如时间延迟神经网络(TDNN),在学习说话人嵌入方面表现出显着的能力。然而,它们同时在存储大小、处理和内存方面带来了巨大的计算成本。发现满足特定约束的专用CNN需要人类专家的大量努力。与手工设计方法相比,神经结构搜索(NAS)作为一种自动化人工结构设计过程的实用技术出现,并在诸如说话人识别的口语处理任务中引起了越来越多的兴趣。在本文中,我们提出了有效的TDNN,一个高效的架构搜索框架,包括一个基于TDNN的超网和TDNN-NAS算法。建议的超网引入时间卷积的不同范围的感受野和功能聚合的各种分辨率从不同的层TDNN。最重要的是,TDNN-NAS算法通过权重共享搜索快速搜索所需的TDNN架构,这在处理具有各种资源需求的大量设备的同时,令人惊讶地减少了计算。在VoxCeleb数据集上的实验结果表明,所提出的EfficientTDNN能够实现关于深度、内核和宽度的近似$10^{13}$架构。在不同的计算约束条件下,该算法在204 M乘加运算时的等误率为2.20%,在571 M乘加运算时的等误率为1.41%,在1.45G乘加运算时的等误率为0.94%。全面的调查表明,经过训练的超网概括了训练期间未采样的子网,并在准确性和效率之间取得了有利的权衡。

英文摘要:

Convolutional neural networks (CNNs), such as the time-delay neural network (TDNN), have shown their remarkable capability in learning speaker embedding. However, they meanwhile bring a huge computational cost in storage size, processing, and memory. Discovering the specialized CNN that meets a specific constraint requires a substantial effort of human experts. Compared with hand-designed approaches, neural architecture search (NAS) appears as a practical technique in automating the manual architecture design process and has attracted increasing interest in spoken language processing tasks such as speaker recognition. In this paper, we propose EfficientTDNN, an efficient architecture search framework consisting of a TDNN-based supernet and a TDNN-NAS algorithm. The proposed supernet introduces temporal convolution of different ranges of the receptive field and feature aggregation of various resolutions from different layers to TDNN. On top of it, the TDNN-NAS algorithm quickly searches for the desired TDNN architecture via weight-sharing subnets, which surprisingly reduces computation while handling the vast number of devices with various resources requirements. Experimental results on the VoxCeleb dataset show the proposed EfficientTDNN enables approximate $10^{13}$ architectures concerning depth, kernel, and width. Considering different computation constraints, it achieves a 2.20% equal error rate (EER) with 204M multiply-accumulate operations (MACs), 1.41% EER with 571M MACs as well as 0.94% EER with 1.45G MACs. Comprehensive investigations suggest that the trained supernet generalizes subnets not sampled during training and obtains a favorable trade-off between accuracy and efficiency.

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。

↑