arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.06426cs.SDcs.LGeess.AS
收藏

文本相关说话人确认中深度表示学习的训练目标和激活函数研究

On Training Targets and Activation Functions for Deep Representation Learning in Text-Dependent Speaker Verification

Indian Institute of Information Technology, Sri City(印度信息技术学院,斯里城);Aalborg University(奥尔堡大学);Pioneer Centre for AI(人工智能先锋中心)

Achintya kr. Sarkar, Zheng-Hua Tan

摘要:

深度表示学习在推进文本相关说话人确认(TD-SV)系统中获得了显著的动力。在设计用于提取瓶颈特征的深度神经网络(DNN)时,关键考虑因素包括训练目标、激活函数和损失函数。在本文中,我们系统地研究了这些选择对TD-SV性能的影响。对于训练目标,我们考虑了说话人身份,时间对比学习(TCL)和自回归预测编码,其中第一个是监督的,最后两个是自监督的。此外,我们研究了一系列的损失函数时,说话人身份作为训练目标。关于激活函数,我们研究了广泛使用的sigmoid函数,校正线性单元(ReLU)和高斯误差线性单元(GELU)。我们的实验表明,GELU是能够降低出错率的TD-SV显着相比,S形,无论训练目标。在三个培训目标中,TCL表现最好。在各种损失函数中,交叉熵,联合softmax和焦点损失函数优于其他损失函数。最后,不同系统的分数级融合也能够降低出错率。实验在RedDots 2016挑战数据库上进行,用于使用短话语的TD-SV。对于说话人分类,著名的高斯混合模型-通用背景模型(GMM-UBM)和i-向量技术被使用。

英文摘要:

Deep representation learning has gained significant momentum in advancing text-dependent speaker verification (TD-SV) systems. When designing deep neural networks (DNN) for extracting bottleneck features, key considerations include training targets, activation functions, and loss functions. In this paper, we systematically study the impact of these choices on the performance of TD-SV. For training targets, we consider speaker identity, time-contrastive learning (TCL) and auto-regressive prediction coding with the first being supervised and the last two being self-supervised. Furthermore, we study a range of loss functions when speaker identity is used as the training target. With regard to activation functions, we study the widely used sigmoid function, rectified linear unit (ReLU), and Gaussian error linear unit (GELU). We experimentally show that GELU is able to reduce the error rates of TD-SV significantly compared to sigmoid, irrespective of training target. Among the three training targets, TCL performs the best. Among the various loss functions, cross entropy, joint-softmax and focal loss functions outperform the others. Finally, score-level fusion of different systems is also able to reduce the error rates. Experiments are conducted on the RedDots 2016 challenge database for TD-SV using short utterances. For the speaker classifications, the well-known Gaussian mixture model-universal background model (GMM-UBM) and i-vector techniques are used.

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。

↑