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 2003.06894eess.AScs.CLcs.SD
收藏

深度神经网络声学模型的说话人自适应高斯混合模型框架

Exploring Gaussian mixture model framework for speaker adaptation of deep neural network acoustic models

University of Avignon(阿维尼翁大学);STC-innovations Ltd(STC创新有限公司)

Natalia Tomashenko, Yuri Khokhlov, Yannick Esteve

摘要:

在本文中,我们研究了用于深度神经网络(DNN)声学模型自适应的GMM衍生(GMMD)特征。在GMMD特征上训练的DNN的适应是通过用于GMMD特征提取的辅助GMM模型的最大后验(MAP)适应来完成的。我们探讨融合的适应GMMD功能与传统的功能,如瓶颈和MFCC功能,在两个不同的神经网络架构:DNN和时间延迟神经网络(TDNN)。我们分析和比较不同类型的自适应技术,如i-向量和特征空间自适应技术的基础上最大似然线性回归(fMLLR)与建议的适应方法,并探讨其互补性,使用各种类型的融合,如特征级,后验级,格级和其他,以发现最佳的组合方式。在TED-LIUM语料库上的实验结果表明,所提出的自适应技术可以有效地集成到不同级别的DNN和TDNN设置中,并在识别性能方面提供额外的增益:相对于基于最大似然线性回归(fMLLR)说话者自适应DNN基线的强特征空间自适应技术,相对字错误率降低(WERR)高达6%,与在传统特征上训练的说话人独立(SI)DNN基线模型相比,相对WERR高达18%。对于TDNN模型,与SI基线相比,所提出的方法实现了高达26%的相对WERR,并且与通过使用i向量调整的模型相比,该方法实现了13%的相对WERR。从不同的角度对适应的GMMD特征进行分析,证明了它们在不同层次上的有效性。

英文摘要:

In this paper we investigate the GMM-derived (GMMD) features for adaptation of deep neural network (DNN) acoustic models. The adaptation of the DNN trained on GMMD features is done through the maximum a posteriori (MAP) adaptation of the auxiliary GMM model used for GMMD feature extraction. We explore fusion of the adapted GMMD features with conventional features, such as bottleneck and MFCC features, in two different neural network architectures: DNN and time-delay neural network (TDNN). We analyze and compare different types of adaptation techniques such as i-vectors and feature-space adaptation techniques based on maximum likelihood linear regression (fMLLR) with the proposed adaptation approach, and explore their complementarity using various types of fusion such as feature level, posterior level, lattice level and others in order to discover the best possible way of combination. Experimental results on the TED-LIUM corpus show that the proposed adaptation technique can be effectively integrated into DNN and TDNN setups at different levels and provide additional gain in recognition performance: up to 6% of relative word error rate reduction (WERR) over the strong feature-space adaptation techniques based on maximum likelihood linear regression (fMLLR) speaker adapted DNN baseline, and up to 18% of relative WERR in comparison with a speaker independent (SI) DNN baseline model, trained on conventional features. For TDNN models the proposed approach achieves up to 26% of relative WERR in comparison with a SI baseline, and up 13% in comparison with the model adapted by using i-vectors. The analysis of the adapted GMMD features from various points of view demonstrates their effectiveness at different levels.

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。

↑