arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.16614eess.AScs.SD
收藏

基于时域GANs的联合域自适应和语音带宽扩展的说话人确认

Joint domain adaptation and speech bandwidth extension using time-domain GANs for speaker verification

Center for Language and Speech Processing(语言与语音处理中心);Human Language Technology Center of Excellence(卓越人类语言技术中心);Johns Hopkins University(约翰斯·霍普金斯大学)

Saurabh Kataria, Jesús Villalba, Laureano Moro-Velázquez, Najim Dehak

摘要:

为特定的声学域和采样频率选择而开发的语音系统不容易转换为其他系统。通常的做法是独立地学习域自适应和带宽扩展模型。与此相反,我们建议一起学习这两项任务。特别地,我们学习将窄带对话电话语音映射到宽带麦克风语音。我们开发了并行和非并行学习解决方案,利用配对和非配对数据。首先,我们首先讨论针对我们的任务的多个生成模型的联合和不联合训练。然后,我们提出了一个两阶段的学习解决方案,在带宽扩展训练中使用预训练的域自适应系统进行预处理。我们评估了我们的计划上的发言人验证下游任务。我们使用JHU-MIT的NIST SRE 21实验装置,包括SRE 16,SRE-CTS超集和SRE 21。我们的研究结果提供了第一个证据,证明学习两项任务比只学习一项任务更好。在SRE 16上,我们最好的系统在等错误率方面实现了22%的相对改进。直接学习基线和8%的w.r.t.强大的带宽扩展系统。

英文摘要:

Speech systems developed for a particular choice of acoustic domain and sampling frequency do not translate easily to others. The usual practice is to learn domain adaptation and bandwidth extension models independently. Contrary to this, we propose to learn both tasks together. Particularly, we learn to map narrowband conversational telephone speech to wideband microphone speech. We developed parallel and non-parallel learning solutions which utilize both paired and unpaired data. First, we first discuss joint and disjoint training of multiple generative models for our tasks. Then, we propose a two-stage learning solution where we use a pre-trained domain adaptation system for pre-processing in bandwidth extension training. We evaluated our schemes on a Speaker Verification downstream task. We used the JHU-MIT experimental setup for NIST SRE21, which comprises SRE16, SRE-CTS Superset and SRE21. Our results provide the first evidence that learning both tasks is better than learning just one. On SRE16, our best system achieves 22% relative improvement in Equal Error Rate w.r.t. a direct learning baseline and 8% w.r.t. a strong bandwidth expansion system.

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。

↑