arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.02680eess.AScs.SDeess.SP
收藏

语言变化与说话人变化:一项比较研究

Language vs Speaker Change: A Comparative Study

Jagabandhu Mishra, S. R. Mahadeva Prasanna

AI总结:

本研究对比口语语言变化检测(LCD)与说话人变化检测(SCD)任务,发现LCD需更长时频谱信息,进而探索两类自动LCD方法,验证了增加时频谱时长可提升LCD性能。

摘要:

口语语言变化检测(LCD)指在多语言语音信号中检测语言切换点。说话人变化检测(SCD)指在多说话人语音信号中定位说话人变化点。本研究的目标是通过将LCD任务与SCD任务对比,明确LCD任务的难点。研究针对LCD和SCD开展了人类主观变化检测实验,结果表明,与SCD相比,LCD需要变化点周围更长时长的时频谱信息。基于此,本研究探索了基于距离和基于模型的两类自动LCD方法,其中基于模型的方法包括高斯混合模型与通用背景模型(GMM-UBM)、注意力机制以及生成对抗网络(GAN)相关方法。人类主观实验和自动LCD任务均表明,纳入越多时长的时频谱信息,LCD任务的性能提升越显著。

英文摘要:

Spoken language change detection (LCD) refers to detecting language switching points in a multilingual speech signal. Speaker change detection (SCD) refers to locating the speaker change points in a multispeaker speech signal. The objective of this work is to understand the challenges in LCD task by comparing it with SCD task. Human subjective study for change detection is performed for LCD and SCD. This study demonstrates that LCD requires larger duration spectro-temporal information around the change point compared to SCD. Based on this, the work explores automatic distance based and model based LCD approaches. The model based ones include Gaussian mixture model and universal background model (GMM-UBM), attention, and Generative adversarial network (GAN) based approaches. Both the human and automatic LCD tasks infer that the performance of the LCD task improves by incorporating more and more spectro-temporal duration.

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。

↑