用于众包语音可懂度实验的有效数据筛选技术:基于IRM的语音增强评测
声音
2023-07-27 v2 音频与语音处理
摘要
为评估客观可懂度度量以开发有效的语音增强与降噪算法,开展由人类听者参与的语音可懂度(SI)实验至关重要。近年来,众包远程测试因能以较小成本在短时间内收集大量多样数据而流行。然而,为获得可靠SI数据,细致的数据筛选必不可少。我们在受控良好的实验室环境与无法直接控制的众包远程环境中,对由“神谕”理想比率掩码(IRM)增强的语音进行了SI实验。我们引入了简单的音调脉冲测试,要求参与者报告可听音调脉冲的数量,以估计其高于听阈的聆听水平。音调脉冲测试对数据筛选极为有效,可降低众包远程结果的变异性,使之接近实验室结果。结果还展示了神谕IRM的SI,给出了基于掩码的单通道语音增强的上限。
引用
@article{arxiv.2203.16760,
title = {Effective data screening technique for crowdsourced speech intelligibility experiments: Evaluation with IRM-based speech enhancement},
author = {Ayako Yamamoto and Toshio Irino and Shoko Araki and Kenichi Arai and Atsunori Ogawa and Keisuke Kinoshita and Tomohiro Nakatani},
journal= {arXiv preprint arXiv:2203.16760},
year = {2023}
}
备注
This paper was submitted to APSIPA ASC 2022 (https://www.apsipa2022.org). The original title [v1] was "Subjective intelligibility of speech sounds enhanced by ideal ratio mask via crowdsourced remote experiments with effective data screening."