search

JusperLee / Dual-Path-RNN-Pytorch

Dual-path RNN: efficient long sequence modeling for time-domain single-channel speech separation implemented by Pytorch
Apache License 2.0
417 stars 65 forks source link

PESQ #24

Closed fanmanqian closed 3 years ago

fanmanqian commented 3 years ago

作者，您好，Dual path的效果SDR很高13.多，听的也很清楚，但是计算pesq和stoi值的时候很低，pesq只有1.9左右，这个可能是什么问题呢？

JusperLee commented 3 years ago

pesq和stoi属于听感上的评价指标，因为这些模型生成的音频的高频都有所缺失，所以在变化的时候都会变的较差。

fanmanqian commented 3 years ago

好的，谢谢您～

---Original--- From: "Kai Li (李凯)"<notifications@github.com> Date: 2020/11/16 21:02:55 To: "JusperLee/Dual-Path-RNN-Pytorch"<Dual-Path-RNN-Pytorch@noreply.github.com>; Cc: "fanmanqian"<2563965485@qq.com>;"Author"<author@noreply.github.com>; Subject: Re: [JusperLee/Dual-Path-RNN-Pytorch] PESQ (#24)

pesq和stoi属于听感上的评价指标，因为这些模型生成的音频的高频都有所缺失，所以在变化的时候都会变的较差。

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.