Le-Xiaohuai-speech / DPCRN_DNS3

Implementation of paper "DPCRN: Dual-Path Convolution Recurrent Network for Single Channel Speech Enhancement"
180 stars 40 forks source link

请教关于real time processing时BatchNormalization问题 #35

Open yyd19948 opened 1 year ago

yyd19948 commented 1 year ago

您好,请教一下real time processing时BatchNormalization,LayerNormalization是保存历史的数据,用之前的多帧数据计算的,还是怎么处理的

Le-Xiaohuai-speech commented 1 year ago

帧独立

---原始邮件--- 发件人: @.> 发送时间: 2022年11月9日(周三) 下午3:17 收件人: @.>; 抄送: @.***>; 主题: [Le-Xiaohuai-speech/DPCRN_DNS3] 请教关于real time processing时BatchNormalization问题 (Issue #35)

请教一下real time processing时BatchNormalization,LayerNormalization是保存历史的数据,用之前的多帧数据计算的,还是怎么处理的

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

yyd19948 commented 1 year ago

帧独立 ---原始邮件--- 发件人: @.> 发送时间: 2022年11月9日(周三) 下午3:17 收件人: @.>; 抄送: @.>; 主题: [Le-Xiaohuai-speech/DPCRN_DNS3] 请教关于real time processing时BatchNormalization问题 (Issue #35) 请教一下real time processing时BatchNormalization,LayerNormalization是保存历史的数据,用之前的多帧数据计算的,还是怎么处理的 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.>

这样会对输出有啥影响吗,我看训练的时候是一段数据来做归一化?

yyd19948 commented 1 year ago

这样会对输出影响大吗,我看训练的时候是一段数据来做归一化?

------------------ 原始邮件 ------------------ 发件人: "Le-Xiaohuai-speech/DPCRN_DNS3" @.>; 发送时间: 2022年11月9日(星期三) 下午3:20 @.>; @.**@.>; 主题: Re: [Le-Xiaohuai-speech/DPCRN_DNS3] 请教关于real time processing时BatchNormalization问题 (Issue #35)

帧独立

---原始邮件--- 发件人: @.> 发送时间: 2022年11月9日(周三) 下午3:17 收件人: @.>; 抄送: @.***>; 主题: [Le-Xiaohuai-speech/DPCRN_DNS3] 请教关于real time processing时BatchNormalization问题 (Issue #35)

请教一下real time processing时BatchNormalization,LayerNormalization是保存历史的数据,用之前的多帧数据计算的,还是怎么处理的

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.> — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.>

Le-Xiaohuai-speech commented 1 year ago

没有影响,因为是在channel维度上。你可以对比一下整体输出和一帧一帧输出的结果

---原始邮件--- 发件人: @.> 发送时间: 2022年11月9日(周三) 下午3:54 收件人: @.>; 抄送: @.**@.>; 主题: Re: [Le-Xiaohuai-speech/DPCRN_DNS3] 请教关于real time processing时BatchNormalization问题 (Issue #35)

这样会对输出影响大吗,我看训练的时候是一段数据来做归一化?

------------------ 原始邮件 ------------------ 发件人: "Le-Xiaohuai-speech/DPCRN_DNS3" @.>; 发送时间: 2022年11月9日(星期三) 下午3:20 @.>; @.**@.>; 主题: Re: [Le-Xiaohuai-speech/DPCRN_DNS3] 请教关于real time processing时BatchNormalization问题 (Issue #35)

帧独立

---原始邮件---
发件人: @.>
发送时间: 2022年11月9日(周三) 下午3:17
收件人:
@.>;
抄送: @.***>;
主题: [Le-Xiaohuai-speech/DPCRN_DNS3] 请教关于real time processing时BatchNormalization问题 (Issue #35)

请教一下real time processing时BatchNormalization,LayerNormalization是保存历史的数据,用之前的多帧数据计算的,还是怎么处理的


Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.Message ID: @.>
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID:
@.> — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>