在使用anaconda装python时,有时候路径可能没有加入到系统路径中,要手动加入:
#将anaconda的bin目录加入PATH,根据版本不同。确定anaconda的名字,如果是默认的,python3.7生成anaconda3
$ echo 'export PATH="~/anaconda3/bin:$PATH"'>>~/.bashrc
#然后更新bashrc,即可立即生效,不用关机,source ~/.bashrc用来重启环境变量。
$ source ~/.bashrc
conda create -n env_name python=3.7
conda install pytorch torchvision cudatoolkit=10.0 -c pytorch
(一)删除整个anaconda目录:
由于Anaconda的安装文件都包含在一个目录中,所以直接将该目录删除即可。到包含整个anaconda目录的文件夹下,删除整个Anaconda目录:
rm -rf anaconda文件夹名
(二)建议清理下.bashrc中的Anaconda路径:
1.到根目录下,打开终端并输入:
sudo gedit ~/.bashrc
2.在.bashrc文件末尾用#号注释掉之前添加的路径(或直接删除):
#export PATH=/home/lq/anaconda3/bin:$PATH
保存并关闭文件
3.使其立即生效,在终端执行:
source ~/.bashrc
4.关闭终端,然后再重启一个新的终端,这一步很重要,不然在原终端上还是绑定有anaconda.
sudo pacman -S jupyter
无法从 mirrors6.tuna.tsinghua.edu.cn : The requested URL returned error: 4 04 获取文件 'mathjax-2.7.6-1-any.pkg.tar.xz'
小龙:
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
3 - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
4 - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
5 - https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
6 - https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/
7 - defaults
星辰:
宋星辰:
把-c去掉就行,-channel
宋星辰:
强制用pytorch的源安装pytorch
github代码地址: https://github.com/r9y9/wavenet_vocoder
2 I try lots of vocoders, some of them are too slow, another of them can not produce good quality voice.
100%|██████████| 133376/133376 [26:05<00:00, 85.20it/s]
github代码地址: https://github.com/fatchord/WaveRNN
colab代码地址: 自己写一个, 先不用这个版本, 期待WaveRNN能实时, 目前研究先用T2+WaveNet版本吧.
总结:
Sentence: I try lots of vocoders, some of them are too slow, another of them can not produce good quality voice.
approximate time of sentence: 0.06s
Tacotron2-model1__WaveNet:
Tacotron2-model1__Mel:
Tacotron1-model2__WaveRNN-u:
Tacotron1-model2__WaveRNN-b:
Tacotron2-model3__Linear:
Tacotron2-model3__Mel:
Tacotron2-model4__Mel:
model1, model2: pre-trained model, 180k steps.
model3: 100ksteps, trained with the latest version code.
model4: 80ksteps, trained with 2018-10 version code.