Building wav2letter++ 安装依赖

安装MKL

• 到官网下载Intel® Math Kernel Library for Linux安装包
• 采用tar -xvzf l_mkl_xxx.tgz命令解压安装包，进入解压后的安装包，运行./install.sh脚本执行安装过程
• 根据提示一路接受，默认安装路径是/opt/intel
• 在目录/opt/intel/mkl/bin下执行./mklvars.sh intel64
• 编辑vi ~/.bashrc设置环境变量
• 在末尾加入代码：
  export PATH=/opt/intel/bin:$PATH
export LD_LIBRARY_PATH= /opt/intel/lib/intel64:/opt/intel/mkl/lib/intel64:$LD_LIBRARY_PATH
• $ source ~/.bashrc 如果没有报错，则path生效

安装Boost

• wget https://dl.bintray.com/boostorg/release/1.64.0/source/boost_1_64_0.tar.gz
• tar zxvf boost_1_64_0.tar.gz
• cd boost_1_64_0
• ./bootstrap.sh
• ./b2 install

安装kenlm

安装kenlm之前需要安装好boost，xz，zlib，bzip，libbz2-dev等依赖。
它们的安装参考博客：https://blog.csdn.net/sinat_33741547/article/details/79996742

• wget https://kheafield.com/code/kenlm.tar.gz
• tar -xvzf kenlm.tar.gz
• cd kenlm
• mkdir build && cd build
• cmake ..
• make -j 4
  这样还无法在python环境中执行
• cd kenlm
• python setup.py install
• python #进入python环境
• import kenlm

安装gflags

sudo apt-get install libgflags-dev

安装glog

• git clone https://github.com/google/glog
• sudo apt-get install autoconf automake libtool
• cd glog
• ./autogen.sh
• ./configure
• make -j 4
• make install

安装google test

• git clone https://github.com/google/googletest.git
• mkdir mybuild
• cd mybuild
• cmake ${GTEST_DIR} #google test的下载目录,比如我clone到了/data/xxx/googletest
• make
• make install

安装fftw

sudo apt-get install fftw3

安装libsndfile:

• git clone git://github.com/erikd/libsndfile.git
• sudo apt install autoconf autogen automake build-essential libasound2-dev \
libflac-dev libogg-dev libtool libvorbis-dev pkg-config python
• cd libsndfile
• ./autogen.sh
• ./configure --enable-werror
• make
• make check
  为了产生wav2letter++需要的.cmake配置文件
• mkdir CMakeBuild
• cd CMakeBuild
• cmake ..
• make
• sudo make install

安装ArrayFire

ArrayFire for Linux官方教程

• sudo apt-get install -y build-essential git cmake libfreeimage-dev
sudo apt-get install -y cmake-curses-gui
• sudo apt-get install libglfw3-dev libfontconfig1-dev libglm-dev
• git clone --recursive https://github.com/arrayfire/arrayfire.git
• cd arrayfire
• mkdir build && cd build
• cmake .. -DCMAKE_BUILD_TYPE=Release
• make -j 4
• make install

然而我build一直有问题，最后也没成功。于是采用方法二：

• 直接下载：ArrayFire-v3.6.1_Linux_x86_64.sh
• bash ArrayFire-v3.6.1_Linux_x86_64.sh
• 之后wav2letter++在cmake时加入选项:
  -DArrayFire_DIR=/xxx/arrayfire/share/ArrayFire/cmake

安装NCCL

到NVIDIA官网下载NCCL，选择对应的版本：

image.png

切换到NCCL文件所在目录，运行以下命令：

• sudo dpkg -i nccl-repo-ubuntu1604-2.3.7-ga-cuda9.2_1-1_amd64.deb
• sudo apt-get update
• sudo apt install libnccl2=2.3.7-1+cuda9.2 libnccl-dev=2.3.7-1+cuda9.2

安装flashlight

注意安装flashlight之前需要安装好arrayfire和nccl [flashlight还需要CUDA>=9.2，CUDNN>=7.2.1，所以我又去装了cuda9.2 :(]

• git clone https://github.com/facebookresearch/flashlight.git
• export MKLROOT=/opt/intel/mkl
• mkdir -p build #in the flashlight project directory:
• cd build
• cmake .. -DCMAKE_BUILD_TYPE=Release -DFLASHLIGHT_BACKEND=CUDA
  • 报错找不到ArrayFire，加上-D ArrayFire_DIR=/data/zd/arrayfire/share/ArrayFire/cmake选项；
  • 报错找不到CUDA，加上-DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.2选项；
• make -j 4 #(or any number of threads)
• make test
• make install
  发现报错

   CMake Error at cmake_install.cmake:36 (file):
     file INSTALL cannot make directory
    "/usr/local/include/flashlight/cereal/cereal": No such file or directory
  

  可能是创建不了文件夹，于是改成sudo make install成功。

安装wav2letter++

终于来到这一步了orz...

• git clone --recursive https://github.com/facebookresearch/wav2letter.git
• mkdir -p build
• cd build
• cmake .. -DCMAKE_BUILD_TYPE=Release -DCRITERION_BACKEND=CUDA
-DArrayFire_DIR=/data/zd/arrayfire/share/ArrayFire/cmake
-DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-9.2
• make -j4

cmake过程依然出错：

-- Looking for KenLM
CMake Error at cmake/Findkenlm.cmake:47 (message):
  kenlm library not found; please set CMAKE_LIBRARY_PATH or KENLM_LIB
Call Stack (most recent call first):
  src/decoder/CMakeLists.txt:5 (find_package)

-- Configuring incomplete, errors occurred!

需要在~/.bashrc中加上环境变量
export KENLM_ROOT_DIR=[your kenlm directory]
然后依然有错：

--   Library mkl: not found
CMake Error at cmake/FindMKL.cmake:265 (MESSAGE):
  MKL library not found.  Please specify library location
Call Stack (most recent call first):
  src/feature/CMakeLists.txt:5 (find_package)

-- Configuring incomplete, errors occurred!

继续在~/.bashrc中加上环境变量
export MKLROOT=/opt/intel/mkl
至此大功告成 :)。

参考文章：
https://www.cnblogs.com/venus024/p/5619373.html --MKL https://blog.csdn.net/u011641865/article/details/73498533 --Boost
https://blog.csdn.net/sinat_33741547/article/details/79996742 --kenlm
http://blog.topspeedsnail.com/archives/5462 --fftw

如果最后始终配置不成功，推荐使用Docker吧：-)