#!/bin/bash

# Copyright 2018 Jarvan Wang Apache 2.0.

# Copyright 2016 Xiaohui Zhang Apache 2.0.

 
 

# This srcipt operates on lattice directories, such as exp/tri4a_lats

# the output is a new lat dir which has lattices from all the input lattice dirs

 
 

# Begin configuration section.

cmd=run.pl

extra_files=

num_jobs=4

# End configuration section.

echo "$0 $@" # Print the command line for logging

 
 

if [ -f path.sh ]; then . ./path.sh; fi

. parse_options.sh || exit 1;

 
 

if [[ $# -lt 3 ]]; then

echo "Usage: $0 [options] <data> <dest-lat-dir> <src-lat-dir1> <src-lat-dir2> ..."

echo "e.g.: $0 --num-jobs 32 data/train exp/tri3_lat_combined exp/tri3_lat_1 exp_tri3_lat_2"

echo "Options:"

echo " --extra-files <file1 file2...> # specify addtional files in 'src-lat-dir1' to copy"

echo " --num-jobs <nj> # number of jobs used to split the data directory."

echo " Note, files that don't appear in the first source dir will not be added even if they appear in later ones."

echo " Other than lattices, only files from the first src lat dir are copied."

exit 1;

fi

 
 

data=$1;

shift;

dest=$1;

shift;

first_src=$1;

 
 

mkdir -p $dest;

rm $dest/{lat.*.gz,num_jobs} 2>/dev/null

 
 

cp $first_src/phones.txt $dest 2>/dev/null

 
 

export LC_ALL=C

 
 

for dir in $*; do

if [ ! -f $dir/lat.1.gz ]; then

echo "$0: check if lattices (lat.*.gz) are present in $dir."

exit 1;

fi

done

 
 

for dir in $*; do

for f in tree; do

diff $first_src/$f $dir/$f 1>/dev/null 2>&1

if [ $? -ne 0 ]; then

echo "$0: Cannot combine lattice directories with different $f files."

fi

done

done

 
 

for f in final.mdl tree cmvn_opts num_jobs $extra_files; do

if [ ! -f $first_src/$f ]; then

echo "combine_lat_dir.sh: no such file $first_src/$f"

exit 1;

fi

cp $first_src/$f $dest/

done

 
 

src_id=0

temp_dir=$dest/temp

[ -d $temp_dir ] && rm -r $temp_dir;

mkdir -p $temp_dir

echo "$0: dumping lattices in each source directory as single archive and index."

for dir in $*; do

src_id=$((src_id + 1))

cur_num_jobs=$(cat $dir/num_jobs) || exit 1;

lats=$(for n in $(seq $cur_num_jobs); do echo -n "$dir/lat.$n.gz "; done)

$cmd $dir/log/copy_lattices.log \

lattice-copy "ark:gunzip -c $lats|" \

ark,scp:$temp_dir/lat.$src_id.ark,$temp_dir/lat.$src_id.scp || exit 1;

done

sort -m $temp_dir/lat.*.scp > $temp_dir/lat.scp || exit 1;

 
 

echo "$0: splitting data to get reference utt2spk for individual lat.JOB.gz files."

utils/split_data.sh $data $num_jobs || exit 1;

 
 

echo "$0: splitting the lattices to appropriate chunks according to the reference utt2spk files."

utils/filter_scps.pl JOB=1:$num_jobs \

$data/split$num_jobs/JOB/utt2spk $temp_dir/lat.scp $temp_dir/lat.JOB.scp

 
 

for i in `seq 1 $num_jobs`; do

lattice-copy scp:$temp_dir/lat.${i}.scp "ark:|gzip -c >$dest/lat.$i.gz" || exit 1;

done

 
 

echo $num_jobs > $dest/num_jobs || exit 1

 
 

echo "$0: checking the lattice files generated have at least 90% of the utterances."

for i in `seq 1 $num_jobs`; do

num_lines=`cat $temp_dir/lat.$i.scp | wc -l` || exit 1;

num_lines_tot=`cat $data/split$num_jobs/$i/utt2spk | wc -l` || exit 1;

python -c "import sys;

percent = 100.0 * float($num_lines) / $num_lines_tot

if percent < 90 :

print ('$dest/lat.$i.gz {0}% utterances missing.'.format(percent))" || exit 1;

done

rm -r $temp_dir 2>/dev/null

 
 

echo "Combined lattices and stored in $dest"

exit 0

combine_lat_dirs.sh的更多相关文章

  1. 数据库备份并分离日志表(按月)sh 脚本

    #!/bin/sh year=`date +%Y` month=`date +%m` day=`date +%d` hour=`date +%H` dir="/data/dbbackup/f ...

  2. Linux下如何查找.sh后缀的文件

    find / -name *.sh或locate *.shfind 与locate的区别:locate 配合数据库查看文件位置 find 实际搜寻硬盘查询文件名称

  3. sh 测试网段在线主机

    yum install nmap nmap -sP  192.168.21.1/24 查看网段在线主机 grep -vwf file1 file2 文件内容比较 #!/bin/bash # day=` ...

  4. Maven打包生成可运行bat/sh脚本文件

    利用Maven的appassembler-maven-plugin插件,就可以实现自动打包可运行的脚本,还可以跨平台.   <plugin>      <groupId>org ...

  5. 记一个简单的保护if 的sh脚本

    真是坑爹,就下面的sh,竟然也写了很久! if [ `pwd` != '/usr/xx/bin/tomcat' ] then echo "rstall is not allowed in c ...

  6. 主机巡检脚本:OSWatcher.sh

    主机巡检脚本:OSWatcher.sh 2016-09-26更新,目前该脚本只支持Linux操作系统,后续有需求可以继续完善. 注意: 经测试,普通用户执行脚本可以顺利执行前9项检查: 第10项,普通 ...

  7. Oracle巡检脚本:ORAWatcher.sh

    Oracle巡检脚本:ORAWatcher.sh #!/usr/bin/ksh echo "" echo "ORAWatcher Version:1.0.1" ...

  8. 解决:/bin/sh: 1: /home/**/custom_app.sh: Permission denied错误

    出现如下错误,一般是执行权限不够. /bin/sh: : /home/custom_app.sh: Permission denied 解决方法是:cd 到此文件目录,对提示的文件赋予可执行权限或读写 ...

  9. SH Script Grammar

    http://linux.about.com/library/cmd/blcmdl1_sh.htm http://pubs.opengroup.org/onlinepubs/9699919799/ut ...

随机推荐

  1. CQOI2019(十二省联考)游记

    CQOI2019(十二省联考)游记 Day -? 自从联赛爆炸,\(THUWC\)爆炸,\(WC\)爆炸(就没有不爆炸的)之后我已经无所畏惧... 听说是考\(4.5 h\)吗? Day -1 \(Z ...

  2. 解决consul覆盖注册

    默认注册consul的服务id为服务名-端口号,相同的服务名和端口号注册会覆盖 解决方式: 1.自定义Consul注册Id import com.ecwid.consul.v1.ConsulClien ...

  3. day 25 面向对象之接口、抽象类、多态、异常处理、反射、断言

    复习 '''继承​1.父类:在类后()中写父类们class A:passclass B:passclass C(A, B):pass​2.属性查找顺序:自己 -> ()左侧的父类 -> 依 ...

  4. 第一次java作业

    学习了c语言半个学期,好像刚开始并不担心Java,因为Java也是类同于C语言的一种语言,可是Java学到现在也将近一个月,在我脑子里现在却还没有一个框架,有点着急了.

  5. 最新:百度春节抢百万游戏--汤圆向前冲--辅助工具v1.0.0.2

    https://www.cnblogs.com/Charltsing/p/ADBJumpTY.html 联系QQ:564955427 本程序为Windows版,不要在手机里面打开. 汤圆向前冲辅助工具 ...

  6. Docker 核心技术之数据管理

    Docker 数据卷简介 为什么用数据卷 宿主机无法直接访问容器中的文件 容器中的文件没有持久化,导致容器删除后,文件数据也随之消失 容器之间也无法直接访问互相的文件 为解决这些问题,docker加入 ...

  7. ISOMAP

    转载 https://blog.csdn.net/dark_scope/article/details/53229427# 维度打击,机器学习中的降维算法:ISOMAP & MDS 降维是机器 ...

  8. MongoDB ODBC Driver for Data Integration with Power BI

    This guide will walk you through connecting Microsoft Power BI to a MongoDB DataSet using our MongoD ...

  9. ZuulFilter 执行顺序

    说明: 创建了两个Filter,分别是 PreFilter public class PreFilter extends ZuulFilter { public PreFilter() { super ...

  10. MAVEN打包同时引入本地jar包

    方法一(pom文件指定jar包目录进行引入) 1.将需要手动引入的包放在项目目录下,如lib目录下: 修改pom文件,引入依赖并且将scope设置为system 2.同时配置maven打包插件 方法二 ...