Praatの音声アノテーション（.TextGrid）を自動生成 ref: https://qiita.com/Syuparn/items/d86e77c39527539b16c5

file0.txt

$ julius --version

file10.txt

$ python convert_label.py path/to/dir

file11.txt

change segmentation unit to mora? (default:phoneme) y/n:

file4.txt

$ file -i sample.txt
sample.txt: text/plain; charset=utf-8

file5.txt

$ perl segment_julius.pl path/to/dir

file7.txt

$ python segment_recursively.py dir/to/path

file8.txt

$ python convert_label.py path/to/dir

file9.txt

change segmentation unit to mora? (default:phoneme) y/n:

hoge.lab

0.0000000 0.1850000 silB
0.1850000 0.2745000 h
0.2745000 0.3870000 o
0.3870000 0.4455000 g
0.4455000 0.6415000 e
0.6410000 0.7097500 silE

hoge.TextGrid

File type = "ooTextFile"
Object class = "TextGrid"

xmin = 0 
xmax = 0.7097500 
tiers? <exists> 
size = 1 
item []: 
    item [1]: 
        class = "IntervalTier" 
        name = "phoneme" 
        xmin = 0 
        xmax = 0.7097500 
        intervals: size = 6 
        intervals [1]:
            xmin = 0.0000000 
            xmax = 0.1850000 
            text = "" 
        intervals [2]:
            xmin = 0.1850000 
            xmax = 0.2745000 
            text = "h" 
        intervals [3]:
            xmin = 0.2745000 
            xmax = 0.3870000 
            text = "o" 
        intervals [4]:
            xmin = 0.3870000 
            xmax = 0.4455000 
            text = "g" 
        intervals [5]:
            xmin = 0.4455000 
            xmax = 0.6415000 
            text = "e" 
        intervals [6]:
            xmin = 0.6410000 
            xmax = 0.7097500 
            text = "" 

sample.txt

きょーわいいてんきだ

segment_julius.pl

## data directory                                                                                                            
if ($ARGV[0]){
    $datadir = $ARGV[0];
} else {
    $datadir = "./wav";
}

segment_recursively.py

import os
import subprocess
import sys

args = sys.argv

for dirpath, dirnames, filenames in os.walk(args[1]):
    subprocess.call(['perl', 'segment_julius.pl', dirpath])