mirror of https://gitlab.com/christosangel/sapo
239 lines
10 KiB
Bash
Executable File
239 lines
10 KiB
Bash
Executable File
#! /bin/bash
|
||
# ▗▄▖
|
||
#▗▛▀▜
|
||
#▐▙ ▟██▖▐▙█▙ ▟█▙
|
||
# ▜█▙ ▘▄▟▌▐▛ ▜▌▐▛ ▜▌
|
||
# ▜▌▗█▀▜▌▐▌ ▐▌▐▌ ▐▌
|
||
#▐▄▄▟▘▐▙▄█▌▐█▄█▘▝█▄█▘
|
||
# ▀▀▘ ▀▀▝▘▐▌▀▘ ▝▀▘
|
||
# ▐▌
|
||
# a tts bash script by Christos Angelopoulos, February 2022
|
||
#converts txt to wav
|
||
#using the all powerful https://github.com/coqui-ai/TTS
|
||
|
||
FILE="$(yad --file-selection --filename=Desktop --height='400' --width='800' --title='Sapo - Select File to Read' --window-icon=$HOME/git/sapo/sapo.png --hscroll-policy=never --vscroll-policy=never)"
|
||
case $? in
|
||
0)
|
||
;;
|
||
1) exit
|
||
;;
|
||
esac
|
||
DIRECTORY=${FILE%/*}/
|
||
NAME=${FILE##*/}
|
||
echo "FILE : ""$FILE"
|
||
echo "DIRECTORY : " "$DIRECTORY"
|
||
echo "NAME : ""$NAME"
|
||
|
||
mkdir "$DIRECTORY""Sapo_""$NAME"/
|
||
|
||
##text delimitation to setences
|
||
echo $(cat "$FILE")>"$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # get rid of new lines
|
||
##sapofonetix: abbreviation, mispronunciation & alphabet substitutions
|
||
sed -i -f $HOME/git/sapo/sapofonetix.sed "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt
|
||
#done < "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt
|
||
|
||
sed -i 's/\.\.\./\.\n/g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # get rid of ...
|
||
sed -i "s/\’/\'/g" "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # get rid of ’
|
||
sed -i 's/[\.\!\?\:\;]/\.\n/g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # delimit to .
|
||
sed -i 's/\…/\,/g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # get rid of …
|
||
sed -i 's/(/\,(/g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # delimit to (
|
||
sed -i 's/)/),/g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # delimit to )
|
||
sed -i 's/^ //g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # delete space at beginning of line
|
||
sed -i 's/^\;$//g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # get rid of just ; lines
|
||
sed -i 's/^\.$//g' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # get rid of just . lines
|
||
sed -i "s/^'//g" "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt #get rid of ' at start of line
|
||
#sed -i 's/\,/\n/3' "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # delimit to 3rd occurence of ,
|
||
sed -i "s/—/ — /g" "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt # put spaces around —, to not interfere with pronunciation
|
||
##################extra delimit to comma and space ###################################
|
||
DELIM=$(cat "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt|wc -l)
|
||
while [ $DELIM -ge 1 ]
|
||
do
|
||
CUR_DELIM_LINE=$(cat "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt|head -$DELIM|tail +$DELIM)
|
||
CHAR_COUNT=$(echo "$CUR_DELIM_LINE"|wc -m)
|
||
while [ $CHAR_COUNT -gt 290 ]
|
||
do
|
||
echo "character count : "$CHAR_COUNT
|
||
|
||
echo "characters > 290"
|
||
COMMA_COUNT=$(echo "$CUR_DELIM_LINE"| sed -e 's/\(.\)/\1\n/g' | grep "," | wc -l)
|
||
echo "comma count : "$COMMA_COUNT
|
||
if [ $COMMA_COUNT -gt 2 ]
|
||
then
|
||
echo "commas > 2"
|
||
echo $DELIM"s/\,/\n/"$(($COMMA_COUNT - 2))>"$DIRECTORY""Sapo_""$NAME"/script.sed
|
||
sed -i -f "$DIRECTORY""Sapo_""$NAME"/script.sed "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt
|
||
fi
|
||
if [ $COMMA_COUNT -le 2 ]
|
||
then
|
||
echo "No commas to delimit to, proceed delimiting with spaces(limit 30)"
|
||
SPACE_COUNT=$(echo "$CUR_DELIM_LINE"| sed -e 's/\(.\)/\1\n/g' | grep " " | wc -l)
|
||
echo "Space count : "$SPACE_COUNT
|
||
|
||
echo $DELIM"s/\ /\n/"$(($SPACE_COUNT - 30))>"$DIRECTORY""Sapo_""$NAME"/script.sed
|
||
sed -i -f "$DIRECTORY""Sapo_""$NAME"/script.sed "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt
|
||
fi
|
||
CUR_DELIM_LINE=$(cat "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt|head -$DELIM|tail +$DELIM)
|
||
CHAR_COUNT=$(echo "$CUR_DELIM_LINE"|wc -m)
|
||
|
||
done
|
||
echo "Task completed for line "$DELIM
|
||
((DELIM--))
|
||
done
|
||
|
||
yad --image "$HOME/git/sapo/sapo.png" --height=40 --width=400 --title="${NAME} - Sapo" --button=gtk-cancel:1 --button=gtk-open:2 --button=gtk-ok:0 --text="The text is prepared, Press:
|
||
1. <span foreground='yellow'><b>Open </b></span>to edit the new file
|
||
2. <span foreground='yellow'><b>OK </b></span>to proceed to speech conversion." --window-icon=$HOME/git/sapo/sapo.png
|
||
case $? in
|
||
0)
|
||
;;
|
||
1) exit
|
||
;;
|
||
2) xed "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt & yad --image "$HOME/git/sapo/sapo.png" --height=40 --width=400 --title="${NAME} - Sapo" --button=gtk-cancel:1 --button=gtk-ok:0 --text="If you are done editing $DIRECTORY\Sapo$NAME/$NAME\sentenced.txt, press OK to continue!" --window-icon=$HOME/git/sapo/sapo.png
|
||
case $? in
|
||
0)
|
||
;;
|
||
1) exit
|
||
;;
|
||
esac
|
||
esac
|
||
TOTALLINES=$(cat "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt|wc -l)
|
||
LINE=1
|
||
(
|
||
while [ $LINE -le $TOTALLINES ]
|
||
do
|
||
CURRENTLINE=$(head -$LINE "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt|tail +$LINE)
|
||
FORMLINE="$(printf "%.6d" $LINE)"
|
||
tts --text "$CURRENTLINE" --out_path "$DIRECTORY""Sapo_""$NAME"/$FORMLINE.wav
|
||
####error detection routine######
|
||
CURRENTLINE_LENGTH=$(echo $CURRENTLINE|wc -m)
|
||
WAV_LENGTH=$(jq -n $(sox "$DIRECTORY""Sapo_""$NAME"/$FORMLINE.wav -n stat 2>&1 |sed -n 's#^Length (seconds):[^0-9]*\([0-9.]*\)$#\1#p')-0.660)
|
||
RATIO=$(jq -n $CURRENTLINE_LENGTH/$WAV_LENGTH|sed 's/\..*$//')
|
||
if [ $RATIO -le 8 ]
|
||
then
|
||
echo $LINE $FORMLINE.wav $CURRENTLINE_LENGTH $WAV_LENGTH $RATIO>>"$DIRECTORY""Sapo_""$NAME"/errors.tsv
|
||
fi
|
||
###################################
|
||
LINE100=$(( $LINE * 100 ))
|
||
PERCENTAGE=$(( $LINE100 / $TOTALLINES))
|
||
###Estimating Estimated time of arrival
|
||
LINESLEFT=$(( $TOTALLINES - $LINE ))
|
||
SECONDS=$(( $LINESLEFT * 8 ))
|
||
HOURS=$(( SECONDS / 3600 ))
|
||
SECHLEFT=$(( $SECONDS - $((HOURS * 3600 )) ))
|
||
MINUTES=$(( $SECHLEFT / 60 ))
|
||
SECMLEFT=$(( $SECHLEFT - $((MINUTES * 60 )) ))
|
||
HOURSTRING="$HOURS"" hrs"
|
||
MINUTESTRING="$MINUTES"" mins"
|
||
## if hours / minutes left are 0 , they are not mentioned
|
||
if [ $HOURS -eq 0 ]
|
||
then
|
||
HOURSTRING=""
|
||
fi
|
||
if [ $MINUTES -eq 0 ]
|
||
then
|
||
MINUTESTRING=""
|
||
fi
|
||
#echo line starting with #, updated in the yad progress bar window
|
||
echo "# Reading line $(( $LINE + 1)) of $TOTALLINES from "$NAME" ($PERCENTAGE%). Roughly remaining : "$HOURSTRING" "$MINUTESTRING" " $SECMLEFT" secs"
|
||
echo "$PERCENTAGE"
|
||
((LINE++))
|
||
done
|
||
) |
|
||
yad --progress --height="40" \
|
||
--title="Sapo - Reading . . . ${NAME}" \
|
||
--percentage=0 \
|
||
--height=40 \
|
||
--width="500" \
|
||
--window-icon=$HOME/git/sapo/sapo.png \
|
||
--image "$HOME/git/sapo/sapo_progress.png" \
|
||
--auto-close
|
||
#--text="Preparing to read..." \
|
||
case $? in
|
||
0)
|
||
;;
|
||
1) exit
|
||
;;
|
||
esac
|
||
##### Error correction routine ######
|
||
(
|
||
TOTAL_ERRORS=$(cat "$DIRECTORY""Sapo_""$NAME"/errors.tsv|wc -l)
|
||
ERROR_LINE=1
|
||
while [ $ERROR_LINE -le $TOTAL_ERRORS ]
|
||
do
|
||
CURRENT_ERROR_LINE=$(cat "$DIRECTORY""Sapo_""$NAME"/errors.tsv|head -$ERROR_LINE|tail +$ERROR_LINE)
|
||
ERROR_TEXT_LINE=$(echo $CURRENT_ERROR_LINE|awk '{print $1}')
|
||
ERROR_WAV=$(echo $CURRENT_ERROR_LINE|awk '{print $2}')
|
||
echo "#Attempting to fix error $ERROR_LINE of $TOTAL_ERRORS, line $ERROR_TEXT_LINE from $NAME.sentenced.txt"
|
||
###################################
|
||
ERROR_LINE100=$(( $ERROR_LINE * 100 ))
|
||
ERROR_PERCENTAGE=$(( $ERROR_LINE100 / $TOTAL_ERRORS))
|
||
echo "$ERROR_PERCENTAGE"
|
||
tts --text "$(cat "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt|head -$ERROR_TEXT_LINE|tail +$ERROR_TEXT_LINE)" --out_path "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV
|
||
|
||
|
||
((ERROR_LINE++))
|
||
done
|
||
)|
|
||
yad --progress --height="40" \
|
||
--title="Sapo - Fixing Errors . . . ${NAME}" \
|
||
--percentage=0 \
|
||
--height=40 \
|
||
--width="500" \
|
||
--window-icon=$HOME/git/sapo/sapo.png \
|
||
--image "$HOME/git/sapo/sapo_progress.png" \
|
||
--auto-close
|
||
####### FIXING ERRORS ONE BY ONE ############
|
||
TOTAL_ERRORS=$(cat "$DIRECTORY""Sapo_""$NAME"/errors.tsv|wc -l)
|
||
ERROR_LINE=1
|
||
while [ $ERROR_LINE -le $TOTAL_ERRORS ]
|
||
do
|
||
CURRENT_ERROR_LINE=$(cat "$DIRECTORY""Sapo_""$NAME"/errors.tsv|head -$ERROR_LINE|tail +$ERROR_LINE)
|
||
ERROR_TEXT_LINE=$(echo $CURRENT_ERROR_LINE|awk '{print $1}')
|
||
ERROR_WAV=$(echo $CURRENT_ERROR_LINE|awk '{print $2}')
|
||
TEXT_TO_CORRECT="$(cat "$DIRECTORY""Sapo_""$NAME"/"$NAME"sentenced.txt|head -$ERROR_TEXT_LINE|tail +$ERROR_TEXT_LINE)"
|
||
GO=false
|
||
while [[ $GO == false ]]
|
||
do
|
||
yad --image "$HOME/git/sapo/sapo.png" \
|
||
--title="Line $ERROR_TEXT_LINE - $NAME" \
|
||
--button=gtk-cancel:1 \
|
||
--button='▶️ Play Audio':2 \
|
||
--button='🔃 Re-Render':3 \
|
||
--button='✂️ Trim Clutter':4 \
|
||
--button='🗡 Split-Render':5 \
|
||
--button='🎵 Edit audio':7 \
|
||
--button='❌ Remove audio':6 \
|
||
--button='😃 Keep/Next':0 \
|
||
--text="Text of line $ERROR_TEXT_LINE :
|
||
|
||
|
||
<span foreground='yellow'><b>$TEXT_TO_CORRECT</b></span>
|
||
|
||
|
||
<i>What would you like to do?</i>" \
|
||
--window-icon=$HOME/git/sapo/sapo.png
|
||
case $? in
|
||
0) GO=true
|
||
;;
|
||
1) GO=true;exit
|
||
;;
|
||
2) celluloid "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV
|
||
;;
|
||
3) s="$(yad --entry --width="600" --text="This is the original text of the line $ERROR_TEXT_LINE.
|
||
Edit as you wish, then hit OK to render." --entry-text="$TEXT_TO_CORRECT" --window-icon=$HOME/git/sapo/sapo.png --title="Line $ERROR_TEXT_LINE - $NAME")";tts --text "$s" --out_path "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV
|
||
;;
|
||
4)sox -V3 "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV "$DIRECTORY""Sapo_""$NAME"/temp.wav silence 1 0.50 0.1% 1 0.5 0.1%;sox "$DIRECTORY""Sapo_""$NAME"/temp.wav "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV pad 0 0.5;rm "$DIRECTORY""Sapo_""$NAME"/temp.wav
|
||
;;
|
||
5)s="$(yad --entry --width="600" --text="Split the printed text roughly in half with the pipe symbol (|), so that it can be rendered in two batches: " --entry-text="$TEXT_TO_CORRECT" --window-icon=$HOME/git/sapo/sapo.png --title="Line $ERROR_TEXT_LINE - $NAME")";s1="$(echo $s|sed 's/|.*$//')";s2="$(echo $s|sed 's/^.*|//')"; echo $s1;echo $s2 ; tts --text "$s1" --out_path "$DIRECTORY""Sapo_""$NAME"/1temp.wav; tts --text "$s2" --out_path "$DIRECTORY""Sapo_""$NAME"/2temp.wav;sox "$DIRECTORY""Sapo_""$NAME"/1temp.wav "$DIRECTORY""Sapo_""$NAME"/2temp.wav "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV ; rm "$DIRECTORY""Sapo_""$NAME"/1temp.wav "$DIRECTORY""Sapo_""$NAME"/2temp.wav
|
||
;;
|
||
6) if [[ -e "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV ]];then rm "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV;notify-send "$DIRECTORY""Sapo_""$NAME"/"$ERROR_WAV"" has been deleted."; else notify-send "There is no file ""$DIRECTORY""Sapo_""$NAME"/"$ERROR_WAV"". Already deleted?";fi
|
||
;;
|
||
7)audacity "$DIRECTORY""Sapo_""$NAME"/$ERROR_WAV
|
||
;;
|
||
esac
|
||
done
|
||
((ERROR_LINE++))
|
||
done
|
||
sox "$DIRECTORY""Sapo_""$NAME"/*.wav "$DIRECTORY""Sapo_""$NAME"/"$NAME".wav
|
||
yad --image "$HOME/git/sapo/sapo.png" --height=40 --width=400 --title="${NAME} - Sapo" --text="Reading of ""$NAME"" is complete!" --window-icon=$HOME/git/sapo/sapo.png
|