#!/bin/bash

shuf metadata.csv > metadata-shuf.csv
cut --complement -f 2 -d\| metadata-shuf.csv > metadata-shuf-2.csv
awk -F '|' -v OFS='|' '$1 {print "wavs/" $1 ".wav" "|" $2}' metadata-shuf-2.csv > metadata-shuf-3.csv
sed -n '1,50p' metadata-shuf-3.csv > ljs_audio_text_val.txt
sed -n '51,200p' metadata-shuf-3.csv > ljs_audio_text_test.txt
sed -n '201,$ p' metadata-shuf-3.csv > ljs_audio_text_train_v3.txt
awk -F '|' -v OFS='|' '$1 {print $1 "|" "pitch/" substr($1,6) ".pt" "|" $2}'  ljs_audio_text_val.txt > ljs_audio_pitch_text_val.txt
awk -F '|' -v OFS='|' '$1 {print $1 "|" "pitch/" substr($1,6) ".pt" "|" $2}'  ljs_audio_text_test.txt > ljs_audio_pitch_text_test.txt
awk -F '|' -v OFS='|' '$1 {print $1 "|" "pitch/" substr($1,6) ".pt" "|" $2}'  ljs_audio_text_train_v3.txt > ljs_audio_pitch_text_train_v3.txt