Random split audio dataset

build-list.sh

FOLDERS=(
    "./AiDataTang/"
    "./AiShell/"
    "./AiShell-3/"
    "./Genshin/"
    "./LibriTTS_R/"
    "./StarRail/"
)

# Clear filelist.txt
echo "" > filelist.txt

for folder in "${FOLDERS[@]}"; do
    find $folder -iname "*.wav" -o -iname "*.flac" >> filelist.txt
    echo "Listed $folder"
done

# Random choice 100 lines to valid and others to train
shuf filelist.txt | uniq > filelist.shuf.txt
echo "Found $(wc -l filelist.shuf.txt) files, spliting..."

head -n 100 filelist.shuf.txt > filelist.split.valid
tail -n +101 filelist.shuf.txt > filelist.split.train
rm filelist.txt filelist.shuf.txt

echo "Done."