#!/bin/bash

datasets=("youtube_music" "genius_hq" "podcasts")
#subsets=("val" "train") 
subsets=("val") 

for subset in "${subsets[@]}"; do 
  for dataset in "${datasets[@]}"; do 

    # Directory containing the MP3 files
    input_directory="/app/suno/data/audio_2ch_48khz_lg/$subset/$dataset"
    # Directory to save the converted WAV files
    output_directory="/app/suno/data/audio_2ch_24khz_lg/$subset/$dataset"

    mkdir -p "$output_directory"
    #echo "$input_directory"
    #echo "$output_directory"

    for file in "$input_directory"/*.wav; do
      # Get the base name of the file
      base_name=$(basename "$file" .wav)
      # Convert and resample to 24 kHz WAV
      ffmpeg -i "$file" -ar 24000 "$output_directory/$base_name.wav"
    done
  done
done