{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import random\n",
    "import shutil\n",
    "import sys\n",
    "\n",
    "libri_path = \"....../LibriTTS\" # absolute path to TensorFlowTTS.\n",
    "dataset_path = \"....../libritts\" # Change to your paths. This is a output of re-format dataset.\n",
    "subset = \"train-clean-100\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(os.path.join(libri_path, \"SPEAKERS.txt\")) as f:\n",
    "    data = f.readlines()\n",
    "    \n",
    "dataset_info = {}\n",
    "max_speakers = 20  # Max number of speakers to train on\n",
    "min_len = 20  # Min len of speaker narration time\n",
    "max_file_len = 11  # max audio file lenght\n",
    "min_file_len = 2  # min audio file lenght"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "possible_dataset = [i.split(\"|\") for i in data[12:] if i.split(\"|\")[2].strip() == subset and float(i.split(\"|\")[3].strip()) >= min_len]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "ids = [i[0].strip() for i in possible_dataset]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import soundfile as sf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "possible_map = {}\n",
    "subset_path = os.path.join(libri_path, subset)\n",
    "for i in os.listdir(subset_path):\n",
    "    if i in ids:\n",
    "        id_path = os.path.join(subset_path, i)\n",
    "        id_dur = 0\n",
    "        id_included = []\n",
    "        \n",
    "        for k in os.listdir(id_path):\n",
    "            for j in os.listdir(os.path.join(id_path, k)):\n",
    "                if \".wav\" in j:\n",
    "                    f_path = os.path.join(id_path, k, j)\n",
    "                    sf_file =  sf.SoundFile(f_path)\n",
    "                    dur = len(sf_file) / sf_file.samplerate\n",
    "                    if max_file_len < dur or dur < min_file_len:\n",
    "                        continue\n",
    "                    else:\n",
    "                        id_included.append(f_path)\n",
    "                        id_dur += dur\n",
    "        possible_map[i] = {\"dur\": id_dur, \"included\": id_included}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "poss_speakers = {k: v[\"included\"] for k, v in possible_map.items() if v[\"dur\"]/60 >= min_len}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "to_move = list(poss_speakers.keys())\n",
    "random.shuffle(to_move)\n",
    "to_move = to_move[:max_speakers]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "for sp_id, v in poss_speakers.items():\n",
    "    if sp_id in to_move:\n",
    "        for j in v:\n",
    "            f_name = j.split(os.path.sep)[-1]\n",
    "            text_f_name = f_name.split(\".wav\")[0] + \".txt\"\n",
    "            os.makedirs(os.path.join(dataset_path, sp_id), exist_ok=True)\n",
    "            shutil.copy(j, os.path.join(dataset_path, sp_id, f_name))\n",
    "            shutil.copy(j.replace(\".wav\", \".normalized.txt\"), os.path.join(dataset_path, sp_id, text_f_name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}