process_audio.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from multiprocessing.pool import Pool \n",
    "from synthesizer import audio\n",
    "from functools import partial\n",
    "from itertools import chain\n",
    "from pathlib import Path\n",
    "from utils import logmmse\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "import librosa"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_audios(datasets_root, out_dir, n_processes, \n",
    "                           skip_existing, hparams, pairs):\n",
    "    # Gather the input directories\n",
    "    \n",
    "    print(\"\\n  Using data from:  \" + datasets_root)\n",
    "    \n",
    "    # Create the output directories for each output file type\n",
    "    out_dir.joinpath(\"mels\").mkdir(exist_ok=True)\n",
    "    out_dir.joinpath(\"audio\").mkdir(exist_ok=True)\n",
    "    \n",
    "    # Create a metadata file\n",
    "    metadata_fpath = out_dir.joinpath(\"train.txt\")\n",
    "    metadata_file = metadata_fpath.open(\"a\" if skip_existing else \"w\", encoding=\"utf-8\")\n",
    "\n",
    "    #print(speaker_dirs)\n",
    "    func = partial(preprocess_speaker, out_dir=out_dir, skip_existing=skip_existing, \n",
    "                   hparams=hparams, datasets_root=datasets_root)\n",
    "    job = Pool(n_processes).imap(func, pairs)\n",
    "    for speaker_metadata in tqdm(job, \"Proprocess\", len(pairs), unit=\"speakers\"):\n",
    "        for metadatum in speaker_metadata:\n",
    "            metadata_file.write(\"|\".join(str(x) for x in metadatum) + \"\\n\")\n",
    "    metadata_file.close()\n",
    "\n",
    "    # Verify the contents of the metadata file\n",
    "    with metadata_fpath.open(\"r\", encoding=\"utf-8\") as metadata_file:\n",
    "        metadata = [line.split(\"|\") for line in metadata_file]\n",
    "    mel_frames = sum([int(m[4]) for m in metadata])\n",
    "    timesteps = sum([int(m[3]) for m in metadata])\n",
    "    sample_rate = hparams.sample_rate\n",
    "    hours = (timesteps / sample_rate) / 3600\n",
    "    print(\"The dataset consists of %d utterances, %d mel frames, %d audio timesteps (%.2f hours).\" %\n",
    "          (len(metadata), mel_frames, timesteps, hours))\n",
    "    print(\"Max input length (text chars): %d\" % max(len(m[5]) for m in metadata))\n",
    "    print(\"Max mel frames length: %d\" % max(int(m[4]) for m in metadata))\n",
    "    print(\"Max audio timesteps length: %d\" % max(int(m[3]) for m in metadata))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_speaker(speaker_dir, out_dir: Path, skip_existing: bool, hparams, datasets_root):\n",
    "    metadata = []\n",
    "    wav_texts = list( (datasets_root + x[0] + '/' + x[1], x[2]) for x in speaker_dir)\n",
    "    wavs = []\n",
    "    texts = []\n",
    "    wav_paths = []\n",
    "    for wav_path, text in wav_texts:\n",
    "        wav = split_on_silences(wav_path, hparams)\n",
    "        wavs.append(wav)\n",
    "        texts.append(text)\n",
    "        wav_paths.append(wav_path)\n",
    "    assert len(wav_paths) == len(wavs) == len(texts)\n",
    "    for i, (wav, text) in enumerate(zip(wavs, texts)):\n",
    "            metadata.append(process_utterance(wav, text, out_dir, wav_paths[i].split('/')[-1], \n",
    "                                            skip_existing, hparams))\n",
    "    return [m for m in metadata if m is not None]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def split_on_silences(wav_fpath, hparams):\n",
    "    # Load the audio waveform\n",
    "    wav, _ = librosa.load(wav_fpath + '.wav', hparams.sample_rate)\n",
    "    if hparams.rescale:\n",
    "        wav = wav / np.abs(wav).max() * hparams.rescaling_max\n",
    "    \n",
    "    return wav"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def process_utterance(wav, text, out_dir, basename, \n",
    "                      skip_existing, hparams):\n",
    "    ## FOR REFERENCE:\n",
    "    # For you not to lose your head if you ever wish to change things here or implement your own\n",
    "    # synthesizer.\n",
    "    # - Both the audios and the mel spectrograms are saved as numpy arrays\n",
    "    # - There is no processing done to the audios that will be saved to disk beyond volume  \n",
    "    #   normalization (in split_on_silences)\n",
    "    # - However, pre-emphasis is applied to the audios before computing the mel spectrogram. This\n",
    "    #   is why we re-apply it on the audio on the side of the vocoder.\n",
    "    # - Librosa pads the waveform before computing the mel spectrogram. Here, the waveform is saved\n",
    "    #   without extra padding. This means that you won't have an exact relation between the length\n",
    "    #   of the wav and of the mel spectrogram. See the vocoder data loader.\n",
    "    \n",
    "    \n",
    "    # Skip existing utterances if needed\n",
    "    mel_fpath = out_dir.joinpath(\"mels\", \"mel-%s.npy\" % basename)\n",
    "    wav_fpath = out_dir.joinpath(\"audio\", \"audio-%s.npy\" % basename)\n",
    "    if skip_existing and mel_fpath.exists() and wav_fpath.exists():\n",
    "        return None\n",
    "    \n",
    "    # Skip utterances that are too short\n",
    "    if len(wav) < hparams.utterance_min_duration * hparams.sample_rate:\n",
    "        return None\n",
    "    \n",
    "    # Compute the mel spectrogram\n",
    "    mel_spectrogram = audio.melspectrogram(wav, hparams).astype(np.float32)\n",
    "    mel_frames = mel_spectrogram.shape[1]\n",
    "    \n",
    "    # Skip utterances that are too long\n",
    "    if mel_frames > hparams.max_mel_frames and hparams.clip_mels_length:\n",
    "        return None\n",
    "    \n",
    "    # Write the spectrogram, embed and audio to disk\n",
    "    np.save(mel_fpath, mel_spectrogram.T, allow_pickle=False)\n",
    "    np.save(wav_fpath, wav, allow_pickle=False)\n",
    "    \n",
    "    # Return a tuple describing this training example\n",
    "    return wav_fpath.name, mel_fpath.name, \"embed-%s.npy\" % basename, len(wav), mel_frames, text"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocess audios parallelly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size=200\n",
    "pairs = []\n",
    "\n",
    "# format in metadata.csv\n",
    "# spk|wav_name|text\n",
    "\n",
    "with open('datasets/vctk/metadata.csv') as f:\n",
    "    subpairs = []\n",
    "    begin = True\n",
    "    for line in f.readlines():\n",
    "        subpairs.append(line.strip().split('|'))\n",
    "        if len(subpairs) % batch_size == 0 and not begin:\n",
    "            pairs.append(subpairs)\n",
    "            subpairs = []\n",
    "            \n",
    "        begin = False\n",
    "        \n",
    "    if len(subpairs) > 0:\n",
    "        pairs.append(subpairs)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "  Using data from:  datasets/vctk/wavs/\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Proprocess: 100%|██████████| 221/221 [01:51<00:00,  1.98speakers/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The dataset consists of 30617 utterances, 6264070 mel frames, 1249113766 audio timesteps (21.69 hours).\n",
      "Max input length (text chars): 186\n",
      "Max mel frames length: 1073\n",
      "Max audio timesteps length: 214400\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "from synthesizer.hparams import hparams\n",
    "from pathlib import Path\n",
    "\n",
    "# rootpath is the path to the wav files\n",
    "rootpath = \"datasets/vctk/wavs/\"\n",
    "\n",
    "# the path to store features, including audio and mel-spectrogram\n",
    "outpath = Path(\"datasets/vctk\").joinpath(\"synthesizer\")\n",
    "outpath.mkdir(exist_ok=True, parents=True)\n",
    "\n",
    "preprocess_audios(rootpath, outpath, 16, True, hparams, pairs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Distribution of the dataset\n",
    "According to the utterance_min_duration set in hparams.py, more than 10, 000 utterances from VCTK dataset would be ignored."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "108 speakers are used for training and testing\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAc8AAAfaCAYAAAAFs/c5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAPYQAAD2EBqD+naQAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdeXhMdx///1f2RBBrSTVC7CoV1VttIbaiFK3lLlpiCb2/lC56t7pRrSripk33qqWEm1hqaa0V6luqIqVVbSyV2EqrSCy3iOT8/vCb+WZkkTOZGYPn47rmaufMmXPeZzLyzlk+5+VhGIYhAABQZJ43uwAAAG41NE8AAEyieQIAYBLNEwAAk2ieAACYRPMEAMAkmicAACbRPAEAMInmCQCASTRPkzZv3iwPDw8tWbLkpq5/8+bNN2X9hYmOjla1atVudhmFunDhgoYOHarKlSvLw8NDzzzzzM0uycb48ePl4eGh06dP3+xSHC41NVUeHh6KjY29aTVcvXpV//73vxUSEiJPT0/16NHjptWCWxvNU5KHh0eRHu7YsFztxIkTGj9+vHbv3n2zS7HL22+/rTlz5uhf//qX5s2bpyeffPJmlwQXmjVrlqZOnapevXpp7ty5evbZZ292SQW67777VLVqVRV2B9UWLVqoUqVKunr1qnVaRkaG3njjDTVs2FAlS5ZUQECAGjRooBdffFEnTpyw/gFelIckzZkzRx4eHkpKSrJZd3p6upo0aSJ/f3+tXbvWOR/Cdfbt26fx48crNTXVJesrjPfNLsAdzJs3z+b5F198oQ0bNuSZXq9ePf3666+uLM3tnDhxQm+88YaqVaumiIgIm9c+++wz5eTk3KTKimbTpk1q2rSpxo0bd7NLwU2wadMmValSRdOnT7/ZpdxQ//799dJLL2nr1q1q1apVntdTU1O1fft2jRw5Ut7e136V//7772rfvr2OHDmi3r17a9iwYfL19dVPP/2kzz//XMuXL9fWrVvz/G4bO3asSpYsqVdeeaVItWVkZOihhx7STz/9pOXLl6tTp07F3+Ai2Ldvn9544w1FRUXd9KNcNE9JTzzxhM3z77//Xhs2bMgzXdJt1zwvX74sX19feXoW/yCEj4+PAypyrj///FP169e/2WXASa5evaqcnBz5+vrm+/qff/6pMmXKFHs5rtCvXz+NHTtWCxYsyLd5Lly4UIZhqH///pKu1fzYY4/p1KlT2rx5s1q2bGkz/8SJEzV58mRVqlQpz++2d955RxUqVMj3d971zp8/r44dO2r37t1atmyZOnfuXIytvHVx2NZOOTk5mjhxou655x75+/urXbt2OnjwYJ75duzYoU6dOikoKEglSpRQ69at9d133xVpHceOHVOPHj0UGBiou+66S88++6wyMzPzzFetWjVFR0fnmR4VFaWoqCjrc8vhmv/+97969dVXVaVKFZUoUUIZGRk6c+aMxowZo/DwcJUsWVKlS5dW586dtWfPHpv3/+Mf/5AkDRo0yHpoZ86cOZLyP+d58eJFPf/88woJCZGfn5/q1Kmj2NjYPIeiPDw8NHLkSH355Zdq0KCB/Pz8dO+99xb5cNCff/6pIUOGqFKlSvL391fDhg01d+7cPNt++PBhffXVV9bab3T4Z/78+WrcuLECAgJUrlw5Pf744zp69KjNPFu3blXv3r1VtWpV+fn5KSQkRM8++6z+97//5Vneb7/9pj59+qhixYoKCAhQnTp18v1r/9y5c4qOjlaZMmUUFBSkQYMG6dKlSzf8HA4cOKCePXuqcuXK8vf31z333KPHH39c6enp1nksn3V8fLzq1Kkjf39/NW7cWN9++22e5R0/flyDBw9WpUqVrD+TWbNm2cxz5coVvf7662rcuLGCgoIUGBioyMhIJSYm3rBewzCse0fLli2z2f5nnnnG+r2pWbOmJk+ebHNkI/c51BkzZqhGjRry8/PTvn378qzHMm9iYqJ++eUXm1MxN1rOjb5b19fywQcfKCwsTCVKlNBDDz2ko0ePyjAMvfnmm7rnnnsUEBCg7t2768yZM4V+NiEhIWrVqpWWLFmirKysPK8vWLBANWrU0IMPPihJWrp0qfbs2aNXXnklT+OUpNKlS2vixImFrvNGLly4oE6dOik5OVlLly5Vly5dbvierKwsvfHGG6pVq5b8/f1Vvnx5tWzZUhs2bLCZ77ffflOvXr1Urlw5+fv764EHHtDKlSutr8+ZM0e9e/eWJLVp0+amn05jz9NO77zzjjw9PTVmzBilp6drypQp6t+/v3bs2GGdZ9OmTercubMaN26scePGydPTU7Nnz1bbtm21detWNWnSpMDl/+9//1O7du105MgRjRo1SnfffbfmzZunTZs2Fbv2N998U76+vhozZowyMzPl6+urffv26csvv1Tv3r1VvXp1nTp1Sp988olat26tffv26e6771a9evU0YcIEvf766xo2bJgiIyMlSc2bN893PYZhqFu3bkpMTNSQIUMUERGhdevW6YUXXtDx48fzHDr7v//3/2rZsmX6P//n/6hUqVJ677331LNnTx05ckTly5cv9LOKiorSwYMHNXLkSFWvXl0JCQmKjo7WuXPnNHr0aNWrV0/z5s3Ts88+q3vuuUfPP/+8JKlixYoFLnfixIl67bXX1KdPHw0dOlR//fWX4uLi1KpVK/3444/WPZiEhARdunRJ//rXv1S+fHn98MMPiouL07Fjx5SQkGBd3k8//aTIyEj5+Pho2LBhqlatmg4dOqRVq1bl+aXWp08fVa9eXZMmTVJycrJmzpypu+66S5MnTy6w3itXrqhjx47KzMzU008/rcqVK+v48eNavXq1zp07p6CgIOu8W7Zs0aJFizRq1Cj5+fnpww8/VKdOnfTDDz+oQYMGkqRTp06padOm1mZbsWJFrVmzRkOGDFFGRob1YquMjAzNnDlTffv2VUxMjM6fP6/PP/9cHTt21A8//JDn8L5Fdna2Bg8erEWLFmn58uXWX8SXLl1S69atdfz4cQ0fPlxVq1bVtm3bNHbsWP3xxx+aMWOGzXJmz56ty5cva9iwYfLz81O5cuXyrKtixYqaN2+eJk6cqAsXLmjSpEmSrp2KsfyRk99yivLdyi0+Pl5XrlzR008/rTNnzmjKlCnq06eP2rZtq82bN+vFF1/UwYMHFRcXpzFjxuT5Q+R6/fv317Bhw7Ru3Tp17drVOv3nn3/W3r179frrr1unWRqNs87jX7x4UZ07d9bOnTu1ZMkSm3oKM378eE2aNElDhw5VkyZNlJGRoaSkJCUnJ6tDhw6SpF9++UUtWrRQlSpV9NJLLykwMFCLFy9Wjx49tHTpUj366KNq1aqVRo0apffee08vv/yy6tWrJ0nW/7qcgTxGjBhhFPTRJCYmGpKMevXqGZmZmdbp7777riHJ+Pnnnw3DMIycnByjVq1aRseOHY2cnBzrfJcuXTKqV69udOjQodAaZsyYYUgyFi9ebJ128eJFo2bNmoYkIzEx0To9NDTUGDhwYJ5ltG7d2mjdunWe2sPCwoxLly7ZzHv58mUjOzvbZtrhw4cNPz8/Y8KECdZpO3fuNCQZs2fPzrO+gQMHGqGhodbnX375pSHJeOutt2zm69Wrl+Hh4WEcPHjQOk2S4evrazNtz549hiQjLi4uz7pys3xW8+fPt067cuWK0axZM6NkyZJGRkaGdXpoaKjRpUuXQpdnGIaRmppqeHl5GRMnTrSZ/vPPPxve3t4206//LA3DMCZNmmR4eHgYaWlp1mmtWrUySpUqZTPNMAyb78e4ceMMScbgwYNt5nn00UeN8uXLF1rzjz/+aEgyEhISCp1PkiHJSEpKsk5LS0sz/P39jUcffdQ6bciQIUZwcLBx+vRpm/c//vjjRlBQkHW7r169avNvwTAM4+zZs0alSpVstuPw4cOGJGPq1KlGVlaW8c9//tMICAgw1q1bZ/PeN9980wgMDDT2799vM/2ll14yvLy8jCNHjtgsr3Tp0saff/5Z6DZbtG7d2rj33nttphW2nKJ+tyzLqFixonHu3DnrvGPHjjUkGQ0bNjSysrKs0/v27Wv4+voaly9fLrTeM2fOGH5+fkbfvn3zfBaSjJSUFOu0Ro0aGUFBQUX6HK5377332vyuyG327NmGJCM0NNTw8fExvvzyS1PLbtiw4Q3/zbVr184IDw+3+TxycnKM5s2bG7Vq1bJOS0hIyPP772bhsK2dBg0aZHM+xLIX9vvvv0uSdu/erQMHDqhfv376+++/dfr0aZ0+fVoXL15Uu3bt9O233xZ6cc3XX3+t4OBg9erVyzqtRIkSGjZsWLFrHzhwoAICAmym+fn5Wc97Zmdn6++//1bJkiVVp04dJScn27Wer7/+Wl5eXho1apTN9Oeff16GYWjNmjU209u3b68aNWpYn993330qXbq09TMtbD2VK1dW3759rdN8fHw0atQoXbhwQVu2bDFd+7Jly5STk6M+ffpYf3anT59W5cqVVatWLZtDkrk/y4sXL+r06dNq3ry5DMPQjz/+KEn666+/9O2332rw4MGqWrWqzbosVzXm9tRTT9k8j4yM1N9//62MjIwCa7bsWa5bt+6Gh3ibNWumxo0bW59XrVpV3bt317p165SdnS3DMLR06VI98sgjMgzD5jPo2LGj0tPTrd8LLy8v67+FnJwcnTlzRlevXtUDDzyQ73fnypUr6t27t1avXq2vv/5aDz30kM3rCQkJioyMVNmyZW3W2759e2VnZ+c5vNyzZ89CjyAUVX7LMfvd6t27t80evuWQ6hNPPGG9qMcy/cqVKzp+/HihNZUtW1YPP/ywVq5cqYsXL0q6dkTnv//9rx544AHVrl3bOm9GRoZKlSplcquL7tSpU/L391dISIip95UpU0a//PKLDhw4kO/rZ86c0aZNm9SnTx+dP3/e+vP++++/1bFjRx04cOCGn9PNwGFbO13/C7Bs2bKSpLNnz0qS9YsycODAApeRnp5ufd/10tLSVLNmzTy/WOvUqWN3zRbVq1fPMy0nJ0fvvvuuPvzwQx0+fFjZ2dnW1wo7ZFqYtLQ03X333Xn+QVsOs6SlpdlMv/4zla59rpbPtLD11KpVK89FTwWtpygOHDggwzBUq1atfF/PfXHUkSNH9Prrr2vlypV5arWca7T8AWA5JHojhX2/Spcune97qlevrueee07/+c9/FB8fr8jISHXr1k1PPPGEzS90SfluV+3atXXp0iX99ddf8vT01Llz5/Tpp5/q008/zXd9f/75p/X/586dq2nTpum3336zOT+X33dt0qRJunDhgtasWWNzTt7iwIED+umnnwpsiLnXW9A67JHfcsx+t67/uVk+9+sbjmX6jb7b0rVDt8uXL9eKFSvUr18/bdu2TampqXkOGRflD83i+OSTT/Tcc8+pU6dO2rp1a5F/F02YMEHdu3dX7dq11aBBA3Xq1ElPPvmk7rvvPknSwYMHZRiGXnvtNb322mv5LuPPP/9UlSpVHLYtjkDztJOXl1e+043//0IYy17l1KlTCzznU7JkSYfUkt+ei3RtDzK/Oq/f65SujX987bXXNHjwYL355psqV66cPD099cwzz7hs+MmNPlNXysnJkYeHh9asWZNvXZafXXZ2tjp06KAzZ87oxRdfVN26dRUYGKjjx48rOjra7s/O3s9i2rRpio6O1ooVK7R+/XqNGjVKkyZN0vfff6977rmnyOu31P3EE08U+Aeg5Zff/PnzFR0drR49euiFF17QXXfdJS8vL02aNEmHDh3K876OHTtq7dq1mjJliqKiouTv759n3R06dNC///3vfNebe29Lyv/7bA9HLKegn1txvttdu3ZVUFCQFixYoH79+mnBggXy8vLS448/bjNf3bp19eOPP+ro0aOm9w6Lon79+vr666/Vrl07dejQQd99912R1tOqVSsdOnTI+p2cOXOmpk+fro8//lhDhw61ftfGjBmjjh075ruMmjVrOnRbHIHm6SSWw4+lS5dW+/btTb8/NDRUe/fulWEYNs0xJSUlz7xly5bVuXPn8kxPS0tTWFhYkda3ZMkStWnTRp9//rnN9HPnzqlChQrW5wU16vyEhoZq48aNOn/+vM3e52+//WZ93RFCQ0P1008/KScnx2YPoTjrqVGjhgzDUPXq1fP8ss7t559/1v79+zV37lwNGDDAOv36KwktP4e9e/earsWs8PBwhYeH69VXX9W2bdvUokULffzxx3rrrbes8+R3CG3//v0qUaKEdY+vVKlSys7OvuH3d8mSJQoLC9OyZctsvh8FjaVt2rSpnnrqKXXt2lW9e/fW8uXLbQ5p1qhRQxcuXLDr342jOeO7ZZafn5969eqlL774QqdOnVJCQoLatm2rypUr28z3yCOPaOHChZo/f77Gjh3rlFqaNGmiL7/8Ul26dFGHDh20devWIh0yL1eunAYNGqRBgwbpwoULatWqlcaPH6+hQ4da/234+Pjc8Gdu5vePs3HO00kaN26sGjVqKDY2VhcuXMjz+l9//VXo+x9++GGdOHHC5jaAly5dyvcQWo0aNfT999/rypUr1mmrV6/OM6SiMF5eXnn+Ck5ISMhzriEwMFCS8m3W+W1Ddna23n//fZvp06dPl4eHh8PGhz388MM6efKkFi1aZJ129epVxcXFqWTJkmrdurXpZT722GPy8vLSG2+8kedzMQxDf//9t6T/t0eRex7DMPTuu+/avKdixYpq1aqVZs2apSNHjuRZniNkZGTY3GlGutZIPT098wxx2r59u835yKNHj2rFihV66KGH5OXlJS8vL/Xs2VNLly7Nt+Hn/v7m9xns2LFD27dvL7DW9u3b67///a/Wrl2rJ5980mYPvU+fPtq+fbvWrVuX533nzp3Ls43O5Izvlj369++vrKwsDR8+XH/99Zd1bGduvXr1Unh4uCZOnJjvZ3/+/Pki3wShMO3atdPChQt18OBBderUqdDz8JKs/1YsSpYsqZo1a1q/k3fddZeioqL0ySef6I8//sjz/tzfNTO/f5yNPU8n8fT01MyZM9W5c2fde++9GjRokKpUqaLjx48rMTFRpUuX1qpVqwp8f0xMjN5//30NGDBAu3btUnBwsObNm6cSJUrkmXfo0KFasmSJOnXqpD59+ujQoUOaP3++zcU3N9K1a1dNmDBBgwYNUvPmzfXzzz8rPj4+z55rjRo1VKZMGX388ccqVaqUAgMD9eCDD+Z7vuiRRx5RmzZt9Morryg1NVUNGzbU+vXrtWLFCj3zzDOm6ivMsGHD9Mknnyg6Olq7du1StWrVtGTJEn333XeaMWOGXRdR1KhRQ2+99ZbGjh2r1NRU9ejRQ6VKldLhw4e1fPlyDRs2TGPGjFHdunVVo0YNjRkzRsePH1fp0qW1dOnSfM9lvffee2rZsqXuv/9+DRs2TNWrV1dqaqq++uorh9zucNOmTRo5cqR69+6t2rVr6+rVq5o3b561EebWoEEDdezY0WaoiiS98cYb1nneeecdJSYm6sEHH1RMTIzq16+vM2fOKDk5WRs3brSOU+zatauWLVumRx99VF26dNHhw4f18ccfq379+vn+4WjRo0cPzZ49WwMGDFDp0qX1ySefSJJeeOEFrVy5Ul27dlV0dLQaN26sixcv6ueff9aSJUuUmppqczTEmZzx3bJH69atdc8992jFihUKCAjQY489lmceHx8fLVu2TO3bt1erVq3Up08ftWjRQj4+Pvrll1+0YMEClS1btthjPSXp0Ucf1WeffabBgwerW7duWrt2bZ7D7xb169dXVFSUGjdurHLlyikpKUlLlizRyJEjrfN88MEHatmypcLDwxUTE6OwsDCdOnVK27dv17Fjx6zjzSMiIuTl5aXJkycrPT1dfn5+atu2re66665ib5Nprr2499ZQlKEq1w8HsFyqfv0Qjh9//NF47LHHjPLlyxt+fn5GaGio0adPH+Obb765YR1paWlGt27djBIlShgVKlQwRo8ebaxduzbfS7WnTZtmVKlSxfDz8zNatGhhJCUlFThUJb+hDJcvXzaef/55Izg42AgICDBatGhhbN++Pc8yDMMwVqxYYdSvX9/w9va22ebrh6oYhmGcP3/eePbZZ427777b8PHxMWrVqmVMnTrVZniGYVwbPjFixIg8dRU0DOd6p06dMgYNGmRUqFDB8PX1NcLDw/MdTlPUoSoWS5cuNVq2bGkEBgYagYGBRt26dY0RI0bYDBHYt2+f0b59e6NkyZJGhQoVjJiYGOswm+tr2Lt3r/Hoo48aZcqUMfz9/Y06deoYr732mvV1y1CVv/76y+Z9luEChw8fLrDW33//3Rg8eLBRo0YNw9/f3yhXrpzRpk0bY+PGjTbzWT7r+fPnG7Vq1TL8/PyMRo0a5Xv5/6lTp4wRI0YYISEhho+Pj1G5cmWjXbt2xqeffmqdJycnx3j77beN0NBQ67JWr16d5/uQe6hKbh9++KEhyRgzZox12vnz542xY8caNWvWNHx9fY0KFSoYzZs3N2JjY40rV64UurzCFDZUpaDlFOW7VdAyCvo3Z/l57ty5s8i1v/DCC4Yko0+fPoXOd/bsWeP11183wsPDjRIlShj+/v5GgwYNjLFjxxp//PFHvu8pylCV/GqNjY01JBldu3a1GYqT21tvvWU0adLEKFOmjBEQEGDUrVvXmDhxovXnaHHo0CFjwIABRuXKlQ0fHx+jSpUqRteuXY0lS5bYzPfZZ58ZYWFhhpeX100dtuJhGDfhagwAN42Hh4dGjBiR53A6gKLjnCcAACbRPAEAMInmCQCASVxtC9xhuMwBKD72PAEAMInmCQCASbftYducnBydOHFCpUqVcqtbOgEAXMswDJ0/f1533313npv82+u2bZ4nTpxwys2RAQC3pqNHj5oKSCjMbds8LbfNOnr0aIERTgCA219GRoZCQkIcejvF27Z5Wg7Vli5dmuYJAHDoKTwuGAIAwCSaJwAAJtE8AQAwieYJAIBJNE8AAEyieQIAYBLNEwAAk2ieAACYRPMEAMAkmicAACbRPAEAMInmCQCASTRPAABMonkCAGASzRMAAJNongAAmETzBADAJJonAAAm0TwBADCJ5gkAgEk0TwAATKJ5AgBgEs0TAACTaJ4AAJhE8wQAwCSaJwAAJtE8AQAwieYJAIBJNE8AAEyieQIAYBLNEwAAk2ieAACYRPMEAMAkmicAACbRPAEAMInmCQCASTRPAABMonkCAGASzRMAAJNongAAmETzBADAJJonAAAm0TwBADCJ5gkAgEk0TwAATKJ5AgBgEs0TAACTaJ4AAJhE8wQAwCSaJwAAJtE8AQAwieYJAIBJNE8AAEyieQIAYBLNEwAAk2ieAACYRPMEAMAkmicAACbRPAEAMInmCQCASTRPAABMonkCAGASzRMAAJNongAAmETzBADAJJonAAAm0TwBADCJ5gkAgEk0TwAATKJ5AgBgEs0TAACTaJ4AAJhE8wQAwCSaJwAAJtE8AQAwieYJAIBJNE8AAEyieQIAYBLNEwAAk2ieAACYRPMEAMAkmicAACbRPAEAMInmCQCASd43uwBnazBunTz9StzsMnAHSX2ny80uAYCTsecJAIBJNE8AAExyaPPMysrSiy++qPDwcAUGBuruu+/WgAEDdOLECes8qampGjJkiKpXr66AgADVqFFD48aN05UrV2zm8fDwyPP4/vvvHVkuAAB2ceg5z0uXLik5OVmvvfaaGjZsqLNnz2r06NHq1q2bkpKSJEm//fabcnJy9Mknn6hmzZrau3evYmJidPHiRcXGxtosb+PGjbr33nutz8uXL+/IcgEAsIvp5hkVFaUGDRpIkubNmycfHx/961//0oQJExQUFKQNGzbYzP/++++rSZMmOnLkiKpWrapOnTqpU6dO1tfDwsKUkpKijz76KE/zLF++vCpXrmzPdgEA4DR2HbadO3euvL299cMPP+jdd9/Vf/7zH82cOTPfedPT0+Xh4aEyZcoUuLz09HSVK1cuz/Ru3brprrvuUsuWLbVy5cpCa8rMzFRGRobNAwAAZ7CreYaEhGj69OmqU6eO+vfvr6efflrTp0/PM9/ly5f14osvqm/fvipdunS+yzp48KDi4uI0fPhw67SSJUtq2rRpSkhI0FdffaWWLVuqR48ehTbQSZMmKSgoyPoICQmxZ9MAALghD8MwDDNviIqKUlhYmGbNmmWdtmLFCvXq1UuXL1+Wl5eXpGsXD/Xs2VPHjh3T5s2b822ex48fV+vWrRUVFVXgnqvFgAEDdPjwYW3dujXf1zMzM5WZmWl9npGRoZCQEIU8s5hxnnApxnkC7iUjI0NBQUFKT08vcEfOLKfcJCErK0t9+vRRWlqaNm3alG+xJ06cUJs2bdS8eXN9+umnN1zmgw8+mOd8am5+fn7y8/MrVt0AABSFXc1zx44dNs+///571apVS15eXtbGeeDAASUmJuZ7hezx48fVpk0bNW7cWLNnz5an542PHu/evVvBwcH2lAsAgEPZ1TyPHDmi5557TsOHD1dycrLi4uI0bdo0ZWVlqVevXkpOTtbq1auVnZ2tkydPSpLKlSsnX19fHT9+XFFRUQoNDVVsbKz++usv63ItV9bOnTtXvr6+atSokSRp2bJlmjVr1g0P7QIA4Ap2Nc8BAwbof//7n5o0aSIvLy+NHj1aw4YNU1pamvWinoiICJv3JCYmKioqShs2bNDBgwd18OBB3XPPPTbz5D79+uabbyotLU3e3t6qW7euFi1apF69etlTLgAADmXXBUMRERGaMWOGs2pyCGecIAYA3Hqc0Q+4ty0AACbRPAEAMMn0Oc/Nmzc7oQznIc8TcE+Mh8WtjD1PAABMonkCAGCSw5qno7I8pWtDVmJjY1W7dm35+fmpSpUqmjhxoqNKBQCgWBx2ez5HZnmOHj1a69evV2xsrMLDw3XmzBmdOXPGUaUCAFAspsZ5Fpbl6eHhkWf+nTt3qkmTJkpLS1PVqlXzXebUqVP10Ucf6ffff5ck/frrr7rvvvu0d+9e1alTx55tkvT/xvVwY3jAPXHBEFzFLcZ5OjvLc9WqVQoLC9Pq1atVvXp1VatWTUOHDr3hnrswJpoAACAASURBVCd5ngAAVzHdPJ2d5fn7778rLS1NCQkJ+uKLLzRnzhzt2rXrhrfmI88TAOAqpptn06ZNbQ7RNmvWTAcOHFB2drZ1miVZxTAMffTRR/ku5/jx4+rUqZN69+6tmJgY6/ScnBxlZmbqiy++UGRkpKKiovT5558rMTFRKSkpBdY1duxYpaenWx9Hjx41u2kAABSJw/M8i5vlGRwcLG9vb9WuXds6rV69epKupbkUdB6UPE8AgKuY3vMsapbnxo0bC8zyjIqKKjDLs0WLFrp69aoOHTpknbZ//35JUmhoqNlyAQBwONPN05LlmZKSooULFyouLk6jR4+2ZnkmJSUpPj7emuV58uRJ6zhOS+OsWrWqNcvTMo9F+/btdf/992vw4MH68ccftWvXLg0fPlwdOnSw2RsFAOBmMX3Y1tlZnp6enlq1apWefvpptWrVSoGBgercubOmTZtm7zYCAOBQpsd53gpZnhJ5ngCAa9xinCcAAHc6micAACaZOud5q2V5AgDgDA4f5+luCMMGXIN71eJOwmFbAABMcmjzdFSm5/jx4+Xh4ZHnERgY6MhyAQCwi0MP2zoq03PMmDF66qmnbJbdrl07/eMf/3BkuQAA2MXUOE/JNZme19uzZ48iIiL07bffKjIyskh1kucJuBbnPOGu3Gacp7MzPa83c+ZM1a5du9DGSZ4nAMBV7Gqezs70vH4Z8fHxGjJkSKE1kecJAHAVu5qnszM9c1u+fLnOnz+vgQMHFloTeZ4AAFdxyjjP4mZ65jZz5kx17dpVlSpVKnSd5HkCAFzFruZZ1EzPxMTEAjM927RpU2Cmp8Xhw4eVmJhoTWsBAMAd2NU8LZmew4cPV3JysuLi4jRt2jRrpmdycrJWr15tzfSUpHLlysnX19ea6RkaGmrN9LSoXLmyzXpmzZql4OBgde7cuRibCACAY9nVPJ2d6SlJOTk5mjNnjqKjo+Xl5WVPmQAAOIVd4zxvhUxPxnkCrsU4T7grZ4zzvO1vDL/3jY6EYQMAHIobwwMAYJLpPU8yPQEAd7rb/rAteZ7A7Y1zrbgZOGwLAIBJNE8AAExyWPN0VBC2JK1bt05NmzZVqVKlVLFiRfXs2VOpqamOKhUAgGJxWPPMHYSdnJysZcuWKSUlRd26dbPOkzsI+5dfftH06dP18ccf6+WXX7bOc/jwYXXv3l1t27bV7t27tW7dOp0+fVqPPfaYo0oFAKBYTN0kwRVB2EuWLFHfvn2VmZlpveftqlWr1L17d2VmZsrHxyff5WRmZiozM9P6PCMjQyEhIdwkAbjNccEQbsQtwrCdHYTduHFjeXp6avbs2crOzlZ6errmzZun9u3bF9g4JfI8AQCuY3rP888//9Qvv/xi3dN86aWXtHLlSu3bt89m3suXL6tFixaqW7eu4uPj813ewYMH1bhxY8XGxtrkeW7ZskV9+vTR33//rezsbDVr1kxff/11oU2YPU/gzsSeJ27ELfY8nR2EffLkScXExGjgwIHauXOntmzZIl9fX/Xq1UuF9Xk/Pz+VLl3a5gEAgDM4/CYJxQ3C/uCDDxQUFKQpU6ZYp82fP18hISHasWOHmjZt6uiSAQAwxXTzdHYQ9qVLl/JMs0SS5eTkmC0XAACHM33Y1hKEnZKSooULFyouLk6jR4+2BmEnJSUpPj7eGoR98uRJ6zhOSxB21apVrUHYlnksunTpop07d2rChAk6cOCAkpOTNWjQIIWGhqpRo0aO23IAAOxkes/T2UHYbdu21YIFCzRlyhRNmTJFJUqUULNmzbR27VoFBATYu50AADiM6attb4UgbMk5V1cBAG49bnG1LQAAdzqaJwAAJpk653krBmGT5wnY4qYCQPGx5wkAgEk0TwAATHLLPM/FixcrIiJCJUqUUGhoqKZOneqoMgEAKDaH3Z4vd55nw4YNdfbsWY0ePVrdunVTUlKSJNs8z5o1a2rv3r2KiYnRxYsXFRsbK0las2aN+vfvr7i4OD300EP69ddfFRMTo4CAAI0cOdJR5QIAYDe3y/Ps16+fsrKylJCQYJ0nLi5OU6ZM0ZEjR/JdT34s43pIVQFsccEQ7jRuMc7T2XmemZmZ8vf3t5knICBAx44dU1paWoHLyczMVEZGhs0DAABnMN08Q0JCNH36dNWpU0f9+/fX008/renTp+eZ7/Lly3rxxRfVt2/fAjv9wYMHFRcXp+HDh1undezYUcuWLdM333yjnJwc7d+/X9OmTZMk/fHHHwXWRRg2AMBV3C7PMyYmRiNHjlTXrl3l6+urpk2b6vHHH79WrGfB5Y4dO1bp6enWx9GjR81uGgAAReLwoSq58zw3bNhgOs/Tw8NDkydP1oULF5SWlqaTJ0+qSZMmkqSwsLAC10sYNgDAVdwuz9PCy8tLVapUkSQtXLhQzZo1U8WKFc2WCwCAw5lunpY8z+HDhys5OVlxcXGaNm2aNc8zOTlZq1evtuZ5SlK5cuXk6+trzfMMDQ215nlaVK5cWZJ0+vRpLVmyRFFRUbp8+bJmz56thIQEbdmyxUGbDABA8bhdnqd07YreMWPGyDAMNWvWTJs3b7YeugUA4GYjzxMAcFtzi3GeAADc6WieAACYRJ4nADgQtz+8M7DnCQCASTRPAABMcmjzLEqmpyR169ZNVatWlb+/v4KDg/Xkk0/mmeenn35SZGSk/P39FRISoilTpjiyVAAA7ObQ5pk70zM5OVnLli1TSkqKunXrZjNfmzZttHjxYqWkpGjp0qU6dOiQevXqZX09IyNDDz30kEJDQ7Vr1y5NnTpV48ePz3MrPwAAbgZT4zwl52R6rly5Uj169FBmZqZ8fHz00Ucf6ZVXXtHJkyfl6+srSXrppZf05Zdf6rfffitSneR5ArgZuGDI/bjNOE9HZnqeOXNG8fHxat68uXx8fCRJ27dvV6tWrayNU7oWVZaSkqKzZ8/muxzyPAEArmJX83REpueLL76owMBAlS9fXkeOHNGKFSusr508eVKVKlWymd/y3HK/3OuR5wkAcBW7mqcjMj1feOEF/fjjj1q/fr28vLw0YMAAmTyCbIM8TwCAq5i+MXxR5M703LRpU77HmCtUqKAKFSqodu3aqlevnkJCQvT999+rWbNmqly5sk6dOmUzv+W5JX3len5+fvLz83P8xgAAcB279jyLmum5cePGfDM9r5eTkyPp2nlL6dqe7LfffqusrCzrPBs2bFCdOnVUtmxZe0oGAMBh7GqelkzPlJQULVy4UHFxcRo9erQ10zMpKUnx8fHWTM+TJ0/qypUrkq413vfff1+7d++27pn27dtXNWrUULNmzSRJ/fr1k6+vr4YMGaJffvlFixYt0rvvvqvnnnvOcVsOAICd7DpsW5xMzxIlSmjZsmUaN26cLl68qODgYHXq1Emvvvqq9bBrUFCQ1q9frxEjRqhx48aqUKGCXn/9dQ0bNqyYmwsAQPHZNc7zVsj0JM8TACC50ThPAADuZDRPAABMMn3O81bM9AQAwJGcMs7TnRCGDQvuOQrAUThsCwCASQ5rno7K8rx8+bKio6MVHh4ub29v9ejRw1ElAgDgEA5rno7K8szOzlZAQIBGjRql9u3bO6o8AAAcxtQ4T1dkeeYWHR2tc+fO6csvvzSzTZLI80RenPME7kxuMc7T2Vme9iLPEwDgKqabp7OzPO1FnicAwFVMN093zPKUyPMEALiOw8d5FjfL017keQIAXMV08yxqlmdiYqJdWZ4AALg7083TkuU5fPhwJScnKy4uTtOmTbNmeSYnJ2v16tXWLE9JKleunHx9fbVjxw7t3LlTLVu2VNmyZXXo0CG99tprNlmekrRv3z5duXJFZ86c0fnz57V7925JeWPOAAC4GUw3T2dneUrSww8/rLS0NOvzRo0aSVKxz4sCAOAIppunj4+PZsyYkedCoGrVqt2wuYWHh2vTpk03XEdqaqrZsgAAcJnb/sbwe9/oSBg2AMChuDE8AAAmmdrzJMsTAIA74LAteZ64HXBfXsC9cNgWAACTHNo8HZXpuXnzZnXv3l3BwcEKDAxURESE4uPjHVkqAAB2c2jzdFSm57Zt23Tfffdp6dKl+umnnzRo0CANGDBAq1evdmS5AADYxVSep+T6TE+LLl26qFKlSpo1a1aR6iTPE7cTznkC9nOLPE/p5mR6pqenq1y5cgW+Tp4nAMBV7Gqers70XLx4sXbu3KlBgwYVOA95ngAAV7Greboy0zMxMVGDBg3SZ599pnvvvbfAmsjzBAC4ilPGeToq03PLli165JFHNH36dA0YMKDQdZLnCQBwFbuapysyPTdv3qyuXbtq8uTJGjZsmD1lAgDgFHY1T2dneiYmJqpr164aPXq0evbsaV2Gr69voRcNAQDgCnad88yd6TlixAhrpufx48e1cuVKHTt2TBEREQoODrY+tm3bJknWTM927dqpTp06GjJkiO677z5t2bLFeth17ty5unTpkiZNmmSzjMcee8xxWw4AgJ3sGucZERGhGTNmOKsmh2CcJ24njPME7OeMcZ63/Y3hyfMEADgaN4YHAMAk03ueZHoCAO50t/1hW/I8gZuHc7W4XXHYFgAAk2ieAACY5JZh2CkpKWrTpo0qVaokf39/hYWF6dVXX1VWVpYjywUAwC5uGYbt4+OjAQMGaP369UpJSdGMGTP02Wefady4cY4sFwAAu9wyYdjPPfecdu7cqa1btxapTm6SANx8XDAEd3DHhmEfPHhQa9euVevWrQusiTBsAICruHUYdvPmzeXv769atWopMjJSEyZMKLAmwrABAK7i1mHYixYtUnJyshYsWKCvvvpKsbGxBdZEGDYAwFXcOgzbsvdYv359ZWdna9iwYXr++efl5eWVZ3mEYQMAXMWuPc+ihmFv3LjR7jDs/ObJysqyzgsAwM3ilmHY8fHx8vHxUXh4uPz8/JSUlKSxY8fqn//8Z4EXFQEA4Cp2Nc/cYdheXl7WMOy0tDStXLlSkhQREWHznsTEREVFRVnDsMeNG6eLFy8qODhYnTp10quvvmo97Ort7a3Jkydr//79MgxDoaGhGjlypJ599tlibi4AAMV324dhO3JcDwDg1uM24zwBALiT0TwBADDptg/DJs8TsMUt84DiY88TAACTaJ4AAJjklnmeuR08eFClSpUq8MbyAAC4mlvmeVpkZWWpb9++ioyMdGSZAAAUi1vneb744os6ceKE2rVrp2eeeUbnzp0rcp3keQL544Ih3GncZpynK/I8N23apISEBH3wwQdFqok8TwCAq7hlnufff/+t6OhozZkzp8h/JZDnCQBwFbfM84yJiVG/fv3UqlWrItdEnicAwFXcMs9z06ZNWrlypTX82jAM5eTkyNvbW59++qkGDx6cZ3nkeQIAXMWu5lnUPM/ExES78jy3b99usxe7YsUKTZ48Wdu2bVOVKlXsKRkAAIdxyzzPevXq2awvKSlJnp6e1qt8AQC4mdwyzxMAAHdGnicA4LbmNuM8AQC4k9E8AQAw6bbP8wQAwNGcMs7TnRCGDbgf7q+LWx2HbQEAMMlhzdORWZ6GYSg2Nla1a9eWn5+fqlSpookTJzqqVAAAisVhzdORWZ6jR4/WzJkzFRsbq99++00rV65UkyZNHFUqAADFYmqcpyuyPH/99Vfdd9992rt3r+rUqWPnZpHnCbgzznnCldxinKezszxXrVqlsLAwrV69WtWrV1e1atU0dOhQnTlzptC6yPMEALiK6ebp7CzP33//XWlpaUpISNAXX3yhOXPmaNeuXXkO7V6PPE8AgKuYbp7OzvLMyclRZmamvvjiC0VGRioqKkqff/65EhMTlZKSUmBd5HkCAFzF4eM8i5vlGRwcLG9vb9WuXds6vyVl5ciRIwWeByXPEwDgKqb3PIua5blx40a7sjxbtGihq1ev6tChQ9Z59u/fL0kKDQ01Wy4AAA5nes/T2Vme7du31/3336/BgwdrxowZysnJ0YgRI9ShQwebvVEAAG4W083T2Vmenp6eWrVqlZ5++mm1atVKgYGB6ty5s6ZNm+aAzQUAoPhMj/O8FbI8JcZ5Au6McZ5wJWeM87ztbwy/942OhGEDAByKG8MDAGCSqT1PsjwBALgDDtuS54nbHecPAdfjsC0AACY5tHk6KtNz/Pjx8vDwyPMIDAx0ZLkAANjFoc3TUZmeY8aM0R9//GHzqF+/vnr37u3IcgEAsIupcZ6SazI9r7dnzx5FRETo22+/VWRkZJHqZJwn7hSc8wQK5xZ5npLzMz2vN3PmTNWuXbvQxkmeJwDAVexqns7O9Lx+GfHx8RoyZEihNZHnCQBwFbuap7MzPXNbvny5zp8/r4EDBxZaE3meAABXcco4z+JmeuY2c+ZMde3aVZUqVSp0neR5AgBcxa7mWdRMz8TERLsyPS0OHz6sxMREa1oLAADuwK7m6exMT4tZs2YpODhYnTt3Lv6WAgDgIHY1T2dnekrX9kbnzJmj6OhoeXl5FWMTAQBwLLuap4+Pj2bMmJHnQqBq1arle9FPbuHh4dq0adMN1+Hp6clFPwAAt3Tb3xiePE8AgKNxY3gAAEwyvedJpicA4E532x+2Jc8TAG5N7nzfZg7bAgBgEs0TAACTHNY8HRWELUnr1q1T06ZNVapUKVWsWFE9e/ZUamqqo0oFAKBYHNY8HRWEffjwYXXv3l1t27bV7t27tW7dOp0+fVqPPfaYo0oFAKBYTIVhuyIIe8mSJerbt68yMzPl6Xmtt69atUrdu3cvMCw7P4RhA8CtzVEXDLlFGLazg7AbN24sT09PzZ49W9nZ2UpPT9e8efPUvn37QhsnYdgAAFcx3TydHYRdvXp1rV+/Xi+//LL8/PxUpkwZHTt2TIsXLy60LsKwAQCuYrp5OjsI++TJk4qJidHAgQO1c+dObdmyRb6+vurVq1eh980lDBsA4CoOv0lCcYOwP/jgAwUFBWnKlCnW+efPn6+QkBDt2LFDTZs2zXe9hGEDAFzFdPN0dhD2pUuXrBcKWVgiySzzAgBwM5k+bGsJwk5JSdHChQsVFxen0aNHW4Owk5KSFB8fbw3CPnnypK5cuSLpWuN9//33tXv3buuead++fW2CsLt06aKdO3dqwoQJOnDggJKTkzVo0CCFhoaqUaNGjt16AADsYLp55g7CHjFihDUI+/jx41q5cqWOHTumiIgIBQcHWx/btm2TJGsQdrt27VSnTh0NGTJE9913n7Zs2WI95Nq2bVstWLBAX375pRo1aqROnTrJz89Pa9euVUBAgGO3HgAAO5ge5xkREaEZM2Y4syaHcMa4HgDArcctxnkCAHCno3kCAGCSqattb8UgbPI8gVuTO2c5Aux5AgBgEs0TAACT3DLPc/HixYqIiFCJEiUUGhqqqVOnOqpMAACKze3yPNesWaP+/fvrqaee0t69e/Xhhx9q+vTpev/99x1VKgAAxeJ2eZ79+vVTVlaWEhISrPPExcVpypQpOnLkSL7ryQ95nsCtjQuG4ChuMc7T2XmemZmZ8vf3t5kvICBAx44dU1paWoF1kecJAHAVt8vz7Nixo5YtW6ZvvvlGOTk52r9/v6ZNmyZJ+uOPPwqsizxPAICruF2eZ0xMjEaOHKmuXbvK19dXTZs21eOPP36tWM+CyyXPEwDgKm6X5+nh4aHJkyfr7bff1smTJ1WxYkV98803kqSwsLAC10ueJwDAVdwuz9PCy8tLVapUkSQtXLhQzZo1U8WKFc2WCwCAw5lunpY8z+HDhys5OVlxcXGaNm2aNc8zOTlZq1evtuZ5SlK5cuXk6+urHTt2aOfOnWrZsqXKli2rQ4cO6bXXXrPJ8zx9+rSWLFmiqKgoXb58WbNnz1ZCQoK2bNni2C0HAMBOpptn7jxPLy8va55nWlqaVq5cKUmKiIiweU9iYqKioqKseZ7jxo3TxYsXFRwcrE6dOunVV1+1OeQ6d+5cjRkzRoZhqFmzZtq8ebOaNGlSzE0FAMAxyPMEANzW3GKcJwAAdzqaJwAAJpHnCaBIuF0e8P+w5wkAgEk0TwAATHJo8yxqpqdFZmamIiIi5OHhod27d9u89tNPPykyMlL+/v4KCQnRlClTHFkqAAB2c2jzLGqmp8W///1v3X333XmmZ2Rk6KGHHlJoaKh27dqlqVOnavz48fr0008dWS4AAHYxfZOEwjI9g4KCtGHDBpv533//fTVp0kRHjhyxyfRcs2aN1q9fr6VLl2rNmjU274mPj9eVK1c0a9Ys+fr66t5779Xu3bv1n//8R8OGDbNnOwEAcBi79jyLm+l56tQpxcTEaN68eSpRIu+VsNu3b1erVq3k6+trndaxY0elpKTo7Nmz+a6HPE8AgKvY1TyLk+lpGIaio6P11FNP6YEHHsh3+SdPnlSlSpVsplmeW+6Xez3yPAEArmJX8yxOpmdcXJzOnz+vsWPHFqPsvMjzBAC4isPzPKXCMz03bdqk7du358nefOCBB9S/f3/NnTtXlStX1qlTp2xetzyvXLlyvuskzxMA4Cp2Nc/iZHq+9957euutt6zPT5w4oY4dO2rRokV68MEHJV3bk33llVeUlZUlHx8fSdKGDRtUp04dlS1b1p6SAQBwGLsO21oyPVNSUrRw4ULFxcVp9OjR1kzPpKQkxcfHWzM9T548qStXrkiSqlatqgYNGlgftWvXliTVqFFD99xzjySpX79+8vX11ZAhQ/TLL79o0aJFevfdd/Xcc885aLMBALCfXXuexcn0LIqgoCCtX79eI0aMUOPGjVWhQgW9/vrrDFMBALgFU3me0q2T6UmeJwBAIs8TAAC3QPMEAMAk0+c8b8VMTwAAHMkp4zzdCWHYuNMRYg04HodtAQAwyWHN01FZnpcvX1Z0dLTCw8Pl7e2tHj16OKpEAAAcwmHN01FZntnZ2QoICNCoUaPUvn17R5UHAIDDmDrn6Yosz8DAQOuN5L/77judO3fOrg0DAMBZTO95OjvL017keQIAXMV083R2lqe9yPMEALiK6ebpjlmeEnmeAADXcfg4z+JmedqLPE8AgKuYbp7OzvIEAMDdmW6elizP4cOHKzk5WXFxcZo2bZo1yzM5OVmrV6+2ZnlKUrly5eTr62tzxa0klSxZUpJtlqck7du3T1euXNGZM2d0/vx56zjQ62POAAC4GUw3T2dneUrSww8/rLS0NOvzRo0aSbp2wREAADeb6ebp4+OjGTNm2FwIJEnVqlUz3dwKek9qaqrZsgAAcJnb/sbwe9/oSBg2AMChuDE8AAAmmdrzJMsTAIA74LAteZ4A7jRkuDofh20BADCJ5gkAgEkObZ6OCsROSUlRmzZtVKlSJfn7+yssLEyvvvqqsrKyHFkuAAB2ceg5z9yB2A0bNtTZs2c1evRodevWTUlJSXnmtwRi79mzx2a6j4+PBgwYoPvvv19lypTRnj17FBMTo5ycHL399tuOLBkAANNMN09XBGKHhYUpLCzM+jw0NFSbN2/W1q1bC6wrMzNTmZmZ1ufkeQIAnMWuw7auDsQ+ePCg1q5dq9atWxc4D3meAABXsat5uioQu3nz5vL391etWrUUGRmpCRMmFDgveZ4AAFexq3m6KhB70aJFSk5O1oIFC/TVV18pNja2wHn9/PxUunRpmwcAAM7glJskOCoQ23LotX79+srOztawYcP0/PPPy8vLyxllAwBQJHY1z5sRiJ2Tk6OsrCzl5OTQPAEAN5VdzdPZgdjx8fHy8fFReHi4/Pz8lJSUpLFjx+qf//ynfHx8irO9AAAUm13N09mB2N7e3po8ebL2798vwzAUGhqqkSNH6tlnn7WnXAAAHMrDMJlgHRUVpYiICM2YMcNZNTlERkaGgoKClJ6ezsVDAHAHc0Y/4N62AACYRPMEAMAk0+c8b7VAbPI8AfuRCwnkjz1PAABMonkCAGCSW+Z55nbw4EGVKlXK5sbyAADcTA5tnrnzPJOTk7Vs2TKlpKSoW7du+c5vyfMsSFZWlvr27avIyEhHlgkAQLG4ZZ6nxauvvqq6deuqXbt22rZtm9lSAQBwCrfN89y0aZMSEhL0wQcfFKmmzMxMZWRk2DwAAHAGt8zz/PvvvxUdHa05c+YU+W4QhGEDAFzFLfM8Y2Ji1K9fP7Vq1arINRGGDQBwFacMVcmd57lhw4YC8zy9vb1Vs2ZNSdfyPAcOHGidJzY2Vt7e3vL29taQIUOUnp4ub29vzZo1K991EoYNAHAVt8zz3L59u81e7IoVKzR58mRt27ZNVapUsadkAAAcxi3zPOvVq2czT1JSkjw9Pa1X+QIAcDO5ZZ4nAADujDxPAMBtjTxPAADcAM0TAACTyPME4NbIFIU7Ys8TAACTaJ4AAJjklnmeqamp8vDwyPP4/vvvHVkuAAB2ces8z40bN+qPP/6wPho3buzIcgEAsItb53mWL19elStXNlsiAABO5bZ5npLUrVs33XXXXWrZsqX1zkUFIc8TAOAqbpnnWbJkSU2bNk0JCQn66quv1LJlS/Xo0aPQBkqeJwDAVdwyz7NChQp67rnn9OCDD+of//iH3nnnHT3xxBOaOnVqge8hzxMA4CpumeeZnwcffFAHDx4s8HXyPAEAruKWeZ752b17t4KDg+0pFwAAh3LLPM+5c+fK19dXjRo1kiQtW7ZMs2bNKvCiJAAAXMlt8zzffPNNpaWlydvbW3Xr1tWiRYvUq1cve8oFAMChyPMEANzWyPMEAMAN0DwBADDpts/zBADA0ey6YOhWQhg2cOciSBvOwmFbAABMcljzdFSWp3Tt/rexsbGqXbu2/Pz8VKVKFU2cONFRpQIAUCwOO2ybO8uzYcOGOnv2rEaPHq1u3bopKSkpz/yWLM89e/bkeW306NFav369YmNjFR4erjNnzujMmTOOKhUAgGIx1TxdkeX566+/6qOPPtLevXtVp04dSVL16tXt2jgAAJzB9GFbZ2d5rlq1SmFhYVq9789E5AAAIABJREFUerWqV6+uatWqaejQoTfc8yTPEwDgKqabp7OzPH///XelpaUpISFBX3zxhebMmaNdu3bd8NZ85HkCAFzFdPN0dpZnTk6OMjMz9cUXXygyMlJRUVH6/PPPlZiYqJSUlALfR54nAMBVHD5UpbhZnsHBwfL29lbt2rWt76tXr56ka2kuBSHPEwDgKqavtnV2lmeLFi109epVHTp0SDVq1JAk7d+/X5IUGhpqtlwAABzOdPN0dpZn+/btdf/992vw4MGaMWOGcnJyNGLECHXo0MFmbxQAgJvFdPN0dpanp6enVq1apaefflqtWrVSYGCgOnfurGnTppktFQAApzDdPH18fDRjxgybC4EkqVq1ajIZDVrge+6++24tXbrUbGkAALjEbX9j+L1vdOTiIQCAQ3FjeAAATDK150mWJwAAd8BhW/I8ARQF2Z8wg8O2AACYRPMEAMAkhzZPRwVijx8/Xh4eHnkegYGBjiwXAAC7OLR55g7ETk5O1rJly5SSkqJu3brlO78lEPt6Y8aM0R9//GHzqF+/vnr37u3IcgEAsIvpC4ZcEYhdsmRJ6637JGnPnj3at2+fPv744wLryszMVGZmpvU5eZ4AAGexa8/T2YHY15s5c6Zq166tyMjIAuchzxMA4Cp2NU9nB2Jfv4z4+HgNGTKk0PnI8wQAuIpdzdPZgdi5LV++XOfPn7fmfRaEPE8AgKs4ZahKcQOxc5s5c6a6du2qSpUqOaNUAABMs+sOQ84OxLY4fPiwEhMTrVFnAAC4A7uap7MDsS1mzZql4OBgde7c2Z4yAQBwCruap7MDsSUpJydHc+bMUXR0tLy8vOwpEwAAp/AwTCZYR0VFKSIiQjNmzHBWTQ6RkZGhoKAgpaenc/EQANzBnNEPuLctAAAm0TwBADDJ9DnPWy0QmzxPADdClifMYs8TAACTaJ4AAJjksObpqCxPSVq3bp2aNm2qUqVKqWLFiurZs6dSU1MdVSoAAMXisObpqCzPw4cPq3v37mrbtq12796tdevW6fTp03rsscccVSoAAMVi6oIhV2R57tq1S9nZ2Xrrrbfk6Xmtt48ZM0bdu3dXVlaWfHx87NpQAAAcxfSep7OzPBs3bixPT0/Nnj1b2dnZSk9P17x589S+fftCG2dmZqYyMjJsHgAAOIPp5unsLM/q1atr/fr1evnll+Xn56cyZcro2LFjWrx4caF1EYYNAHAV083T2VmeJ0+eVExMjAYOHKidO3dqy5Yt8vX1Va9evVTYnQQJwwYAuIpdN4YvTO4sz02bNhWY5ZnbAw88oP79+2vu3Ln64IMPFBQUpClTplhfnz9/vkJCQrRjxw41bdo03/X6+fnlWS4AAM5gunk6O8vz0qVL1guFLCypKjk5OWbLBQDA4Uw3T2dneXbp0kXTp0/XhAkT1LdvX50/f14vv/yyQkND1ahRo+JuLwAAxWa6eTo7y7Nt27ZasGCBpkyZoilTpqhEiRJq1qyZ1q5dq4CAALPlAgDgcKbyPG+VLE+JPE8AwDXkeQIA4AZongAAmGTqnOetluUpkecJOBM5mLhTsecJAIBJNE8AAExyyzzPxYsXKyIiQiVKlFBoaKimTp3qqDIBACg2t8vzXLNmjfr376+nnnpKe/fu1Ycffqjp06fr/fffd1SpAAAUi+lxngXleea+WbzFzp071aRJE6WlpeXJ83zuuee0dOlS3Xvvvfrxxx//P/buPaqqOuH/+AcEjqhAmhfMTiDT8tbjfSLJLG9JpYOudCqtNFPURo10LHVGLZlJU0ItLJunvORIpibPY1JeMNB80hqQkby00EoxJzFNQ53kiODvD3+c4Sgg+7DPcYvv11pnLQ9ne853//Vd+/Z9OxdWGDJkiIqKirRmzRrn9klJSZo7d66OHDlS7u+Up/S5HvsLq7lhCPAQbhjCjcASz3l6uufpcDhUu3Ztl78FBgbq6NGjysvLq3Bc9DwBAN5iuZ5ndHS0UlJS9Nlnn6mkpEQHDhxQYmKiJOnYsWMVjoueJwDAWyzX84yNjdW4cePUr18/BQQEqEuXLnriiScuD9a34uHS8wQAeIvpj6qU7XmmpaVV2PP08/PTnXfeKelyz3PYsGGSJB8fH82ZM0fnzp1TXl6e8vPzFRkZKUmKiIio8HdtNpuCg4NdXgAAeILlep6latWqpWbNmkmSVq5cqaioKDVq1MjocAEAMJ3lep4nT57URx99pO7du6uwsFBLly7VmjVrtG3bturuKwAAprBcz1O6fEfvpEmTdOnSJUVFRWnr1q3OU7cAAFxv9DwBADWaJZ7zBADgZsfkCQCAQTW+5wkAgNkM3zB0oyGGjZsRa84CnsVpWwAADDJ18qxq0zM8PFw+Pj4ur9dee81lm6+//lrdunVT7dq1ZbfbNXfuXDOHCgCA20w9bVu26dm+fXudPn1acXFxiomJUVZWlsu28fHxio2Ndb4PCgpy/vvMmTPq06ePevfurXfeeUd79uzRs88+q1tuuUWjRo0yc8gAABhmePKsrOkZEhKitLQ0l+0XLlyoyMhIHTlyxGWFoaCgIIWGhpb7G8nJybpw4YKWLFmigIAA3XXXXdq9e7fmzZvH5AkAuO7cOm1b3aanJL322mu69dZb1bFjRyUkJOjixYvOz3bu3Kn7779fAQEBzr9FR0crNzdXp0+fLvd36HkCALzFrdO2pU1PHx8ftWzZUnv27NH8+fNdTsNK5Tc9Jen5559Xp06d1KBBA+3YsUNTp07VsWPHNG/ePElSfn6+mjdv7vJdTZo0cX5Wv379q8Y0e/ZszZw5053dAQDAELeOPKvT9JSkiRMnqnv37mrXrp3GjBmjxMREJSUlyeFwuLkb9DwBAN7jkec8yzY909PTr7mW4D333KOLFy/q8OHDatmypUJDQ3X8+HGXbUrfV3Sd1GazyWazmbMDAABUwq0jz6o2Pbds2XJV07M8u3fvlq+vrxo3bizp8pHs559/rqKiIuc2aWlpatmyZbmnbAEA8Ca3Js/Spmdubq5WrlyppKQkxcXFOZueWVlZSk5OdjY98/PzdeHCBUmXbwZasGCBcnJy9P333ys5OVkTJkzQU0895ZwYhwwZooCAAI0YMUL79u3TqlWr9MYbb2jixInm7TkAAG5y67RtdZqeNptNH374oV555RU5HA41b95cEyZMcJkYQ0JCtHnzZo0dO1adO3dWw4YNNWPGDB5TAQBYgluTp7+/vxYsWHDVjUDh4eG6Vh60U6dO+vLLL6/5G+3atdP27dvdGR4AAB5V4xeG3zszmhg2AMBULAwPAIBBho88aXoCAG52Nf60LT1P4NrofwLGcNoWAACDTJs8zWp5FhYW6plnnlHbtm3l5+enAQMGmDVEAABMYdrkWbblmZ2drZSUFOXm5iomJuaqbePj43Xs2DHna/z48c7PiouLFRgYqOeff169e/c2a3gAAJjG0DVPb7Q869at63x+9IsvvtAvv/xiaIcAAPA0w0eenm55uoueJwDAWwzfbevplqe76HkCALzF8JGnFVueEj1PAID3mP6cZ3Vbnu6i5wkA8BbDk2dVW54ZGRlutTwBALA6w5Nnactz9OjRys7OVlJSkhITE50tz+zsbKWmpjpbnpLUoEEDBQQEaOfOnfrqq6/Uo0cPBQUFaefOnVe1PCVp//79unDhgk6dOqWzZ89q9+7dkq7OnAEAcD0Ynjw93fKUpEceeUR5eXnO9x07dpSka+bOAADwBsOTpzdanocPHzY6LAAAvKbGLwxPzxMAYDYWhgcAwCBDR560PAEAuAlO29LzBKqOridQNZy2BQDAICZPAAAMMnXyNCuIvXXrVvXv319NmzZV3bp11aFDByUnJ5s5VAAA3GbqNc+yQez27dvr9OnTiouLU0xMjLKysly2jY+PdymxBAUFOf+9Y8cOtWvXTpMnT1aTJk2UmpqqoUOHKiQkRP369TNzyAAAGGZ48vRGEPtPf/qTy/u4uDht3rxZKSkpTJ4AgOvOrdO21yOIXVBQoAYNGlT4OTFsAIC3uHXa1ttB7NWrVyszM1N/+9vfKhwTMWwAgLf4XDK42nr37t0VERGhJUuWOP+2bt06DRo0SIWFhapVq5akyzcPDRw4UEePHtXWrVsrXSJvyZIlGj16tM6dO3dVkzMjI0P9+vXTokWLNHTo0Aq/w+FwuAS1z5w5I7vdLvsLq3nOE6ginvNETXTmzBmFhISooKDAtOVaPbJIgllB7G3btul3v/ud5s+fX+nEKRHDBgB4j1uTpzeC2Fu3blW/fv00Z84cjRo1yp1hAgDgEW5Nnp4OYpeeqo2Li9PAgQOd3xEQEFDpTUMAAHiDW5Onp4PY77//vn799VfNnj1bs2fPdv79gQceYHF6AMB159YNQx06dNCCBQs8NSZTeOICMQDgxuOJ+YC1bQEAMIjJEwAAgwxf87zRrjnS88TNiOc1Ac/iyBMAAIOYPAEAMMiSPc/c3Fz16NFDTZo0Ue3atRUREaFp06apqKjIzOECAOAWS/Y8/f39NXToUHXq1Em33HKLcnJyFBsbq5KSEs2aNcvMIQMAYJgle54RERGKiIhwvg8LC9PWrVu1fft2o8MFAMB0N0TP89tvv9XGjRv1wAMPVLgNPU8AgLdYuud57733Kjs7Ww6HQ6NGjVJ8fHyFY6LnCQDwFkv3PH/44QedPXtWOTk5evHFF/X888/rpZdeKvc76HkC/8FznsB/3HQ9T7vdLklq06aNiouLNWrUKP3xj390TtBl0fMEAHiLW9c8q9rz3LJli9s9zyuVlJSoqKhIJSUl7gwZAADTWLLnmZycLH9/f7Vt21Y2m01ZWVmaOnWqHn/8cfn7+5u39wAAuMGSPU8/Pz/NmTNHBw4c0KVLlxQWFqZx48ZpwoQJ1dhVAADMQc8TAFCj0fMEAMACmDwBADCoxvc8AQAwm0ee87QSYtgAiyYAZuO0LQAABlmy51nWt99+q6CgoKsWlgcA4HoxdfIs2/PMzs5WSkqKcnNzFRMTc9W28fHxOnbsmPM1fvz4q7YpKirS4MGD1a1bNzOHCQBAtViy51lq2rRpatWqlXr16qUdO3YYHSoAAB5h2Z5nenq61qxZo7feeqtKY6LnCQDwFkv2PH/++Wc988wzWrFiRZVXg6DnCQDwFreOPLt06SIfHx/n+6ioKB08eFDFxcXOv5XWVS5duqRFixa5/P+JEyeqe/fuateuncaMGaPExEQlJSU5e5yxsbEaMmSI7r///iqPaerUqSooKHC+fvjhB3d2DQCAa/LIoyple55paWmGep7S5VO2r7/+uvz8/OTn56cRI0aooKBAfn5+LhHusmw2m4KDg11eAAB4glunbava88zIyHCr57lz506Xo9h169Zpzpw52rFjh5o1a+bOkAEAMI0le56tW7d2+b2srCz5+vo67/IFAOB6smTPEwAAK6vxPU/7C6tZ2xY3Pda2xc3MEz3PGr8w/N6Z0dw8BAAwFQvDAwBgED1PAAAMqvGnbel5AjcertHC6jhtCwCAQZbseR4+fPiqz318fPTll1+aOVwAANxi6mnbsj3P9u3b6/Tp04qLi1NMTIyysrJcto2Pj3dZSD4oKOiq79uyZYvuuusu5/uqrFYEAICnWbrneeutt15zGwAAvM2yPU9JiomJUePGjXXfffc5Vy6qCD1PAIC3WLLnWa9ePSUmJqpr167y9fXV2rVrNWDAAP3v//6vYmJiyh0TPU8AgLe4tTxfRESESxps3bp1GjRokAoLC1WrVi1Jl28eGjhwoI4ePaqtW7dWusrPkiVLNHr0aJ07d042m63cbYYOHapDhw5p+/bt5X7ucDicPVDp8nJMdrud5fmAGxCPqsBMnliez5I9z4q2+fbbbyv8nJ4nAMBbLNnzrGibpk2bujNcAABMZcme5/vvv6+AgAB17NhRkpSSkqIlS5ZUeFMSAADeZNme51/+8hfl5eXJz89PrVq10qpVqzRo0CA3dxMAAPPQ8wRgOdwwBDPR83QDPU8AgNlYGB4AAIPoeQIAYFCNP21LzxPwDK5L4mbGaVsAAAxi8gQAwCDTJk+zQtiSdOnSJb3++utq0aKFbDabmjVrpldffdWsoQIAUC2mXfM0M4QdFxenzZs36/XXX1fbtm116tQpnTp1yqyhAgBQLYYmT2+EsL/55hstWrRIe/fuVcuWLSVJzZs3v+bYyquqAADgCYZP23o6hL1+/XpFREQoNTVVzZs3V3h4uEaOHHnNI8/Zs2crJCTE+bLb7UZ3DQCAKjF82tbTIezvv/9eeXl5WrNmjZYvX67i4mJNmDBBgwYNUnp6eoXjmjp1qsv6uKU9TwAAzGZ48uzSpYt8fHyc76OiopSYmKji4mKXEPZjjz2mS5cuadGiRS7/v+wE165dOwUEBGj06NGaPXu2bDabSkpK5HA4tHz5crVo0UKStHjxYnXu3Fm5ubnOU7lXstlsFYa0AQAwk+mPqlQ3hN20aVP5+fk5J05Jat26taTLKTQAAK43w5NnVUPYW7ZscSuE3bVrV128eFHfffedc5sDBw5IksLCwowOFwAA0xk+bevpEHbv3r3VqVMnPfvss1qwYIFKSko0duxYPfjggy5HowAAXC+GJ09Ph7B9fX21fv16jR8/Xvfff7/q1q2rhx9+WImJidXcVQAAzGEohn2jhLAlz8RPAQA3Hk/MB6xtCwCAQUyeAAAYZOia540YwqbniRsVvUzAujjyBADAICZPAAAMMnXyNKvp+corr1z1uY+Pj+rWrWvmcAEAcItpPU/JvKbnpEmTNGbMGJfte/XqpbvvvtvM4QIA4BbDk6c3mp716tVTvXr1nO9zcnK0f/9+vfPOO0aHCwCA6dw6bevppueV3nvvPbVo0ULdunWrcBuHw6EzZ864vAAA8AS3Ttt6uul55XckJydrypQplY5p9uzZmjlzpju7AwCAIYaW55Mun7aNiIjQkiVLnH9bt26dBg0apMLCQpem58CBA3X06FFt3bq10iWRlixZotGjR+vcuXNXNTlXrlypoUOH6ujRo2rSpEmF3+FwOORwOJzvS2PY9hdW85wnbkg85wmYwxPL85l6w1Cpsk3P9PR0Q03PK2PX7733nvr161fpxCkRwwYAeI9bk2dVm54ZGRluNT1LHTp0SBkZGc5aCwAAVuDW5OnppmepJUuWqGnTpnr44Yerv6cAAJjErcnT001PSSopKdGyZcv0zDPPOK+jAgBgBW7dMHQjND3peQIAJHqeAABYApMnAAAGGb7meaM1Pel5Au7hOVOgYhx5AgBgEJMnAAAGmTZ5mtXylKRNmzapS5cuCgoKUqNGjTRw4EAdPnzYrKECAFAtpk2eZVue2dnZSklJUW5urmJiYq7aNj4+XseOHXO+xo8f7/zs0KFD6t+/v3r27Kndu3dr06ZNOnnypB599FGzhgoAQLUYumHIGy3PXbt2qbi4WH/961/l63t5bp80aZL69++voqIi+fv7G9pBAADMZvjI09Mtz86dO8vX11dLly5VcXGxCgoK9Pe//129e/eudOKk5wkA8BbDj6p4uuXZvHlzbd68WY899phGjx6t4uJiRUVF6dNPP610XPQ8AQDeYmh5Pm+0PPPz83X//fdrwIABGjx4sM6ePasZM2bIz89PaWlp8vHxKfd76HkC5uI5T9QUN0TPs7otz7feekshISGaO3euc5sVK1bIbrfrq6++UpcuXcr9HnqeAABvMTx5errl+euvvzpvFCpVekRbUlJidLgAAJjO8A1DpS3P3NxcrVy5UklJSYqLi3O2PLOyspScnOxseebn5+vChQuSpJ07d2rBggXKycnR999/r+Tk5Ktann379lVmZqbi4+N18OBBZWdna/jw4QoLC1PHjh3N3XsAANxg+MjT0y3Pnj176oMPPtDcuXM1d+5c1alTR1FRUdq4caMCAwOrubsAAFSf4RuGboSWp0TPEwBwGT1PAAAsgMkTAACDDF3zvNFangAAeILpz3laDTFsAFbGYhQ3Jk7bAgBgkCV7nqtXr1aHDh1Up04dhYWFKSEhwaxhAgBQbaadti3b82zfvr1Onz6tuLg4xcTEKCsry2Xb+Ph4l4Xkg4KCnP/esGGDnnzySSUlJalPnz765ptvFBsbq8DAQI0bN86s4QIA4DbL9Tz//ve/a8CAARozZowkKSIiQlOnTtWcOXM0duzYCheGBwDAWyzX83Q4HKpdu7bL9oGBgTp69Kjy8vIqHBc9TwCAtxiePEt7ni1bttSTTz6p8ePHa/78+VdtV1nP88MPP1RGRoZGjx6tWbNm6aWXXnJ+Hh0drZSUFH322WcqKSnRgQMHlJiYKEk6duxYheOaPXu2QkJCnC+73W501wAAqBLDk2eXLl1cTp1GRUXp4MGDKi4udv6ttK5y6dIlLVq0yOX/T5w4Ud27d1e7du00ZswYJSYmKikpydnijI2N1bhx49SvXz8FBASoS5cueuKJJy4P1rfi4U6dOlUFBQXO1w8//GB01wAAqBLTH1Up2/NMS0sz1POUJB8fH82ZM0fnzp1TXl6e8vPzFRkZKeny9c+K2Gw2BQcHu7wAAPAEy/U8S9WqVUvNmjWTJK1cuVJRUVFq1KiR0eECAGA6w5Nnac9z9OjRys7OVlJSkhITE509z+zsbKWmpjp7npLUoEEDBQQEaOfOnfrqq6/Uo0cPBQUFaefOnVf1PE+ePKmPPvpI3bt3V2FhoZYuXao1a9Zo27Zt5u45AABuslzPU7p8R++kSZN06dIlRUVFaevWrc5TtwAAXG+GJ09/f38tWLDgqhuBwsPDda00aKdOnfTll19Wuk3Dhg21c+dOo8MCAMBravzC8HtnRnPzEADAVCwMDwCAQfQ8AQAwqMaftqXniRsRjUfA2jhtCwCAQUyeAAAY5PUYtiR98sknuueeexQYGKj69etrwIABLp8fOXJEffv2VZ06ddS4cWO9+OKLLuUVAACuJ6/HsNeuXavY2FjNmjVLPXv21MWLF7V3717n58XFxerbt69CQ0O1Y8cOHTt2TEOHDpW/v79mzZpl1nABAHCbz6VrrWxQRmUx7PIi1ZmZmYqMjFReXp7uuOMOXbx4UeHh4Zo5c6ZGjBhR7m9s2LBB/fr1048//qgmTZpIkt555x1NnjxZJ06cUEBAQLn/z+FwOMssknTmzBnZ7XbZX1jNDUO44XDDEGCeM2fOKCQkRAUFBaY99+/VGHZ2drb+9a9/ydfXVx07dlTTpk318MMPuxx57ty5U23btnVOnNLlxueZM2e0b9++CsdFzxMA4C1ejWF///33kqRXXnlF06ZNU2pqqurXr6/u3bvr1KlTkqT8/HyXiVOS833pQvPloecJAPAWr8awS0pKJEl//vOfNXDgQHXu3FlLly6Vj4+P1qxZU539oOcJAPAar8awmzZtKklq06aN8282m00RERE6cuSIJCk0NFTHjx93+c7S96GhoWYPFwAAwwxPnlWNYW/ZsuWqGHbnzp1ls9mUm5vr/FtRUZEOHz6ssLAwSZePZPfs2aOffvrJuU3pJFx20gUA4Hrxagw7ODhYY8aM0csvvyy73a6wsDAlJCRIkn7/+99Lkvr06aM2bdro6aef1ty5c5Wfn69p06Zp7NixstlsJu46AADu8WoMW5ISEhLk5+enp59+WufPn9c999yj9PR01a9fX5JUq1Ytpaam6rnnnlNUVJTq1q2rYcOGKT4+vpq7CgCAOQw/59mhQwctWLDAk2MyhSee6wEA3Hgs8ZwnAAA3OyZPAAAMqvExbHqeAIxieURcC0eeAAAYxOQJAIBBlux5Pv/8884FFa587AUAgOvNtMmzbM8zOztbKSkpys3NVUxMjMt2a9eu1dNPP63hw4crJydHX3zxhYYMGXLV9z377LN6/PHHzRoeAACmMXTDUGU9z5CQEKWlpblsv3DhQkVGRurIkSPOnmdcXJwSEhJcep5XLrv35ptvSpJOnDihr7/+2q0dAwDAUyzX83SXw+HQmTNnXF4AAHiC5Xqe7iKGDQDwlhrT8ySGDQDwFsMLw19L2Z5nenq64Z6nu2w2G9UVAIBXWK7nCQCA1Vmu5ylJ3377rc6dO6f8/HydP39eu3fvlnT5iDUgIMCM/QYAwG2W63lK0siRI7Vt2zbn+44dO0qSDh06pPDwcKNDBgDAVPQ8AQA1Gj1PAAAsgMkTAACD6HkCwBXoeeJaOPIEAMAgJk8AAAyyXM8zJydHgwcPlt1uV2BgoFq3bq033njDrGECAFBtpi3PV7bn2b59e50+fVpxcXGKiYlRVlaWc7u1a9cqNjZWs2bNUs+ePXXx4kWXqsquXbvUuHFjrVixQna7XTt27NCoUaNUq1YtjRs3zqzhAgDgNsPPeVbU8yy7WHypzMxMRUZGKi8vz9nzDA8P18yZM116ntcyduxYffPNN0pPT6/y/yl9rsf+wmpuGAJgCDcM1SyWeM7zevQ8CwoK1KBBg0q3oecJAPAWy/c8d+zYoVWrVmnUqFGVjoueJwDAWyzd89y7d6/69++vl19+WX369Kl0XPQ8AQDeYtme5/79+9WrVy+NGjVK06ZNu+bv0vMEAHiLJXue+/btU48ePTRs2DC9+uqrRocIAIBHWa7nuXfvXvXs2VPR0dGaOHGi8ztq1aqlRo0ambXfAAC4zXI9z48++kgnTpzQihUrtGLFCud3hIWF6fDhw27uJgAA5qHnCQCo0SzxnCcAADc7Jk8AAAyq8T1PAADMZvpznlZDDBuAFbBebs3CaVsAAAyyXM/z559/1kMPPaTbbrtNNptNdrtd48aNY6F3AIBlmDZ5lu15ZmdnKyUlRbm5uYqJiXHZbu3atXr66ac1fPhw5eTk6IsvvtCQIUP+MyBfX/Xv318ff/yxDhw4oGXLlmnLli0aM2aMWUMFAKBaboie55tvvqmEhARDi73T8wRgJVy4sfKRAAAgAElEQVTzvH4s8Zynt3ueP/74o1JSUvTAAw9UOi56ngAAb7Fsz3Pw4MGqU6eOmjVrpuDg4Aon6FL0PAEA3mLZnuf8+fOVnZ2tdevW6bvvvtPEiRMrHRc9TwCAt1i25xkaGqrQ0FC1atVKDRo0ULdu3TR9+nTnd1yJnicAwFss2fO8UukRq8PhMDpcAABMZ7me56effqrjx4/r7rvvVr169bRv3z69+OKL6tq1q8LDw83bcwAA3GS5nmdgYKDeffddTZgwQQ6HQ3a7XY8++qimTJlSzV0FAMAcNb7nyXOeAKyA5zyvH08851njF4bfOzOaGDYAwFQsDA8AgEH0PAEAMKjGn7al5wnAKK5P4lo4bQsAgEGW63mW9fPPP+v222+Xj4+PfvnlF7OGCgBAtViu51nWiBEj1K5dO7OGCACAKQxd86ys5xkSEqK0tDSX7RcuXKjIyEgdOXLE2fOMi4tTQkKCS8+z7Fq3pRYtWqRffvlFM2bM0IYNG9zZNwAAPMKSPc/9+/crPj5ey5cvl69v1YZIzxMA4C2W63k6HA4NHjxYCQkJuuOOO6o8LnqeAABvsVzPc+rUqWrdurWeeuopQ+Oi5wkA8BbTH1Up2/NMS0sz3PNMT0/XmjVr5OfnJz8/P/Xq1UuS1LBhQ7388ssV/q7NZlNwcLDLCwAATzC8SEJVe54ZGRmV9jzvu+8+SVf3PNeuXavz5887/09mZqaeffZZbd++Xb/5zW8M7yAAAGazXM/zygny5MmTkqTWrVs7bzwCAOB6slzPEwAAq6PnCQBXYG3bmoWepxvoeQIAzMbC8AAAGETPEwAAg2r8aVt6ngA8geuiNzdO2wIAYBCTJwAABlkyhu3j43PV68MPPzRrqAAAVItp1zzLxrDbt2+v06dPKy4uTjExMcrKynJut3btWsXGxmrWrFnq2bOnLl68eFWSTJKWLl2qhx56yPme1YUAAFZh2Rj2LbfcotDQUHf2CQAAj7JkDFuSxo4dq4YNGyoyMlJLlizRtRZCIoYNAPAWy8WwJSk+Pl6rV69WWlqaBg4cqD/84Q9KSkqqdFzEsAEA3mK5GLYkTZ8+XV27dlXHjh01efJkvfTSS876SkWIYQMAvMVyMezy3HPPPTp69KgcDkeF2xDDBgB4i+HJs6ox7C1btlQawy51ZQy7PLt371b9+vVls9mMDhcAANNZLoa9fv16HT9+XF26dFHt2rWVlpamWbNmadKkSSbuNgAA7rNcDNvf319vvfWWJkyYoEuXLunOO+/UvHnzFBsbW81dBQDAHDU+hm1m/BQAcOPxxHzA2rYAABjE5AkAgEE1PoZNzxOoGE1KwD0ceQIAYBCTJwAABlmy5ylJy5YtU7t27VS7dm01btxYY8eONWuoAABUiyV7nvPmzVNiYqISEhJ0zz336N///rcOHz5s1lABAKgWw895VtTzLLtYfKnMzExFRkYqLy/P2fMMDw/XzJkzXXqeZZ0+fVrNmjXT+vXr1atXLzd36z/P9dhfWM0NQ0AFuGEINwNLPOfp6Z5nWlqaSkpK9K9//UutW7fW7bffrscee+yalRR6ngAAb7Fcz/P7779XSUmJZs2apQULFuijjz7SqVOn9OCDD+rChQsVjoueJwDAWyzX8ywpKVFRUZHefPNNRUdHq0uXLlq5cqUOHjyojIyMCsdFzxMA4C2m3TBUqmzPMz093XDPs7xtGjVqpIYNG1ba/LTZbCTLAABeYbmeZ9euXSXJZZtTp07p5MmTlTY/AQDwFsOTZ2nPMzc3VytXrlRSUpLi4uKcPc+srCwlJyc7e575+fnOa5Vle56bN29Wbm6unnvuOUn/6Xm2aNFC/fv3V1xcnHbs2KG9e/dq2LBhatWqlXr06GHirgMA4B7L9Twlafny5ZowYYL69u0rX19fPfDAA9q4caP8/f2rsasAAJiDnicAoEazxHOeAADc7Jg8AQAwqMb3PAEAMJvpz3laDTFswDjWvAUqx2lbAAAMslzPc9myZfLx8Sn39dNPP5k1XAAA3Ga5nufjjz+uhx56yOW7n3nmGRUWFqpx48ZmDRcAALcZmjwr63mGhIQoLS3NZfuFCxcqMjJSR44ccfY84+LilJCQ4NLzLLuObWBgoAIDA53vT5w4ofT0dC1evNitHQQAwGyW63leafny5apTp44GDRpU6bjoeQIAvMVyPc8rLV68WEOGDHE5Gi0PPU8AgLdYrudZ1s6dO/XNN9+4nOKtCD1PAIC3WK7nWdZ7772nDh06qHPnztf8XXqeAABvsVzPs9S5c+e0evXqKh11AgDgTYaPPEt7nqNHj1Z2draSkpKUmJjo7HlmZ2crNTXV2fOUpAYNGiggIMCl52m32xUWFqaEhARJ/+l5llq1apUuXryop556yoTdBADAPJbseUqXbxR69NFHnXfqAgBgFTW+52l/YTVr2wIGsbYtahJP9Dxr/MLwe2dGE8MGAJiKheEBADCInicAAAbV+NO29DwBeBvXjGs+TtsCAGCQ5XqekpSZmalevXrplltuUf369RUdHa2cnByzhgoAQLWYNnmW7XlmZ2crJSVFubm5iomJcdlu7dq1evrppzV8+HDl5OToiy++0JAhQ5yfnzt3Tg899JDuuOMOffXVV/q///s/BQUFKTo6WkVFRWYNFwAAtxl+zrOinmfZxeJLZWZmKjIyUnl5ec6eZ3h4uGbOnFnhsntZWVm6++67deTIEWcZZc+ePWrXrp0OHjyoO++8s0pj5TlPANcL1zytxRPPeVqu59myZUvdeuutWrx4sS5cuKDz589r8eLFat26tcLDwyscFz1PAIC3WK7nGRQUpK1bt2rFihUKDAxUvXr1tHHjRm3YsEF+fhXfHEzPEwDgLZbreZ4/f14jRoxQ165d9eWXX+qLL77Qf/3Xf6lv3746f/58heOi5wkA8BbL9Tw/+OADHT58WDt37pSvr6/zb/Xr19e6dev0xBNPlPu79DwBAN5iuZ7nr7/+Kl9fX5ej29L3pUeuAABcT4Ynz9KeZ25urlauXKmkpCTFxcU5e55ZWVlKTk529jzz8/N14cIFSXLpeW7evFm5ubl67rnnJP2n5/nggw/q9OnTGjt2rL755hvt27dPw4cPl5+fn3r06GHirgMA4B7L9TxbtWql9evXa+bMmYqKinLembtx40bnaV8AAK4nep4AYDKe87QWep5uoOcJADAbC8MDAGAQPU8AAAyq8adt6XkCnsc1PtxsOG0LAIBBTJ4AABhkyRj2Z599pnvvvVdBQUEKDQ3V5MmTdfHiRbOGCgBAtVguhp2Tk6NHHnlEDz30kP75z39q1apV+vjjjzVlyhSzhgoAQLVYLob9pz/9SWlpacrMzHT+bf369Xrsscf0008/KSgoqEpjZZEEwHu4YQhWdlPEsB0Oh2rXru3yPYGBgSosLNSuXbsqHBcxbACAt1guhh0dHa0dO3Zo5cqVKi4u1r/+9S/Fx8dLko4dO1bhuIhhAwC8xXIx7D59+ighIUFjxoyRzWZTixYt9Mgjj1werG/FwyWGDQDwFtMfVSkbw05LSzMcw5akiRMn6pdfftGRI0d08uRJ9e/fX5IUERFR4e/abDYFBwe7vAAA8ATLxbBL+fj46LbbblNgYKBWrlwpu92uTp06GR0uAACmM7w8X2kMe/To0crOzlZSUpISExOdMezs7GylpqY6Y9iS1KBBAwUEBLjEsO12u8LCwpSQkCDpPzFs6XLz86GHHpKvr69SUlL02muvafXq1apVq5ZJuw0AgPssF8OWpA0bNujVV1+Vw+FQ+/bttW7dOj388MPV2E0AAMxT42PYZj7XAwC48VjiOU8AAG52TJ4AABhU42PY9DwB1AQsgWgtHHkCAGAQkycAAAaZOnlWpem5detW+fj4lPsqW1L5+uuv1a1bN9WuXVt2u11z5841c6gAALjN1MmzKk3Pe++9V8eOHXN5jRw5Us2bN9dvf/tbSZdvK+7Tp4/CwsK0a9cuJSQk6JVXXtF///d/mzlcAADcYniRhMqaniEhIUpLS3PZfuHChYqMjNSRI0d0xx13KCAgQKGhoc7Pi4qKtG7dOo0fP9654HxycrIuXLigJUuWKCAgQHfddZd2796tefPmadSoUdXZXwAAqs2tI8/qND2v9PHHH+vnn3/W8OHDnX/buXOn7r//fgUEBDj/Fh0drdzcXJ0+fbrc76HnCQDwFrcmz+o0Pa+0ePFiRUdH6/bbb3f+LT8/X02aNHHZrvR96Xq5V6LnCQDwFrcmz+o0Pcs6evSoNm3apBEjRrgzDBf0PAEA3mL4mmdVlG16pqenV3jUuXTpUt16660uNxRJUmhoqI4fP+7yt9L3Za+XlmWz2WSz2UwYPQAAlXPryLM6Tc9Sly5d0tKlSzV06FD5+/u7fBYVFaXPP/9cRUVFzr+lpaWpZcuWLvUVAACuB7cmz9KmZ25urlauXKmkpCTFxcU5m55ZWVlKTk52Nj3z8/N14cIFl+9IT0/XoUOHNHLkyKu+f8iQIQoICNCIESO0b98+rVq1Sm+88YYmTpzo3l4CAGAit07bVrfpKV2+Uejee+9Vq1atrvr+kJAQbd68WWPHjlXnzp3VsGFDzZgxg8dUAACWYKjnKd04TU96ngAAiZ4nAACWwOQJAIBBhq953ohNTwAAzOSR5zythBg2gOuNkHXNw2lbAAAMMm3yNKvlWVhYqGeeeUZt27aVn5+fBgwYYNYQAQAwhWmTp1ktz+LiYgUGBur5559X7969zRoeAACmMXTN0xstz7p16zoXkv/iiy/0yy+/VGsHAQAwm+EjT0+3PN1FzxMA4C2GJ09PtzzdRc8TAOAthidPK7Y8JXqeAADvMf05z+q2PN1FzxMA4C2GJ8+qtjwzMjLcankCAGB1hifP0pbn6NGjlZ2draSkJCUmJjpbntnZ2UpNTXW2PCWpQYMGCggIcH5HZS1PSdq/f78uXLigU6dO6ezZs9q9e7ekqzNnAABcD4YnT0+3PCXpkUceUV5envN9x44dJV0+YgUA4Hoz1PO8UVqeEj1PAMBl9DwBALAAJk8AAAwydM2TlicAAPQ8AVgcLUxYEadtAQAwyNTJ06ym59atW9W/f381bdpUdevWVYcOHZScnGzmUAEAcJupk6dZTc8dO3aoXbt2Wrt2rb7++msNHz5cQ4cOVWpqqpnDBQDALYae85Qqb3qWXTC+VGZmpiIjI5WXl6c77rjjqs+LiorUrFkzjR8/XtOnT6/wd/v27asmTZpoyZIlVRpn6XM99hdWc80TuIFxzRPVZZnnPK9H07OgoEANGjSo8HN6ngAAb3Fr8vR203P16tXKzMysdIKl5wkA8Ba3Jk9vNj0zMjI0fPhwvfvuu7rrrrsq3I6eJwDAWzzynKdZTc9t27bpd7/7nebPn6+hQ4dW+pv0PAEA3uLW5OmNpufWrVvVr18/zZkzR6NGjXJnmAAAeIRbk6enm54ZGRnq16+f4uLiNHDgQOd3BAQEVHrTEAAA3uDW5Onppuf777+vX3/9VbNnz9bs2bOdf3/ggQdYXxcAcN259ZznjdD05DlPoGbgOU9Ulyee86zxC8PvnRlNDBsAYCoWhgcAwCDDR55ccwQA3Oxq/Glbep6Ae7jWCFSM07YAABjE5AkAgEGWjGHn5uaqR48eatKkiWrXrq2IiAhNmzZNRUVFZg4XAAC3mHrNs2wMu3379jp9+rTi4uIUExOjrKwsSf+JYZc1ffp0ffbZZ84Ytr+/v4YOHapOnTrplltuUU5OjmJjY1VSUqJZs2aZOWQAAAwzPHlWFsMOCQlRWlqay/YLFy5UZGSkjhw5ojvuuEMBAQEKDQ11fl5UVKR169Zp/PjxzlJLRESEIiIinNuEhYVp69at2r59e4Xjcjgccjgczvf0PAEAnnJDxLC//fZbbdy4UQ888ECF29DzBAB4i1vL8/3000/at2+f80hxypQp+vjjj7V//36XbQsLC9W1a1e1atVKycnJ5X7fI488Ikn69NNPr/rs3nvvVXZ2thwOh0aNGqVFixbJ17f8+b68I0+73c7yfICbeFQFNYUnluezdAx71apVys7O1gcffKBPPvlEr7/+eoVjstlsCg4OdnkBAOAJlo5hl556bdOmjYqLizVq1Cj98Y9/VK1atTwxbAAAqsStI8+qxrC3bNnidgz7SiUlJSoqKlJJSYk7QwYAwDSWjGEnJyfL399fbdu2lc1mU1ZWlqZOnarHH3+8ShMtAACeZMkYtp+fn+bMmaMDBw7o0qVLCgsL07hx4zRhwgR3hgsAgKlqfAzbzLurAAA3HsvcbQsAwM2MyRMAAINqfAybnicAeMbNvJAGR54AABjE5AkAgEGW7HmW9e233yooKKjCheUBAPA2UyfPsj3P7OxspaSkKDc312X5vdKeZ9nXyJEj1bx5c2fPs1RRUZEGDx6sbt26mTlMAACqxZI9z1LTpk1Tq1at1KtXL+3YscOd/QMAwHSW7Xmmp6drzZo1euutt6o0JofDoTNnzri8AADwBLeW57Pb7Zo/f758fHzUsmVL7dmzR/Pnz1dsbKzLdoWFhZo8ebIGDx5c4aoOixcvVnR0tG6//Xbn337++Wc988wzWrFiRZVXg5g9e7Zmzpzpzu4AAGCIJXuesbGxGjJkiO6///4qj2nq1KkqKChwvn744QeDewUAQNV45FGVsj3PtLQ0wz3P9PR0vf766/Lz85Ofn59GjBihgoIC+fn5acmSJeV+FzFsAIC3uHXatqo9z4yMDLd6njt37nQ5il23bp3mzJmjHTt2qFmzZu4MGQAA01iy59m6dWuX91lZWfL19XXe5QsAwPVkyZ4nAABWRs8TAFCj0fMEAMACmDwBADCInieAKruZ+41AWRx5AgBgEJMnAAAGWbLnefjw4XI///LLL80cLgAAbnHrOc+KlO15tm/fXqdPn1ZcXJxiYmKUlZUl6T89z7KmT5+uzz777Kqe55YtW3TXXXc531e0WhEAAN5k6Z7nrbfe6rItAABWYNmepyTFxMSocePGuu+++5wrF1WEnicAwFvcmjxLe54tW7bUk08+qfHjx2v+/PlXbeduz7NevXpKTEzUmjVr9Mknn+i+++7TgAEDKp1AZ8+erZCQEOfLbre7s2sAAFyTW8vzRUREuKTB1q1bp0GDBqmwsFC1atWSdPl07MCBA3X06FFt3bq13Mnz6NGjCgsL0+rVqzVw4MBKf3fo0KE6dOiQtm/fXu7nDodDDofD+f7MmTOy2+2yv7Ca5zwBk/CcJ25Enliez9QbhkqV7Xmmp6cb7nmW55577rnqempZNptNNpvN7TEDAFBVlux5lmf37t1q2rSpO8MFAMBUlux5vv/++woICFDHjh0lSSkpKVqyZEmFNyUBAOBNlu15/uUvf1FeXp78/PzUqlUrrVq1SoMGDXJnuAAAmIqeJwCgRqPnCQCABTB5AgBgUI3veQIAYDaPPOdpJcSwAdxMWMjCOzhtCwCAQaZNnma1PKXLCyi8/vrratGihWw2m5o1a6ZXX33VrKECAFAtpp22NbPlGRcXp82bN+v1119X27ZtderUKZ06dcqsoQIAUC2GJk9vtDy/+eYbLVq0SHv37lXLli0lSc2bN6/WTgIAYCbDp2093fJcv369IiIilJqaqubNmys8PFwjR4685pEnPU8AgLcYnjw93fL8/vvvlZeXpzVr1mj58uVatmyZdu3adc2l+eh5AgC8xfDk2aVLF+cpVkmKiorSwYMHVVxc7PxbaVnl0qVLWrRoUbnfc/ToUW3atEkjRoxw+XtJSYkcDoeWL1+ubt26qXv37lq8eLEyMjKUm5tb4bimTp2qgoIC5+uHH34wumsAAFSJ6c95Vrfl2bRpU/n5+alFixbOv7Vu3VrS5ZpL6XXQK9HzBAB4i+Ejz6q2PLds2eJWy7Nr1666ePGivvvuO+ffDhw4IEkKCwszOlwAAExnePIsbXnm5uZq5cqVSkpKUlxcnLPlmZWVpeTkZGfLMz8/XxcuXHD5jspanr1791anTp307LPP6p///Kd27dql0aNH68EHH3Q5GgUA4HoxfNrW0y1PX19frV+/XuPHj9f999+vunXr6uGHH1ZiYqLRoQIA4BGGJ09/f38tWLDgqhuBwsPDVdU06AcffFDp57fddpvWrl1rdGgAAHhFjV8Yfu/MaGLYAABTsTA8AAAGGTrypOUJAMBNcNqWnieAytC/hDs4bQsAgEFMngAAGGTq5GlWEPuVV14p9/O6deuaOVwAANxi6uRZNoidnZ2tlJQU5ebmuqxfWxrELvsaOXKkmjdv7gxiT5o06apt2rRpo9///vdmDhcAALcYvmHIG0HsevXqqV69es5tcnJytH//fr3zzjsVjsvhcMjhcDjf0/MEAHiKW0eeng5iX+m9995TixYt1K1btwq3oecJAPAWn0tVXVPv/+vevbt++ukn7du3z3mkOGXKFH388cfav3+/y7aFhYXq2rWrWrVqpeTk5HK/75FHHpEkffrpp+V+XlhYqNtuu01TpkzRSy+9VOG4yjvytNvtsr+wmkdVAFSIR1VqvjNnzigkJEQFBQWmrTjn1pGnp4PYZf3P//yPzp49q2HDhlU6JpvNpuDgYJcXAACe4JFFEqobxC7rvffeU79+/dSkSRNPDBUAAMPcmjyrGsTOyMhwK4hd6tChQ8rIyHCmzgAAsAK3Ttt6OohdasmSJWratKkefvhhd4YJAIBHuHXk6ekgtiSVlJRo2bJleuaZZ1SrVi13hgkAgEe4dbdthw4dtGDBAk+NyRSeuLsKAHDjsczdtgAA3MyYPAEAMMjwNc8bLYhNzxMwDwsKAJdx5AkAgEFMngAAGGTa5GlWy1OSNm3apC5duigoKEiNGjXSwIEDdfjwYbOGCgBAtZg2eZrV8jx06JD69++vnj17avfu3dq0aZNOnjypRx991KyhAgBQLYZuGPJGy3PXrl0qLi7WX//6V/n6Xp7bJ02apP79+6uoqKjCpfwAAPAWw0eenm55du7cWb6+vlq6dKmKi4tVUFCgv//97+rdu3elE6fD4dCZM2dcXgAAeILhydNut2v+/Plq2bKlnnzySY0fP17z58+/arvCwkJNnjxZgwcPrnBFh8WLFys6Olq3336782/NmzfX5s2b9ac//Uk2m0233HKLjh49qtWrV1c6LmLYAABvMTx5errlmZ+fr9jYWA0bNkyZmZnatm2bAgICNGjQIFW2kuDUqVNVUFDgfP3www9Gdw0AgCoxvedZ3ZbnW2+9pZCQEM2dO9f5txUrVshut+urr75Sly5dyv0+m80mm81m3o4AAFABw5Onp1uev/76q/NGoVKlVZWSkhKjwwUAwHSGT9t6uuXZt29fZWZmKj4+XgcPHlR2draGDx+usLAwdezY0f09BQDAJIaPPD3d8uzZs6c++OADzZ07V3PnzlWdOnUUFRWljRs3KjAw0OhwAQAwnaGe543S8pToeQIALqPnCQCABTB5AgBgkKFrnjday1Oi54kbHw1NwHo48gQAwCAmTwAADLJkz3P16tXq0KGD6tSpo7CwMCUkJJg1TAAAqs1yPc8NGzboySef1JgxY7R37169/fbbmj9/vhYuXGjWUAEAqBbDz3lW1PMsu1h8qczMTEVGRiovL0933HHHVZ8XFRWpWbNmGj9+vKZPny5JGjJkiIqKirRmzRrndklJSZo7d66OHDlS7u+Up/S5HvsLq7lhCDc0bhgCqscSz3l6uufpcDhUu3Ztl+0CAwN19OhR5eXlVTguep4AAG+xXM8zOjpaKSkp+uyzz1RSUqIDBw4oMTFRknTs2LEKx0XPEwDgLZbrecbGxmrcuHHq16+fAgIC1KVLFz3xxBOXB+tb8XDpeQIAvMX0R1XK9jzT0tIM9zx9fHw0Z84cnTt3Tnl5ecrPz1dkZKQkKSIiosLftdlsCg4OdnkBAOAJlut5lqpVq5aaNWsmSVq5cqWioqLUqFEjo8MFAMB0hifP0p7n6NGjlZ2draSkJCUmJjp7ntnZ2UpNTXX2PCWpQYMGCggIcH5HZT3PkydP6qOPPlL37t1VWFiopUuXas2aNdq2bVs1dhMAAPNYrucpXb6jd9KkSbp06ZKioqK0detW56lbAACuN3qeAIAazRLPeQIAcLNj8gQAwKAa3/MEAMBshm8YutEQwwZQk7DWsTVw2hYAAIO82vOUpAMHDqh///5q2LChgoODdd999ykjI8NlmyNHjqhv376qU6eOGjdurBdffFEXL140a6gAAFSLV3uektSvXz9dvHhR6enp2rVrl9q3b69+/fo5F1QoLi5W3759deHCBe3YsUPvv/++li1bphkzZpg1VAAAqsWrPc+TJ0+qUaNG+vzzz9WtWzdJ0tmzZxUcHKy0tDT17t1bGzZsUL9+/fTjjz+qSZMmkqR33nlHkydP1okTJ1xWKqoMPU8ANRHXPI2zxHOe1el53nrrrWrZsqWWL1+uf//737p48aL+9re/qXHjxurcubMkaefOnWrbtq1z4pQuZ8rOnDmjffv2VTguep4AAG8xfLdtac/Tx8dHLVu21J49ezR//nzFxsa6bFdez9PHx0dbtmzRgAEDFBQUJF9fXzVu3FgbN25U/fr1JUn5+fkuE6ck5/vSU7vlmT17tmbOnGl0dwAAMMyrPc9Lly5p7Nixaty4sbZv365//OMfGjBggH73u99VGrquCnqeAABvMf05z7I9z/T0dJfzy+np6UpNTdXp06edf3/77beVlpam999/X1OmTFFoaKj+8Y9/uHzn8ePHJUmhoaEV/q7NZpPNZjN7dwAAuIrhI8+q9jy3bNlyVc/z119/vfyjvq4/6+vrq5KSEkmXj2T37Nmjn376yfl5aVS7TZs2RocLAIDpDE+epT3P3NxcrVy5UklJSYqLi3P2PLOyspScnOzseebn5+vChf60oBgAACAASURBVAuSLk+M9evX17Bhw5STk6MDBw7oxRdf1KFDh9S37+U7yPr06aM2bdro6aefVk5OjjZt2qRp06Zp7NixHFkCACzBqz3Phg0bauPGjfrzn/+snj17qqioSHfddZfWrVun9u3bS5Jq1aql1NRUPffcc4qKilLdunU1bNgwxcfHm7C7AABUX43vefKcJ4CahOc8jfPEc541fmH4vTOjiWEDAEzFwvAAABhEzxMAAINq/Glbep4AqovrjLgSp20BADDIkj3P559/Xp07d5bNZrvqsRcAAK43y/U8Sz377LN6/PHHzRoeAACmMXTNs7KeZ0hIiNLS0ly2X7hwoSIjI3XkyBFnz/PgwYNavHix2rVrJ0l67bXX9Pbbb2vv3r3OtWvffPNNSdKJEyf09ddfV3snAQAwk+V6nu6i5wkA8BbL9TzdRc8TAOAt9DwBADDIcj1Pd9HzBAB4i+HJs6o9z4yMDLd6ngAAWJ3hybO05zl69GhlZ2crKSlJiYmJzp5ndna2UlNTnT1PSWrQoIECAgJcep4zZsxQYGCg3n33XZeepyR9++23OnfunPLz83X+/Hnt3r1bktSmTRsFBASYtOsAALjHcj1PSRo5cqS2bdvmfN+xY0dJ0qFDhxQeHu7OfgIAYBrDk6e/v78WLFjgciOQJIWHh6sqadDf/va32rRpU6XbsAA9AMDKavzC8PQ8AQBmY2F4AAAMoucJAIBBNf60LT1PwDx0LYHLOG0LAIBBTJ4AABhkuRh2Tk6OBg8eLLvdrsDAQLVu3VpvvPGGWcMEAKDaLBfD3rVrlxo3bqwVK1Zo3759+vOf/6ypU6dq4cKFZg0VAIBq8blUlZUN/r/KYthlSyulMjMzFRkZqby8PGcMu1GjRvr888/VrVs3SdLZs2cVHBystLQ09e7du9zfHTt2rL755hulp6dXecfOnDmjkJAQ2V9YzQ1DgEm4YQg3otL5oKCgwLTn/m+IGHZBQYEaNGhQ6biIYQMAvMXyMewdO3Zo1apV+uSTTyodFzFsAIC3WDqGvXfvXvXv318vv/yy+vTpU+m4iGEDALzFsjHs/fv3q1evXho1apSmTZt2zd8lhg0A8BZLxrD37dunnj17atiwYXr11VeNDhEAAI8yfNq2NIadm5urlStXKikpSXFxcc4YdlZWlpKTk50x7Pz8fF24cEGSXGLYOTk5OnDggF588UWXGPbevXvVo0cP9enTRxMnTnR+x4kTJ8zdcwAA3GS5GPZHH32kEydOaMWKFVqxYoXzO8LCwnT48OFq7CoAAOYw/Jxnhw4dtGDBAk+OyRSeeK4HAHDjscRzngAA3OyYPAEAMKjGx7DpeQLXF0v6oSbiyBMAAIOYPAEAMMhyPc+ff/5ZDz30kG677TbZbDbZ7XaNGzeOhd4BAJZhuZ6nr6+v+vfvr48//lgHDhzQsmXLtGXLFo0ZM8asoQIAUC03RM/zzTffVEJCgqHF3ul5AtbADUO43izxnKe3e54//vijUlJS9MADD1Q6LnqeAABvMTx5lvY8W7ZsqSeffFLjx4/X/Pnzr9qusp7nP//5TwUFBal27dqaN29euT3PwYMHq06dOmrWrJmCg4MrnKBLzZ49WyEhIc6X3W43umsAAFSJZXue8+fPV3Z2ttatW6fvvvtOEydOrHRc9DwBAN5i2Z5naGioQkND1apVKzVo0EDdunXT9OnT1bRp03J/l54nAMBbLNnzvFLpZw6Hw+hwAQAwneHJs7TnOXr0aGVnZyspKUmJiYnOnmd2drZSU1OdPU9JatCggQICAlx6njNmzFBgYKDeffddl57np59+quPHj+vuu+9WvXr1tG/fPr344ovq2rWrwsPDTd15AADcYbmeZ+mEOmHCBDkcDtntdj366KMup3QBALie6HkCAGo0SzznCQDAzY7JEwAAg2p8zxMAALOZ/pyn1RDDBqqP9WkBV5y2BQDAIMv1PMv6+eefdfvtt8vHx0e//PKLWUMFAKBaLNfzLGvEiBFq166dWUMEAMAUhq55VtbzDAkJUVpamsv2CxcuVGRkpI4cOeLseR48eFCLFy92Toqvvfaa3n77be3du1ehoaHO/7to0SL98ssvmjFjhjZs2FDd/QQAwDSW7Hnu379f8fHxWr58+VXr4FaEnicAwFsM321b2vP08fFRy5YttWfPHs2fP1+xsbEu21XW8xwwYICCgoLk6+urxo0bu/Q8HQ6HBg8erISEBN1xxx36/vvvqzSu2bNna+bMmUZ3BwAAwyzX85w6dapat26tp556ytC46HkCALzF9EdVyvY809LSyu15fvjhh+ratas6deqkt99+W4GBgXr//fed26xZs0Z+fn7y8/NTr169JEkNGzbUyy+/XOHv2mw2BQcHu7wAAPAEy/U8165dq/Pnzzs/y8zM1LPPPqvt27frN7/5jdHhAgBgOsv1PK+cIE+ePClJat26tfPGIwAArifL9TwBALA6w5Onv7+/FixY4HIjkCSFh4erKmnQ3/72t9q0aVOVf6979+5V+l4AALylxi8Mv3dmNDcPAQBMxcLwAAAYRM8TAACDavxpW3qeuBnR3wQ8i9O2AAAYZMmep4+Pz1WvDz/80KyhAgBQLZbteS5dulTHjh1zvgYMGGDWUAEAqBbL9jxvueUWl/cAAFiFJXuekjR27Fg1bNhQkZGRWrJkyTUXSqDnCQDwFsv1PCUpPj5ePXv2VJ06dbR582b94Q9/0Llz5/T8889XOC56ngAAb7Fcz1OSpk+frq5du6pjx46aPHmyXnrpJSUkJFQ6LnqeAABvsVzPszz33HOPjh49KofDUeE29DwBAN5iuZ5neXbv3q369evLZrMZHS4AAKazXM9z/fr1On78uLp06aLatWsrLS1Ns2bN0qRJk8zdcwAA3GS5nqe/v7/eeustTZgwQZcuXdKdd96pefPmXXVDEgAA14vPJQOxzO7du6tDhw5asGCBJ8dkijNnzigkJET2F1azti1uOqxtC/xH6XxQUFBg2v0wNX5heHqeAACzsTA8AAAG0fMEAMCgGn/alp4nACvj+vSNidO2AAAYxOQJAIBBloxhS9KyZcvUrl071a5dW40bN9bYsWPNGioAANViyRj2vHnz9Oc//1lTpkzRvn37tGXLFkVHR5s1VAAAqsXwIgkVxbDLllZKZWZmKjIyUnl5ec4YdqNGjfT555+rW7dukqSzZ88qODhYaWlp6t27t06fPq3/x969hlVVJv7//4AbNyge80SGqKmIpaM5lYc85QHLvtpVTilWpoQ2XzXTySksM53LzCMWOjXVNnRUBi2/mZQmjnYwpRTEAxlmJU4qJuJhREMU/w/87/1zKyhrs/Z2ge/Xde0HyGLvez26XWuv+343bNhQq1evVs+ePT0+MTZJAFAe8MCQ93ljkwTLxbBTUlJUVFSkgwcPKiIiQrfddpsee+yx6ybGiGEDAHzF8OTpjGGHh4dryJAhGjNmjOLi4q467lox7O3bt6tatWoKDAzU3Llz3WLYP//8s4qKivT6669r3rx5+vDDD5WXl6fevXvr3LlzJY5r+vTpqlGjhusVGhpq9NQAACgVy8Wwi4qKVFhYqLfeekuRkZHq0KGDEhMTXZmzkhDDBgD4iumbJFwew96wYUOxMezjx4+7/v3vf/+7UlJStGjRIr300ksKCQmRJLVq1cr1d3Xr1lWdOnV04MCBEj/XbrfT+wQA+IThK8/SxrDXr1/vUQy7c+fOkqSsrCzX7/Py8pSbm6uwsDCjwwUAwHSGJ09nDDsrK0uJiYmKj4/X2LFjXTHsbdu2aenSpa4Ydk5Ojuu7ystj2Dt27NDevXs1YcIEtxh2ixYtNGDAAI0dO1abN2/W7t27NXToULVs2VI9evQw9+wBAPCA5WLYkrR48WKNGzdO/fr1k7+/v7p166a1a9cqICCgjKcLAEDZVfgYtpnregAA5Y8l1nkCAHCzY/IEAMCgCh/DpucJlA7bxAGlx5UnAAAGMXkCAGCQ5XqeCQkJ8vPzK/b122+/mTVcAAA8Zrme5+OPP67Dhw+7vSIjI9WtWzfVq1fPrOECAOAxy/U8r3T06FE1bNhQDodDTz75ZKlPjJ4nYAwPDKGissQ6T2/3PK+0ePFiValSRQMHDrzmuOh5AgB8xfD2fM6ep5+fn8LDw7Vr1y7FxcUpJibG7bhr9TwffvhhVatWTf7+/qpXr55bz/NKDodDUVFRCgoKuua4pk+frilTphg9HQAADLNcz/NyW7Zs0Z49exQdHX3dcdHzBAD4iuV6npd7//331bZt2xJv6V6OnicAwFcs1/N0On36tJYvX16qq04AAHzJcj1Pp6SkJJ0/f15PPPGEOWcKAIBJLNnzlC49KPTII4+4ntQFAMAq6HkCACo0S6zzBADgZsfkCQCAQfQ8AZRLbCeIG4krTwAADGLyBADAIMv1PKVLNZaePXuqZs2aqlWrliIjI7Vjxw6zhgoAQJlYrud5+vRp9e3bV40aNdK3336rTZs2qVq1aoqMjFRhYaFZwwUAwGOW63lu27ZNd999tw4cOKDQ0FBJ0q5du9SmTRv9+OOPatasWanGSs8TqNh4YAilZYl1nt7ueYaHh+uWW26Rw+HQuXPndPbsWTkcDkVERKhx48YljoueJwDAVyzX86xWrZq++OILPfzww/rb3/4mSWrevLk+//xz2WwlD5eeJwDAVyzX8zx79qyio6PVuXNnpaam6ptvvtGdd96pfv366ezZsyWOi54nAMBXLNfzXLZsmfbv368tW7a40mXLli1TrVq1tGrVKg0aNKjYz6XnCQDwFcOTZ2l7nhs3bvSo53nmzBn5+/u7Xd06f76y+QkAwI1guZ5n7969dfz4cY0aNUp79uxRZmamhg0bJpvNph49eph79gAAeMByPc+WLVtq9erVmjJlijp27Ch/f3+1a9dOa9euVUhIiAmnDABA2dDzBABUaJZY5wkAwM2OyRMAAIMqfM8TAACzmb7O02qIYeNmxv6vgHdw2xYAAIN83vNMT09X7969VbNmTd1yyy0aMWKETp8+7fr9jh07NHjwYIWGhiooKEgRERF68803zRomAABl5tOe56FDh9SrVy81a9ZM3377rdauXavMzEw9/fTTrmPS0tJUr149LVmyRJmZmXr55ZcVGxur+fPnmzVUAADKxKc9z3fffVeTJk3S4cOHXVv0labV6dxtaMOGDaU+MXqeAN95ApJF1nmWpedZUFCgypUru+1tGxQUJEnatGlTiZ958uRJ1a5d+5rjoucJAPAVw5Ons+cZHh6uIUOGaMyYMYqLi7vquOJ6nvfff79ycnI0a9YsnTt3TsePH9dLL70kSa4k2ZU2b96spKQkjRgx4prjmj59umrUqOF6hYaGGj01AABKxac9zzvuuEOLFi3SnDlzVKVKFTVo0EBNmjRR/fr1ryqtSNLu3bs1YMAATZ48WX369LnmuOh5AgB8xac9T0mKiopSVFSUjhw5oqpVq8rPz09z585V06ZN3Y77/vvv1bNnT40YMUKvvPLKdT+XnicAwFd82vO8XP369SVJCxcuVGBgoHr37u36XWZmpu6//34NHTpU06ZNMzpEAAC8yvDk6ex5jhw5Uunp6YqPj9ecOXNcPc/09HQlJye7ep6SVLt2bVWuXFmSNH/+fHXq1EnBwcFKSUnRhAkT9MYbb7geKtq9e7fuv/9+RUZGavz48a73qFSpkurWrWvWeQMA4DGf9jwl6bvvvtPkyZN1+vRptWzZUv/4xz/05JNPuo798MMPdfToUS1ZskRLlixx/XtYWJj279/vwSkCAGCuCt/zZJ0nbmas8wS8s86zwm8Mv3tKJDFsAICp2BgeAACD6HkCAGBQhb9tS88TqFj4HhdWwG1bAAAMYvIEAMAgy8Wwjx07pr59++rWW2+V3W5XaGioRo8eTSUFAGAZloth+/v7a8CAAfrkk0+0d+9eJSQkaP369Xr22WfNGioAAGVSLmLYb731lmbNmnXNUkpBQYEKCgpcP586dUqhoaFskgBUMDwwBKNuyhj2oUOHtHLlSnXr1u2a46LnCQDwFcvGsAcPHqwqVaqoYcOGql69eokTtBM9TwCAr1g2hh0XF6f09HStWrVKP/30k8aPH3/NcdntdlWvXt3tBQCAN1g2ht2gQQM1aNBALVu2VO3atdWlSxdNmjRJISEhZg8ZAABDLBnDvlJRUZEkuT0QBADAjWK5GPZnn32mI0eO6O6771ZwcLAyMzM1YcIEde7cWY0bNzbvzAEA8JDlYthBQUF67733NG7cOBUUFCg0NFSPPPKI68EiAAButAofwzZzXQ8AoPyxxDpPAABudkyeAAAYVOFj2PQ8AXdsbweUHVeeAAAYxOQJAIBBlut5Xu7YsWO67bbb5OfnpxMnTpg1VAAAysRyPc/LRUdHq02bNmYNEQAAUxh6YOhaPc8aNWooJSXF7fj58+frnnvu0YEDB9SoUSMlJycrICBACxYscG0E/84776hNmzbat2+fW8/z7bff1okTJ/Tqq69qzZo1ZT1PAABMY8me5/fff6+pU6dq8eLFV9VWSlJQUKBTp065vQAA8AbL9TwLCgo0ePBgzZo1S40aNSr1uIhhAwB8xXI9z9jYWEVEROiJJ54wNC5i2AAAXzF9qcrlPc+UlJRie545OTk6ePCgjh07ptdee01Hjx519Tw3bNigFStWyGazyWazqWfPnpKkOnXqaPLkySV+LjFsAICvWK7n+dFHH+ns2bOu47Zu3arhw4fr66+/1u233250uAAAmM5yPc8rJ8jc3FxJUkREhOsYAABuJMv1PAEAsDp6ngCACo2eJwAAFsDkCQCAQfQ8AZQ7NElxo3HlCQCAQUyeAAAYZMmep5+f31Wvf/3rX2YNFQCAMrFsz/ODDz7Q4cOHXa+HH37YrKECAFAmlu151qxZUw0aNCjr+QEAYDpL9jwladSoUapTp47uueceLVy4UNfby4GeJwDAVyzX85SkqVOnavny5UpJSdGjjz6q//3f/1V8fPw1x0XPEwDgK5breUrSpEmT1LlzZ7Vr104vvvii/vrXv2rWrFnXHBc9TwCAr1iu51mce++9V7/++qsKCgpKPIaeJwDAVyzX8yxORkaGatWqJbvdbnS4AACYznI9z9WrV+vIkSPq0KGDAgMDlZKSotdff10vvPCCiacNAIDnLNfzdC5lGTdunC5evKhmzZpp7ty5iomJKcNpAgBgHnqeAIAKjZ4nAAAWwOQJAIBBFb7nCQCA2Qw/MFTeEMMGcD3EtWEUt20BADDIcj3PhISEYnuefn5++u2338waLgAAHrNcz/Pxxx9363gePnxYkZGR6tatm+rVq2fWcAEA8Jjlep5BQUGuTJkkHT16VBs2bJDD4SjruQIAYArL9jydFi9erCpVqmjgwIHXHBc9TwCAr1iy53k5h8OhqKgot6vR4tDzBAD4iiV7nk5btmzRnj17FB0dfd1x0fMEAPiK6es8L+95btiwodieZ1RUlI4cOaKqVavKz89Pc+fOLbbn+f7776tt27Zq3779dT/XbreTLAMA+IThK8/S9jzXr19/3Z5ncHCwkpKSiu15nj59WsuXLy/VVScAAL5kuZ6nU1JSks6fP68nnnjChNMEAMA8lut5OjkcDj3yyCNXTaoAANxohifPgIAAzZs3z+1BIElq3LixSpMGXbx4cak+Z/PmzUaHBgCAT1T4jeF3T4kkhg0AMBUbwwMAYBA9TwAADKrwt23peQLlD31NWB23bQEAMMhyPU9J2rp1q3r27KmaNWuqVq1aioyM1I4dO8waKgAAZWK5nufp06fVt29fNWrUSN9++602bdqkatWqKTIyUoWFhWYNFwAAj/ldLM3izP/ftXqel28W77R161bdc889ys7OVqNGjfTuu+9q0qRJOnz4sGsj+F27dqlNmzb68ccf1axZM23btk133323Dhw44CqjXHlMaZw6depSXeX55XznCZQzfOcJMznng5MnT5q2dNFyPc/w8HDdcsstcjgcOnfunM6ePSuHw6GIiAg1bty4xHHR8wQA+Irlep7VqlXTF198oSVLligoKEjBwcFau3at1qxZI5ut5IeD6XkCAHzFcj3Ps2fPKjo6Wp07d1Zqaqq++eYb3XnnnerXr5/Onj1b4rjoeQIAfMVyPc9ly5Zp//792rJli2tCXbZsmWrVqqVVq1Zp0KBBxX4uPU8AgK8YnjxL2/PcuHHjdXuekrRw4UK3nueZM2fk7+/vdnXr/LmoqMjocAEAMJ3h27bOnmdWVpYSExMVHx+vsWPHunqe27Zt09KlS109z5ycHJ07d8719/Pnz1d6err27t2rBQsWaPTo0Zo+fbrroaLevXvr+PHjGjVqlPbs2aPMzEwNGzZMNptNPXr0MO/MAQDwkOV6ni1bttTq1as1ZcoUdezYUf7+/mrXrp3Wrl2rkJCQMpwqAADmMLzOs23btpo3b543x2QK1nkC5RfrPGEmb6zzrPAbw9PzBACYjY3hAQAwiJ4nAAAGVfjbtvQ8AWvh+0xUBNy2BQDAICZPAAAMsmQM+9///rc6deqkatWqqUGDBnrxxRd1/vx5s4YKAECZWC6GvWPHDj344IPq27evtm/frqSkJH3yySeu+goAADea5WLYEydOVEpKirZu3ep6n9WrV+uxxx7Tb7/9pmrVqpVqrGySAFgTDwzB126KGHZBQYECAwPd3icoKEi///670tLSShwXMWwAgK9YLoYdGRmpzZs3KzExURcuXNDBgwc1depUt2OKQwwbAOArloth9+nTR7NmzdKzzz4ru92uFi1a6MEHH7w0WP+Sh0sMGwDgK6YvVbk8hp2SklJsDDsnJ0cHDx7UsWPH9Nprr+no0aOuGLYkjR8/XidOnNCBAweUm5urAQMGSJLbMVey2+2qXr262wsAAG+wXAzbyc/PT7feeqskKTExUaGhobrrrruMDhcAANMZnjydMeyRI0cqPT1d8fHxmjNnjiuGnZ6eruTkZFcMW5Jq166typUrS7oUw+7UqZOCg4OVkpKiCRMm6I033nA9VCRJs2bNUt++feXv76+VK1fqjTfe0PLly1WpUiWTThsAAM9ZLoYtSWvWrNG0adNUUFCgP/zhD1q1apUeeOABD08RAABzVfgYtpnregAA5Y8l1nkCAHCzY/IEAMCgCh/DpucJ4EZjS8KKhytPAAAMYvIEAMAg0yfP1157TS1btlTVqlVVq1Yt9erV66qNFfLy8jRkyBBVr15dNWvWVHR09FVNz507d6pLly4KDAxUaGioZs6cafZQAQDwiOmTZ4sWLTR//nzt2rVLmzZtUuPGjdWnTx8dPXrUdcyQIUOUmZmplJQUJScn66uvvtKIESNcvz916pT69OmjsLAwpaWladasWXrttdf07rvvmj1cAAAMM7TOUzLe9HSur1m/fr169uypPXv2qFWrVtq6dav++Mc/SpLWrl2rBx98UL/++qtuvfVWvf3223r55ZeVk5Pj2pnopZde0scff6wffvihVOOk5wnAKnhg6MayzDrP0jY9z507p3fffVc1atTQH/7wB0nSli1bVLNmTdfEKUm9evWSv7+/6/buli1b1LVrV9fEKV1KlWVlZen48ePFjomeJwDAVzyaPK/X9ExOTlZwcLACAwMVFxenlJQU1alTR5KUk5OjevXqub2fzWZT7dq1XXvh5uTkuDaOd3L+7DzmSvQ8AQC+4tHkeb2mZ48ePZSRkaHNmzerb9++euyxx/Tbb7+ZM+IS0PMEAPiKV5aqVK1aVc2aNVOHDh3kcDhks9nkcDgkSQ0aNLhqIj1//rzy8vLUoEED1zFHjhxxO8b5s/OYK9HzBAD4ikeT57WansUpKipSQUGBpEtXqSdOnFBaWprr9xs2bFBRUZHuvfde1zFfffWVCgsLXcekpKQoPDxctWrV8mTIAACYxqPJ09n0zMrKUmJiouLj4zV27Fjl5+dr4sSJSk1NVXZ2ttLS0jR8+HAdPHhQf/rTnyRJERER6tu3r2JiYvTdd9/pm2++0ejRozVo0CBX/DoqKkqVK1dWdHS0MjMzlZSUpDfffFPjx48378wBAPCQ4Z6nVHLTs6CgQD/88IMWLVqk3Nxc3XLLLbr77rv19ddf64477nD9/dKlSzV69Gj17NlT/v7+evTRR/XWW2+5fl+jRg2tW7dOo0aNUvv27VWnTh29+uqrbmtBAQC4UTxa51kemp70PAEAkoXWeQIAcDNj8gQAwCDD33mWx6YnAABm8uiBofKEGDYAK2B/24qF27YAABhk6uRpRsvz999/19NPP63WrVvLZrPp4YcfNnOIAACUmamTpxktzwsXLigoKEjPPfecevXqZebwAAAwhaF1nr5oeV7u6aef1okTJ/Txxx8bPjF6ngCshO88bxxLrPP0dsvTU/Q8AQC+YvhpW2fL08/PT+Hh4dq1a5fi4uIUExMj6VLLc9CgQTpz5oxCQkIMtzw9NX36dE2ZMqVM7wEAQGkYvvK0YstToucJAPAd05eqlLXl6Sl6ngAAXzE8eXq75QkAgNUZ/s7T2fIcOXKk0tPTFR8frzlz5ig/P1/Tpk1T//79FRISotzcXC1YsKDEluc777yjwsLCq1qekvT999/r3LlzysvL03//+19lZGRIktq2bWvSaQMA4DnDk6e3W56S9OCDDyo7O9v1c7t27SRJButpAAB4heHJMyAgQPPmzdPbb7/t9u+BgYFauXLldf++du3aWrZs2TWP2b9/v9FhAQDgMxV+Y/jdUyJ5eAgAYCo2hgcAwCBDV560PAEAuAlu29LzRFmwHymA4nDbFgAAg0yfPM1oen7xxRcaMGCAQkJCVLVqVbVt21ZLly41e6gAAHjE9MnTjKbn5s2b1aZNG3300UfauXOnhg0bpqeeekrJyclmDxcAAMMM9Twl3zc9nfr1WRAI1gAAIABJREFU66f69etr4cKFpRonPU+Yge88gfLPEj1P6cY0PU+ePKnatWuX+Ht6ngAAX/HoaVtfNz2XL1+urVu36h//+EeJY6LnCQDwFY+uPH3Z9Ny4caOGDRum9957z22P3CvR8wQA+IpXlqqY1fT88ssv9T//8z+Ki4vTU089dc3PpOcJAPAVjyZPXzQ9v/jiC/Xr108zZsxwexIXAIAbzaPJ09n0zMrKUmJiouLj4zV27Fjl5+dr4sSJSk1NVXZ2ttLS0jR8+PASm57fffedvvnmm6uanhs3blS/fv303HPP6dFHH1VOTo5ycnKUl5dn3pkDAOAhjx4Y8nbTc9GiRTpz5oymT5+u6dOnu/69W7du7K8LALjhPFrn2bZtW82bN89bYzIF6zxhBtZ5AuWfN9Z5VviN4el5AgDMxsbwAAAYZPjKk+8cAQA3uwp/25aeJ4Dr4bttGMVtWwAADGLyBADAIEvGsLOystSjRw/Vr19fgYGBatq0qV555RUVFhaaPVwAAAwz/TtPZwy7adOmOnv2rOLi4tSnTx/t27dPdevWlXQphn348GGlpKSosLBQw4YN04gRI7Rs2TJJUkBAgJ566indddddqlmzpnbs2KGYmBgVFRXp9ddfN3vIAAAYUm5i2OPHj9fWrVv19ddfl2qcbJIAoLR4YKhiu2lj2Pv27dPatWvVrVu3EsdEDBsA4CseTZ7OGHZ4eLiGDBmiMWPGKC4uzvX75ORkBQcHKzAwUHFxcR7HsDt16qTAwEA1b95cXbp00dSpU0sc0/Tp01WjRg3XKzQ01JNTAwDguiwdw05KSlJ6erqWLVumTz/9VLNnzy7xWGLYAABf8comCc4YtjOI3bx5czkcDsXGxhqKYTuvHlu1aqULFy5oxIgR+stf/lJsN9Rut8tut3vjdAAAcGPZGHZx71FYWKiioiJPhgwAgGk8uvJ0xrBHjhyp9PR0xcfHa86cOcrPz9e0adPUv39/hYSEKDc3VwsWLCgxhv3OO++osLDwqhj20qVLFRAQoNatW8tut2vbtm2KjY3V448/roCAAPPOHgAAD1gyhm2z2TRjxgzt3btXFy9eVFhYmEaPHq1x48aV/YwBACijCh/DNnNdDwCg/LHMOk8AAG5mTJ4AABhU4WPY9DyBq7EdHVA2XHkCAGAQkycAAAZZsud5uX379qlatWqqWbOm2UMFAMAjpk+ezp7nrl27tGnTJjVu3Fh9+vTR0aNHXccMGTJEmZmZSklJUXJysr766iuNGDHiqvcqLCzU4MGD1aVLF7OHCQCAxyzd83zxxRd16NAh9ezZU88//7xOnDhR6nHS8wRKxgNDuJlYZp2nL3qeGzZs0IoVK7RgwYJSjYmeJwDAVzzans/Z8/Tz81N4eLh27dqluLg4xcTESLrU8xw0aJDOnDmjkJAQwz3PY8eO6emnn9aSJUtK/b+E6dOna8qUKZ6cDgAAhliy5xkTE6OoqCh17dq11H9DzxMA4CteWari7Hl26NBBDodDNptNDodDkkrV89ywYYNmz54tm80mm82m6OhonTx5UjabTQsXLiz2M+12u6pXr+72AgDAGzy6bWtWz7N9+/aSru55btmyxXUVK0mrVq3SjBkztHnzZjVs2NCTIQMAYBpL9jwjIiLcPm/btm3y9/d3PeULAMCNZMmeJwAAVkbPEwBQoVlmnScAADczJk8AAAyq8D1PAADM5tEDQ+UJMWwAvsS+wTcHbtsCAGCQJXue+/fvl5+f31Wv1NRUs4cLAIBhlu55rl+/XocPH3a9nDsSAQBwI1my57l//341adJE27dvV9u2bT06MXqeAG4EvvO0Hsus8/RFz1OS+vfvr3r16um+++7TJ598cs0x0fMEAPiKJXuewcHBmjNnjjp37ix/f3999NFHevjhh/Xxxx+rf//+xY6JnicAwFcs2fOsU6eOxo8fr3vvvVd333233njjDT3xxBOaNWtWiX9DzxMA4CuW7HkW595779W+fftK/D09TwCAr3g0eZrV83S6sudZnIyMDIWEhHgyXAAATGXJnueiRYtUuXJltWvXTpK0cuVKLVy4sNiHkgAA8DXL9jz/9re/KTs7WzabTS1btlRSUpIGDhxYtrMFAMAEFb7nyTpPAL7EOk/r8cY6zwq/MfzuKZE8PAQAMBUbwwMAYBA9TwAADKrwt23peQKoSPhO1Rq4bQsAgEGmTp5mtDwl6eLFi5o9e7ZatGghu92uhg0batq0aWYOFQAAj5l629bZ8mzatKnOnj2ruLg49enTR/v27VPdunUlXWp5Hj58WCkpKSosLNSwYcM0YsQILVu2zPU+Y8eO1bp16zR79my1bt1aeXl5ysvLM3OoAAB4zNA6T1+0PPfs2aM2bdpo9+7dCg8P9/jEWOcJoCLiO0/jLNHz9HbLc/Xq1WratKmSk5PVpEkTNW7cWM8888x1rzzpeQIAfMXw5OlseYaHh2vIkCEaM2aM4uLiXL9PTk5WcHCwAgMDFRcXZ7jl+fPPPys7O1srVqzQ4sWLlZCQoLS0tOtuzTd9+nTVqFHD9QoNDTV6agAAlIrhydPbLU9ngWXx4sXq0qWLunfvLofDoY0bNyorK6vEv6PnCQDwFdOXqpS15RkSEiKbzaYWLVq4jomIiJB0qeZSEnqeAABfMTx5ervl2blzZ50/f14//fST65i9e/dKksLCwowOFwAA0xmePJ0tz6ysLCUmJio+Pl5jx45Vfn6+Jk6cqNTUVGVnZystLU3Dhw8vseX53Xff6Ztvvrmq5dmrVy/dddddGj58uLZv3660tDSNHDlSvXv3drsaBQDgRjG8ztPbLU9/f3+tXr1aY8aMUdeuXVW1alU98MADmjNnjjlnDABAGRle51keWp4S6zwBVEys8zSOnqcH6HkCAMzGxvAAABhk6MqTlicAADfBbVt6ngCsiO8uyzdu2wIAYBCTJwAABpk+eZoRxH7ttdfk5+d31atq1apmDxcAAMNMnzydQexdu3Zp06ZNaty4sfr06aOjR4+6jhkyZIgyMzOVkpKi5ORkffXVVxoxYoTr9y+88IIOHz7s9mrVqpVrpyIAAG4kQ5skSL4JYl9px44datu2rb766it16dKl2HEVFBS49tB1fm5oaCibJACwJB4Y8h1LxLAl7wexr/T++++rRYsWJU6cEj1PAIDveDR5ejuIfbnff/9dS5cuVXR09DXHRM8TAOArHk2e3g5iX+7//u//9N///ldDhw695nH0PAEAvuKVpSplDWJf7v3339dDDz2k+vXre2OoAAAY5tHk6e0gttMvv/yijRs3XveWLQAAvuTR5OntILbTwoULFRISogceeKDsZwoAgEk82tvW20Fs6dLVakJCgp5++ukSr2gBALgRPFrnWR6C2N5Y1wMAKH8ss84TAICbGZMnAAAGGf7Os7wFsel5AtbBlnSoKLjyBADAICZPAAAMMnXyNKPlKUmff/65OnTooGrVqqlu3bp69NFHtX//fjOHCgCAx0ydPM1oef7yyy8aMGCA7r//fmVkZOjzzz9Xbm6uHnnkETOHCgCAxwyt8/RFy/PDDz/U4MGDVVBQIH//S3P76tWrNWDAABUUFCggIKBUY3V+Nj1PwDp4YAg3giXWeXq75dm+fXv5+/vrgw8+0IULF3Ty5En985//VK9eva45cRYUFOjUqVNuLwAAvMHw5OntlmeTJk20bt06TZw4UXa7XTVr1tSvv/6q5cuXX3NcxLABAL5iePL0dsszJydHMTExGjp0qLZu3aovv/xSlStX1sCBA3WtO8zEsAEAvuLRxvDX4mx5OnuezZs3l8PhUGxsbKlangsWLFCNGjU0c+ZM1zFLlixRaGiovv32W3Xo0KHYz7Xb7bLb7WafDgAAVzF85entlueZM2dcDwo5Od+7qKjI6HABADCd4cnT2y3Pfv36aevWrZo6dap+/PFHpaena9iwYQoLC1O7du3MPXsAADxg+Latt1ue999/v5YtW6aZM2dq5syZqlKlijp27Ki1a9cqKCjInLMGAKAMDK/zLA8tT4meJwDgEkus8wQA4GbH5AkAgEGGvvMsby1PiZ4ncDNg2z/4GleeAAAYxOQJAIBBPu95Tps2TZ06dVKVKlVUs2bNYt/nwIED6tevn6pUqaJ69eppwoQJOn/+vJlDBQDAYz7veZ47d05/+tOf9Oc//7nY97hw4YL69eunc+fOafPmzVq0aJESEhL06quvmjlUAAA85tOe5+USEhL0/PPP68SJE27/vmbNGj300EM6dOiQ6tevL0l655139OKLL+ro0aOqXLlyqcZKzxO4efDAEK7FEus8y9LzLI0tW7aodevWrolTkiIjI3Xq1CllZmaW+Hf0PAEAvmJ4ez5nz9PPz0/h4eHatWuX4uLiFBMTI+lSz3PQoEE6c+aMQkJC3HqepZGTk+M2cUpy/exsfhZn+vTpmjJlitHTAQDAMMv1PD1FzxMA4CumL1Vx9jw7dOggh8Mhm80mh8NR6r9v0KCBjhw54vZvzp+dzc/i2O12Va9e3e0FAIA3+LTnWRodO3bUrl273K5WU1JSVL16dbVq1crocAEAMJ1Pe57Ov8/IyNCBAwd04cIFZWRkKCMjQ6dPn5Yk9enTR61atdKTTz6pHTt26PPPP9crr7yiUaNGyW63m3fmAAB4yOc9z1dffVWLFi1y/ewMXG/cuFHdu3dXpUqVlJycrD//+c/q2LGjqlatqqFDh2rq1KkmnC4AAGVHzxMAUKFZYp0nAAA3OyZPAAAMqvA9TwAAzGb4gaHyhhg2UDL2hAU8w21bAAAMsmTP87nnnlP79u1lt9vVtm1bM4cIAECZWa7n6TR8+HA9/vjjZg4PAABTGPrO83o9z6ioKLfj586dK4fDoZ07d7p6ns7ySUJCQomf89Zbb0mSjh49qp07dxoZIgAAXme5nqen6HkCAHzFcj1PT9HzBAD4Cj1PAAAMslzP01P0PAEAvmK5nicAAFZn+DtPZ89z5MiRSk9PV3x8vObMmaP8/HxNmzZN/fv3V0hIiHJzc7VgwYJie555eXluPU9JatasmYKDgyVJ+/bt0+nTp5WTk6OzZ8+6jmnVqpUqV65sxnkDAOAxy/U8JemZZ57Rl19+edUxv/zyixo3buzJeQIAYJoK3/MMfX45e9sCJWBvW9wMvNHzrPAbw++eEsnDQwAAU7ExPAAABtHzBADAoAp/25aeJ+CO7zmBsuO2LQAABjF5AgBgkOVi2Dt27NDgwYMVGhqqoKAgRURE6M033zRzmAAAlInlYthpaWmqV6+elixZoszMTL388suKjY3V/PnzzRwqAAAeM7xJwrVi2FdyLkxdv369K4btlJCQoOeff14nTpy47ueOGjVKe/bs0YYNG0o8pqCgwG0P3VOnTik0NJRNEoAr8MAQbjbe2CShXMSwT548qdq1a1/zmOnTp6tGjRquV2hoaJk+EwCAklg+hr1582YlJSXp008/veZxsbGxGj9+vOtn55UnAABms3QMe/fu3RowYIAmT56sPn36XPNYep4AAF+xbAz7+++/V8+ePTVixAi98sorZg8TAACPWTKGnZmZqR49emjo0KGaNm2a0SECAOBVloth7969W/fff78iIyM1fvx45eTkSJIqVaqkunXrmnTaAAB4znIx7A8//FBHjx7VkiVLtGTJEtdxYWFh2r9/fxlOFQAAc1T4GLaZ63oAAOWPJdZ5AgBws2PyBADAoAofw6bniYqKbfaAG4crTwAADGLyBADAIMv1PI8dO6a+ffvq1ltvld1uV2hoqEaPHq1Tp06ZOVQAADxmuZ6nv7+/BgwYoE8++UR79+5VQkKC1q9fr2effdbMoQIA4LFy0fN86623NGvWLP3nP/8p7VBdn03PExUVDwwBpWOJdZ6+7nkeOnRIK1euVLdu3a55XEFBgU6dOuX2AgDAGwxPns6eZ3h4uIYMGaIxY8YoLi7O9fvk5GQFBwcrMDBQcXFxHvc8Bw8erCpVqqhhw4aqXr16sRP05YhhAwB8xbI9z7i4OKWnp2vVqlX66aef3ELXxYmNjdXJkyddLyO3eAEAMMLwxvDX4+x5OpuezZs3l8PhUGxsrKH3adCggRo0aKCWLVuqdu3a6tKliyZNmqSQkJBij7fb7bLb7WacAgAA12TJnmdx7yGpzO8DAIAZLNfz/Oyzz3TkyBHdfffdCg4OVmZmpiZMmKDOnTurcePGpp04AACeslzPMygoSO+9957GjRungoIChYaG6pFHHtFLL71kwukCAFB29DwBABWaJdZ5AgBws2PyBADAIHqeACC2O4QxXHkCAGAQkycAAAZZrud5uWPHjum2226Tn59fqeorAAD4guV6npeLjo5WmzZtzBwiAABlZmjy7N69u0aPHq3Ro0erRo0aqlOnjiZNmiTnUtGoqCj16tVLTZs21R133KG5c+fq1KlT2rlzp+s9pkyZonHjxql169bX/Ky3335bJ06c0AsvvODBaQEA4D2W7Hl+//33mjp1qhYvXix//9INkZ4nAMBXLNfzLCgo0ODBgzVr1iw1atSo1H9HzxMA4CuW63nGxsYqIiJCTzzxhKFx0fMEAPiK6UtVnD3PDh06yOFwyGazyeFwlPrvN2zYoBUrVshms8lms6lnz56SpDp16mjy5Mkl/p3dblf16tXdXgAAeIPhqoq3e54fffSRzp496/p569atGj58uL7++mvdfvvtRocLAIDpLNfzvHKCzM3NlSRFRERcd10oAAC+YLmeJwAAVkfPEwBQodHzBADAApg8AQAwqML3PAEAMJvhB4bKG2LYQNkQiQauxm1bAAAMsmzPMyEhQW3atFFgYKDq1aunUaNGmTlUAAA8Zsme59y5c/Xyyy/rpZdeUmZmptavX6/IyEgzhwoAgMcMr/O88847JUn//Oc/FRAQoD//+c+aOnWq22bxTs61NevXr3ftUeuUkJCg559/XidOnHD79+PHj6thw4ZavXr1VX9jhPOzQ59fzneeQBnwnSfKO0us8/R2zzMlJUVFRUU6ePCgIiIidNttt+mxxx67biWFnicAwFcs1/P8+eefVVRUpNdff13z5s3Thx9+qLy8PPXu3Vvnzp0r8e/oeQIAfMVyPc+ioiIVFhbqrbfeUmRkpDp06KDExET9+OOP2rhxY4l/R88TAOArlut5hoSESJJatWrl+re6deuqTp06OnDgQIl/R88TAOArhidPb/c8O3fuLEnKyspy/VteXp5yc3MVFhZmdLgAAJjO8OTp7HlmZWUpMTFR8fHxGjt2rPLz8zVx4kSlpqYqOztbaWlpGj58eLE9z4yMDLeeZ0ZGhk6fPi3p0nKXAQMGaOzYsdq8ebN2796toUOHqmXLlurRo4d5Zw4AgIcs2fNcvHixxo0bp379+snf31/dunXT2rVrFRAQUMbTBQCg7Cp8z5N1nkDZsM4T5Z031nlW+I3hd0+J5OEhAICp2BgeAACD6HkCAGBQhb9tS88TqBj47hVWwm1bAAAMslzPMyEhQX5+fsW+jGzzBwCAt1iu5/n444/r8OHDbq/IyEh169ZN9erVM3O4AAB4xNB3ntfreUZFRbkdP3fuXDkcDu3cudPV5pwyZYqkS1eYxQkKClJQUJDr56NHj2rDhg2G9scFAMCbLNfzvNLixYtVpUoVDRw48JrH0fMEAPiK4adtnT1PPz8/hYeHa9euXYqLi1NMTIykSz3PQYMG6cyZMwoJCTHc87ySw+FQVFSU29VocaZPn+66qgUAwJss1/O83JYtW7Rnzx5FR0df91h6ngAAX7Fcz/Ny77//vtq2bav27dtf91h6ngAAX7Fcz9Pp9OnTWr58eamuOgEA8CXD33k6e54jR45Uenq64uPjNWfOHOXn52vatGnq37+/QkJClJubqwULFhTb88zLy3PreUpSs2bNFBwc7DouKSlJ58+f1xNPPGHCaQIAYB5L9jylSw8KPfLII8VupAAAwI1EzxNAucDetvAUPU8P0PMEAJiNjeEBADCInicAAAZV+Nu29DyBioXvPmEF3LYFAMAgJk8AAAyyXAxbkrZu3aqePXuqZs2aqlWrliIjI7Vjxw4zhwoAgMcsF8M+ffq0+vbtq0aNGunbb7/Vpk2bVK1aNUVGRqqwsNDM4QIA4BHDmyRcK4Z9JefC1PXr17ti2E4JCQl6/vnndeLECbd/37Ztm+6++24dOHBAoaGhkqRdu3apTZs2+vHHH9WsWbNSjZVNEoCKiQeGYJQ3NkmwXAw7PDxct9xyixwOh86dO6ezZ8/K4XAoIiJCjRs3LvHviGEDAHzF8OTpjGGHh4dryJAhGjNmjOLi4ly/T05OVnBwsAIDAxUXF2c4hl2tWjV98cUXWrJkiYKCghQcHKy1a9dqzZo1stlKXlkzffp01ahRw/VyXrUCAGA2y8Wwz549q+joaHXu3Fmpqan65ptvdOedd6pfv346e/ZsiX9HDBsA4Cumb5LgjGE7g9jNmzeXw+FQbGxsqf5+2bJl2r9/v7Zs2SJ/f3/Xv9WqVUurVq3SoEGDiv07u90uu91u2nkAAFASy8Wwz5w5I39/f7erW+fPRUVFRocLAIDpDE+ezhh2VlaWEhMTFR8fr7Fjxyo/P18TJ05UamqqsrOzlZaWpuHDhxcbw87IyHCLYWdkZOj06dOSpN69e+v48eMaNWqU9uzZo8zMTA0bNkw2m009evQw78wBAPCQ5WLYLVu21OrVqzVlyhR17NhR/v7+ateundauXauQkBATThkAgLKp8DFsM9f1AADKH0us8wQA4GbH5AkAgEEVPoZNzxMAboyKvJUiV54AABjE5AkAgEGmT57Xa3ru379f0dHRatKkiYKCgnT77bdr8uTJOnfunOuYrKws9ejRQ/Xr11dgYKCaNm2qV155hSQZAMASTN+ez9n0bNq0qc6ePau4uDj16dNH+/btU926dfXDDz+oqKhI//jHP9SsWTPt3r1bMTExys/P1+zZsyVJAQEBeuqpp3TXXXepZs2a2rFjh2JiYlRUVKTXX3/d7CEDAGCIoXWekrlNT6dZs2bp7bff1s8//1zi544fP15bt27V119/Xapx0vMEgBvLKg8MWWadp9lNz5MnT6p27dol/n7fvn1au3atunXrVuIx9DwBAL7i0eRpZtNz3759io+P18iRI6/6XadOnRQYGKjmzZurS5cumjp1aoljoucJAPAVjyZPs5qeBw8eVN++ffWnP/1JMTExV/0+KSlJ6enpWrZsmT799FPXd6LFoecJAPAV0x8YkkrX9Dx06JB69OihTp066d133y32fZxXj61atdKFCxc0YsQI/eUvfyk2f0bPEwDgKx5deZa16Xnw4EF1795d7du31wcffOCKXl9LUVGRCgsLaXoCAG44j648nU3PkSNHKj09XfHx8ZozZ47y8/M1bdo09e/fXyEhIcrNzdWCBQvcmp7OiTMsLEyzZ8/W0aNHXe/boEEDSdLSpUsVEBCg1q1by263a9u2bYqNjdXjjz+ugIAAE04bAADPeTR5lqXpmZKSon379mnfvn267bbb3N7XuWrGZrNpxowZ2rt3ry5evKiwsDCNHj1a48aNK+PpAgBQdh6t8ywPTU96ngAAyULrPAEAuJkxeQIAYJDh7zzLY9MTAAAzeWWdp5UQwwauzSr7jwLlCbdtAQAwyNTJ04yWp3Rpycrs2bPVokUL2e12NWzYUNOmTTNzqAAAeMzU27ZmtDwlaezYsVq3bp1mz56t1q1bKy8vT3l5eWYOFQAAjxla5+mLlueePXvUpk0b7d69W+Hh4Z6ck9tn0/MEro3vPFHRWWKdp7dbnqtXr1bTpk2VnJysJk2aqHHjxnrmmWeue+VJzxMA4CuGJ09vtzx//vlnZWdna8WKFVq8eLESEhKUlpamgQMHXnNc9DwBAL5iePL0dsvTWWBZvHixunTpou7du8vhcGjjxo3KysoqcVz0PAEAvmL6UhVny7NDhw5yOByy2WxyOBxux1yr5RkSEiKbzaYWLVq4/i0iIkLSpZpLSex2u6pXr+72AgDAGwxPnt5ueXbu3Fnnz5/XTz/95Pq3vXv3SpLCwsKMDhcAANMZnjydLc+srCwlJiYqPj5eY8eOVX5+viZOnKjU1FRlZ2crLS1Nw4cPL7bl2ahRI1fLMycnRzk5Oa7379Wrl+666y4NHz5c27dvV1pamkaOHKnevXu7XY0CAHCjGF7n6e2Wp7+/v1avXq0xY8aoa9euqlq1qh544AHNmTPHhNMFAKDsDK/zLA8tT4l1nkBpsc4TFZ031nlW+I3hd0+J5OEhAICp2BgeAACDDF150vIEAOAmuG1LzxOAN/Bd8c2N27YAABhkyZ7n8uXL1bZtW1WpUkVhYWGaNWuWmcMEAKBMLNfzXLNmjYYMGaL4+Hj16dNHe/bsUUxMjIKCgjR69GgzhwsAgEcs1/OMiopSYWGhVqxY4TomPj5eM2fO1IEDB4r9nOKwzhOAN/GdZ/lxU/Q8CwoKFBgY6HZMUFCQfv31V2VnZ5f4PvQ8AQC+YrmeZ2RkpFauXKl///vfKioq0t69e11b8x0+fLjEcdHzBAD4iuV6njExMRo9erQeeughVa5cWR06dNCgQYMuDda/5OHS8wQA+Irlep5+fn6aMWOGTp8+rezsbOXk5Oiee+6RJDVt2rTEz6XnCQDwFcNP25rR8+zRo0eJPU+nSpUqqWHDhpKkxMREdezYUXXr1jU6XAAATGd48nT2PEeOHKn09HTFx8drzpw5ys/P17Rp09S/f3+FhIQoNzdXCxYsKLbnGRYW5up5OjVo0ECSlJubqw8//FDdu3fX77//rg8++EArVqzQl19+adIpAwBQNpbreUqXnuh94YUXdPHiRXXs2FFffPGF69YtAAA3Gj1PAPAA6zzLD3qeHqDnCQAwGxvDAwBgED1PAAAMqvC3bel5AgCxzqoIAAAgAElEQVTf0ZqN27YAABjE5AkAgEGmT57XC2JLUv/+/dWoUSMFBgYqJCRETz75pA4dOuR2zM6dO9WlSxcFBgYqNDRUM2fONHuoAAB4xPTJ0xnE3rVrlzZt2qTGjRurT58+brsJ9ejRQ8uXL1dWVpY++ugj/fTTTxo4cKDr96dOnVKfPn0UFhamtLQ0zZo1S6+99tpV++ACAHAjGNokQfJOEPuTTz7Rww8/rIKCAgUEBOjtt9/Wyy+/rJycHFWuXFmS9NJLL+njjz/WDz/8UKpxskkCAPw/N/MDQ5aIYUvmBrHz8vK0dOlSderUSQEBAZKkLVu2qGvXrq6JU7rU+czKytLx48eLfR9i2AAAX/Fo8jQjiP3iiy+qatWquuWWW3TgwAGtWrXK9bucnBzVr1/f7Xjnzzk5OcWOiRg2AMBXPJo8zQhiT5gwQdu3b9e6detUqVIlPfXUUzJ4B9kNMWwAgK94ZZMEZxDbGcVu3ry5HA6HYmNjXcfUqVNHderUUYsWLRQREaHQ0FClpqaqY8eOatCggY4cOeL2ns6fnemyK9ntdtntdm+cDgAAbjy68ixrELu430tyHdOxY0d99dVXKiwsdB2TkpKi8PBw1apVy5MhAwBgGo8mT2cQOysrS4mJiYqPj9fYsWOVn5+viRMnKjU1VdnZ2UpLS9Pw4cPdgtjffvut5s+fr4yMDGVnZ2vDhg0aPHiwbr/9dnXs2FGSFBUVpcqVKys6OlqZmZlKSkrSm2++qfHjx5t35gAAeMij27ZlCWJXqVJFK1eu1OTJk5Wfn6+QkBD17dtXr7zyiuu2a40aNbRu3TqNGjVK7du3V506dfTqq69qxIgR5p05AAAe8midZ3kIYntjXQ8AoPyxzDpPAABuZkyeAAAYZPg7z/IWxKbnCdwYN/N2cKj4uPIEAMAgJk8AAAyyZM8zKytLPXr0UP369RUYGKimTZvqlVdecds0AQCAG8WSPc+AgAA99dRTWrdunbKysjRv3jy99957mjx5stnDBQDAMEv2PIszfvx4bd26VV9//XWpxknPE7ixeGAIVmGZdZ7e7nlead++fVq7dq26detW4pjoeQIAfMWSPU+nTp06KTAwUM2bN1eXLl00derUEsdEzxMA4CuW7nkmJSUpPT1dy5Yt06effqrZs2eXOCZ6ngAAX7Fkz9PJefXYqlUrXbhwQSNGjNBf/vKXYtNn9DwBAL5iyZ5nSccUFha6jgUA4Ebx6MrT2fMcOXKk0tPTFR8frzlz5ig/P1/Tpk1T//79FRISotzcXC1YsOCqnufWrVt13333qVatWvrpp580adIkt57n0qVLFRAQoNatW8tut2vbtm2KjY3V448/XuJDRQAA+Iole542m00zZszQ3r17dfHiRYWFhWn06NEaN26ceWcOAICH6HkCACo0y6zzBADgZsbkCQCAQfQ8gQqMLfIA7+DKEwAAg5g8AQAwyJI9z8vt27dP1apVU82aNc0eKgAAHrFkz9OpsLBQgwcPVpcuXcweJgAAHrN0z/PFF1/UoUOH1LNnTz3//PM6ceJEqcdJzxPggSFAstA6T1/0PDds2KAVK1ZowYIFpRoTPU8AgK9Ysud57NgxPf3000pISCj1/xLoeQIAfMWSPc+YmBhFRUWpa9eupR4TPU8AgK94ZamKs+fZoUMHORwO2Ww2ORwOt2OcLc/evXvrX//6lz777DOlpqZKunTLdvbs2bLZbLLZbIqOjtbJkydls9m0cOHCYj/TbrerevXqbi8AALzBo6qKt3ueW7ZscV3FStKqVas0Y8YMbd68WQ0bNvRkyAAAmMaSPc+IiAi3z9u2bZv8/f1dT/kCAHAjWbLnCQCAldHzBABUaJZZ5wkAwM2MyRMAAIMqfM8TAACzefTAUHlCDBtAecXexNbFbVsAAAyyZM9z//798vPzu+rl3IEIAIAbydI9z/Xr1+vw4cOuV/v27c0eLgAAhlmy57l//341adJE27dvV9u2bT04LXqeAMo/vvM0h2XWefqi5yldur1br1493Xffffrkk0+uOSZ6ngAAX7FkzzM4OFhz5szRihUr9Omnn+q+++7Tww8/fM0JlJ4nAMBXPLpt27RpU7c02KpVqzRw4ED9/vvvqlSpkvLz83X48GHl5ubqvffe04YNG/Ttt9+qXr16rr/Jzc1VXl6esrOzNWXKFNWoUUPJycnF3vqVLu2n+8svv+jrr78u9vcFBQVu5ZZTp04pNDSU27YAyi1u25rDG7dtvbLO09nzdDY9mzdvLofDodjYWNcxderUcTU9IyIiFBoaqtTUVFdZ5Ur33nuvUlJSSvxMu93OxvIAAJ/w6Latt3uexcnIyFBISIgHowUAwFyW7HkuWrRIlStXVrt27SRJK1eu1MKFC4t9KAkAAF+zbM/zb3/7m7Kzs2Wz2dSyZUslJSUVuxYUAABfq/A9Tx4YAlBe8cCQOcrNA0NWsntKJDFsAICp2BgeAACD6HkCAGBQhb9tS88T5Q3fcwHWx21bAAAMYvIEAMAgUydPM0LYknTx4kXNnj1bLVq0kN1uV8OGDTVt2jQzhwoAgMdMnTzNCmGPHTtW77//vmbPnq0ffvhBn3zyie655x4zhwoAgMcMbZLgixD2nj171KZNG+3evVvh4eGlPhGqKqgoeGAIMJclYtjeDmGvXr1aTZs2VXJyspo0aaLGjRvrmWeeUV5e3jXHRc8TAOArhidPb4ewf/75Z2VnZ2vFihVavHixEhISlJaWdt19bWNjY3Xy5EnX6z//+Y/RUwMAoFQMT54dOnRwu0XbsWNH/fjjj7pw4YKkS99pZmRkaPPmzerbt68ee+wx/fbbb27vMWHCBG3fvl3r1q1TpUqV9NRTT8l599iZL1u8eLG6dOmi7t27y+FwaOPGjcrKyipxXHa7XdWrV3d7AQDgDaYvVXGGsDt06CCHwyGbzSaHw+F2jDOC3bt3b/3rX//SZ599ptTUVElSSEiIbDabWrRo4To+IiJC0qUUGgAAN5rhydPbIezOnTvr/Pnz+umnn1zH7N27V5IUFhZmdLgAAJjO8PZ83g5h9+rVS3fddZeGDx+uefPmqaioSKNGjVLv3r3drkYBALhRDE+e3g5h+/v7a/Xq1RozZoy6du2qqlWr6oEHHtCcOXPMPXMAADxkeJ1neQhhS95Z1wMAKH8ssc4TAICbHZMnAAAGGfrOszyGsOl5wmxsnweAK08AAAxi8gQAwCBL9jyXL1+utm3bqkqVKgoLC9OsWbPMHCYAAGViuZ7nmjVrNGTIED377LPavXu3/v73vysuLk7z5883c6gAAHjMcj3PqKgoFRYWasWKFa5j4uPjNXPmTB04cKDYzymO87PpecJsPDAElC+WWOfp7Z5nQUGBAgMD3Y4LCgrSr7/+quzs7BLHVVBQoFOnTrm9AADwBsv1PCMjI7Vy5Ur9+9//VlFRkfbu3evamu/w4cMljosYNgDAVyzX84yJidHo0aP10EMPqXLlyurQoYMGDRp0abD+JQ+XGDYAwFcs1/P08/PTjBkzdPr0aWVnZysnJ0f33HOPJKlp06Ylfi4xbACArxiuqni75+lUqVIlNWzYUJKUmJiojh07qm7dukaHCwCA6SzX88zNzdWHH36o7t276/fff9cHH3ygFStW6MsvvzT3zAEA8JDlep7SpSd6X3jhBV28eFEdO3bUF1984bp1CwDAjUbPEwBQoVlinScAADc7Jk8AAAyi5wnLYNs7AOUFV54AABjE5AkAgEGmT56laXo6FRQUqG3btvLz81NGRobb73bu3KkuXbooMDBQoaGhmjlzptlDBQDAI6ZPnqVpejr99a9/1a233nrVv586dUp9+vRRWFiY0tLSNGvWLL322mt69913zR4uAACGGZ48u3fvrtGjR2v06NGqUaOG6tSpo0mTJrk2do+KilKvXr3UtGlT3XHHHZo7d65OnTqlnTt3ur3PmjVrtG7dOs2ePfuqz1i6dKnOnTunhQsX6o477tCgQYP03HPPae7cuR6eJgAA5vHoyrOsTc8jR44oJiZG//znP1WlytVPwm7ZskVdu3ZV5cqVXf8WGRmprKwsHT9+vNgx0fMEAPiKR5NnWZqeFy9e1NNPP61nn31Wf/zjH4t9/5ycHNWvX9/t35w/5+TkFPs39DwBAL7i0eRZlqZnfHy8/vvf/yo2NtaE4f8/9DwBAL7ilaUq12p6btiwQVu2bJHdbpfNZlOzZs0kSX/84x81dOhQSVKDBg105MgRt/d0/tygQYNiP5OeJwDAVzyaPMvS9Hzrrbe0Y8cOZWRkKCMjQ5999pkkKSkpSdOmTZN06Ur2q6++UmFhoes9UlJSFB4erlq1ankyZAAATOPR5OlsemZlZSkxMVHx8fEaO3as8vPzNXHiRKWmpio7O1tpaWkaPny4W9OzUaNGuvPOO12vFi1aSJJuv/123XbbbZIuPbFbuXJlRUdHKzMzU0lJSXrzzTc1fvx4k04bAADPGe55SmVrepZGjRo1tG7dOo0aNUrt27dXnTp19Oqrr2rEiBGeDBcAAFMZ6nlK5afpSc8TACDR88T/1969h8d85/0ffw05IcTZBJFIL4I2jtXcTttUcifV7mLb3lVaqk6tOlu9tLuUchfdkLS1brv3btCurmJpKd1V6nC3hLbizJWKJe7eQpV1KEEk398f/WXWyPEzmZlM4vm4rrkuM/OZ77y/n2smb9/5Hl4AAJ9A8wQAwJDxPs/KmOkJAIA7uXTAUGVCGDaAikTIe9XEz7YAABhya/N0R5bnjRs3NHToUEVHR8vPz0/9+/d3Z4kAAJSbW5unO7I88/LyVKNGDY0fP17x8fHuLA8AALcw2ucZGxurBx54QJL05z//Wf7+/ho9erRmzZolm82mQYMGOY1PTk5WamqqDh48qLi4OMfjBVmea9as0d/+9jen19SqVUuLFy+WJO3cuVOXLl1yacUAAPAU4y1PT2d5uoo8TwCAtxg3T09nebqKPE8AgLcYN09fzPKUyPMEAHiP209VKW+Wp6vI8wQAeIvxRRLKm+X5n//5n47nzpw5o8TERK1cuVIxMTGmpQAAUCGMm2dBlueLL76o9PR0LVy4UAsWLNC1a9f05ptvqm/fvgoNDdUPP/ygRYsWFcryvFNwcLAk5yxPSTp69Khu3bqlixcv6urVq47zQDt27OjyigIA4C7GzdPTWZ6S9NhjjykrK8txv1OnTpJ+OuAIAICKZtw8/f399fbbbzvOxSwQFBSktWvXGi0rIiKiyIZ46tQp07IAAPCaKn9h+MNvJHLwEADArbgwPAAAhoy2PMnyBADgHvjZljxPwLvIr8S9gJ9tAQAwRPMEAMCQ25unOwKxMzIy9Mgjj6hJkyYKCgpSZGSkpk2bptzcXHeXCwCAMbfv8ywIxI6MjFROTo5SUlKUkJCgzMxMNWrUyGlsQSD2gQMHnB739/fXkCFD1LlzZ9WtW1cHDhzQyJEjlZ+frzlz5ri7ZAAAjBg3T28EYkdGRioyMtJxPzw8XNu3b9cXX3xRbF03b950XENXEnmeAACPcelnW28HYmdmZurvf/+7Hn744WLHkOcJAPAWl5qntwKxu3fvrqCgILVq1Uq9evXSrFmzih1LnicAwFtcap7eCsReuXKl0tPT9Ze//EUbN27U/Pnzix1LnicAwFs8cqqKuwKxw8LC1K5dOw0cOFDz5s3TzJkzHQ0aAICK4tLRthURiJ2fn6/c3Fzl5+cX+z4AAHiDS83T04HYH3zwgfz9/RUdHa3AwEB98803eu211zRgwAD5+/uXZ30BACg3l5qnpwOx/fz89NZbb+nbb7+VZVkKDw/X2LFjNWnSJFfKBQDArWxWUWnUJYiNjVXHjh319ttve6omt7hy5YpCQkJ0+fJlDh4CgHuYJ/oB17YFAMAQzRMAAEPG+zwrWyA2eZ7wJWRdAlUDW54AABiieQIAYMgn8zxPnTolm81W6LZ79253lwsAgDG3N8+CPM9Dhw7pyy+/VEREhBISEnT+/PlCYwvyPIuzZcsWZWdnO25dunRxd7kAABjzyTzPAg0aNJDdbjctEQAAj/LpPM++ffuqcePG6tmzp9avX19iTTdv3tSVK1ecbgAAeIJP5nkGBwdrwYIFWr16tTZu3KiePXuqf//+JTZQwrABAN7ik3meDRs21OTJkxUTE6OuXbtq3rx5eu6555SUlFTsawjDBgB4i0/ned4pJiZGmZmZxT5PGDYAwFsqTZ7n/v37FRoa6kq5AAC4lU/meb733nsKCAhQp06dJElr167VkiVLijwoCQAAb/PJPE9Jmj17trKysuTn56c2bdpo5cqVeuqpp1wpFwAAtyLPEwBQpZHnCQCAD6B5AgBgiDxPAFUGeanwFrY8AQAwRPMEAMCQT+Z5zpw5s8g8z1q1arm7XAAAjPlknueUKVOccjyzs7PVrl07x4UWAACoSMbNMzY2VmPHjtXYsWMVEhKihg0bavr06So4XXTQoEGKj49XZGSk7r//fiUnJ+vKlSs6ePCg03IK8jznz59f6D2Cg4Nlt9sdt3Pnzuno0aMaPny4i6sJAID7+HSeZ4E//elPat26tXr16lXsGPI8AQDe4pN5nne6ceOGPvjgg1K3OsnzBAB4i0/med7po48+0tWrV0uMK5PI8wQAeI/P53n+6U9/0s9//nM1adKkxPckzxMA4C0+ned58uRJbdu2TevXr3elTAAAPMIn8zwLLFmyRKGhoerTp48rZQIA4BE+m+eZn5+vZcuWaejQocVu0QIAUBHI8wQAVGnkeQIA4ANongAAGKryeZ4AALibSwcMVSaEYaOyI+AZ8D38bAsAgCG3Ns+yZHlGREQUyumcN2+e4/kbN25o6NChio6Olp+fn/r37+/OEgEAKDe3Ns+yZnnOmjXLKatz3Lhxjufy8vJUo0YNjR8/XvHx8e4sDwAAtzDa5xkbG6sHHnhAkvTnP/9Z/v7+Gj16tGbNmiWbzaZBgwY5jU9OTlZqaqoOHjyouLg4x+O1a9eW3W4v8j1q1aqlxYsXS5J27typS5cuGa0QAACeZrzlWd4sT0maN2+eGjRooE6dOikpKUm3b992fQ3+P/I8AQDeYny0bUGWp81mU1RUlA4dOqSUlBSNHDlS0k9Zns8884yuX7+u0NBQpyxPSRo/frw6d+6s+vXra9euXXrttdeUnZ2t5OTkcq3I3Llz9cYbb5RrGQAAlIXxlmd5sjwlafLkyYqNjVX79u310ksvacGCBVq4cKEjdcVV5HkCALzF7aeqlJTlWZSYmBjdvn1bp06dKtf7kucJAPAW4+ZZnizPouzfv1/VqlVT48aNTUsBAKBCGO/zLE+WZ1pamvbs2aNHHnlEtWvXVlpamiZNmqTnnntO9erVc7zH0aNHdevWLV28eFFXr17V/v37JUkdO3Z002oDAOA64+ZZnizPwMBAffjhh5o5c6Zu3rypli1batKkSZo8ebLTezz22GPKyspy3O/UqZMkyTA9DQAAjzBunv7+/nr77bcd52IWCAoK0tq1a0t8befOnbV79+5S36O8+z8BAPCkKn9h+MNvJHLwEADArbgwPAAAhoy2PMnyBADgHvjZljxPlAdZmgCKws+2AAAY8nqepyRt3LhRMTExqlGjhurVq1cos/P06dN6/PHHVbNmTTVu3FivvPKKWy4eDwCAO7j1Z9uCPM/IyEjl5OQoJSVFCQkJyszMVKNGjSRJa9as0ciRIzVnzhz17t1bt2/f1uHDhx3LyMvL0+OPPy673a5du3YpOztbQ4YMkb+/v+bMmePOcgEAcInNMrjyQGl5nne7cuWKQkJCtGXLFsXFxen27duKiIjQG2+8oeHDhxf5Hn/729/085//XGfOnFGTJk0kSb///e81depUnT9/XgEBAWWqteC9wyauYp8nXMY+T6DyK+gHly9fdtupi17N80xPT9f//d//qVq1aurUqZNCQ0PVp08fpy3PtLQ0RUdHOxqnJCUmJurKlSs6cuRIsXWR5wkA8Bbj5lmQ5xkVFaVnn31W48aNU0pKiuP5DRs2KDg4WEFBQUpJSXHK8/zHP/4h6ad9o9OmTdOGDRtUr149xcbG6uLFi5Kks2fPOjVOSY77Z8+eLbauuXPnKiQkxHELCwszXTUAAMrEq3me+fn5kqTf/OY3evLJJ9WlSxctXbpUNptNq1evLteKkOcJAPAWr+Z5hoaGSpLatWvnGB8YGKjIyEidPn1akmS323Xu3DmnZRbct9vtxb4veZ4AAG/xap5nly5dFBgYqIyMDMfzubm5OnXqlMLDwyX9tCV76NAhx9aqJG3evFl16tRxaroAAFQU4+ZZkOeZkZGhFStWaOHChZowYYKuXbumX//619q9e7eysrK0d+9eDRs2zCnPs06dOnrppZc0Y8YMffbZZ8rIyNDo0aMlyTEmISFB7dq10+DBg3XgwAFt2rRJ06ZN05gxYxQYGOjGVQcAwDVezfOUpKSkJPn5+Wnw4MHKyclRTEyMtm7d6gjDrl69ujZs2KDRo0erW7duqlWrlp5//nnNmjXLfWsNAEA5GJ/n2bFjR7399tuerMktOM8T7sB5nkDl54nzPKv8heHJ8wQAuBsXhgcAwBB5ngAAGKryP9uS53lvYR8lAG/gZ1sAAAzRPAEAMOSTYdjjx493XI2oY8eO7iwRAIByc2vzLAjDPnTokL788ktFREQoISFB58+fd4xZs2aNBg8erBdeeEEHDhzQzp07NWjQoELLGjZsmAYMGODO8gAAcAufC8O+08yZM/Xxxx9r//79Bqvk/N5cJOHewgFDAO52T4Rhu4owbACAt/hcGLarCMMGAHgLYdgAABjyuTBsVxGGDQDwFp8LwwYAwNcZX56vIAz7xRdfVHp6uhYuXKgFCxbo2rVrevPNN9W3b1+Fhobqhx9+0KJFi4oNww4LC1N4eLiSkpIk/SsMW5IyMzP1448/6uzZs8rJyXEcbduuXTsFBAS4Y70BAHCZz4VhS9KIESO0Y8cOx/1OnTpJkk6ePKmIiIhyrC4AAOVX5cOw3XleDwCg8vGJ8zwBALjX0TwBADBU5cOwyfNEVcMlCIGKx5YnAACGaJ4AABjyuTzPAwcOaODAgQoLC1ONGjXUtm1bvfPOO+4sEwCAcjE+z7MkBXmekZGRysnJUUpKihISEpSZmalGjRpJ+inPc+TIkZozZ4569+6t27dvO6Wq7N27V40bN9by5csVFhamXbt2adSoUapevbrGjh3rznIBAHCJT+d5FhgzZoyOHTumrVu3lvk15HmiquKAIcCMT5znWRF5npcvX1b9+vVLHEOeJwDAW3w+z3PXrl1auXKlRo0aVWJd5HkCALzFp/M8Dx8+rH79+mnGjBlKSEgosS7yPAEA3uKzeZ5Hjx5VXFycRo0apWnTppX6vuR5AgC8xSfzPI8cOaJHHnlEzz//vN58803TEgEA8Cify/M8fPiwevfurcTERE2ePFlnz56VJFWvXt1xugsAABXJ5/I8//rXv+r8+fNavny5li9f7nhdeHi4Tp06Vf41BgCgnMjzBABUaT5xnicAAPc6micAAIaqfJ4nAADu5tYLw/siwrDvbVwHFoAn8LMtAACGfC7P88KFC3r00UfVtGlTBQYGKiwsTGPHjuVC7wAAn+HW5lmQ53no0CF9+eWXioiIUEJCgs6fP+8Ys2bNGg0ePFgvvPCCDhw4oJ07d2rQoEH/KqhaNfXr10/r16/Xt99+q2XLlmnLli166aWX3FkqAAAuqxR5nu+++66SkpKMLvZOnick9nkC8JHzPL2d53nmzBmtXbtWDz/8cIl1kecJAPAWn83zHDhwoGrWrKlmzZqpTp06RTboO5HnCQDwFp/N80xJSVF6errWrVunEydOaPLkySXWRZ4nAMBb3H6eZ0GeZ0GmZ6tWrZSamqrXXnvNKM/TbrfLbrerTZs2ql+/vnr16qXp06c7lnG3wMBABQYGunt1AAAoxCfzPItahiTHcgAAqEg+l+f56aef6ty5c+ratauCg4N15MgRvfLKK+rRo4ciIiLct+YAALjI5/I8a9SooT/+8Y+aNGmSbt68qbCwMD3xxBN69dVX3bfWAACUQ5XP8+Q8z3sb53kC8MR5nlX+wvCH30gkDBsA4FZcGB4AAEPkeQIAYKjK/2xLnifchf2nAArwsy0AAIZ8Ls/zThcuXFDz5s1ls9l06dIld5YKAIDLfC7P807Dhw9X+/bt3VkiAADl5rN5nosXL9bKlSv1+uuvKy4uTv/85z9Vt27dMq8Y53nC3djnCVRO90ye59GjRzVr1iy9//77qlatbCWS5wkA8Bafy/O8efOmBg4cqKSkJLVo0aLMdZHnCQDwFp/L83zttdfUtm1bPffcc0Z1kecJAPAWt5+qUpDn+W//9m9KTU2Vn5+fUlNTJalMeZ5bt27V6tWr5efnJz8/P8XFxUmSGjZsqBkzZhT7voGBgapTp47TDQAATzC+SIK78jx79uwpqXCe55o1a5STk+N4/ddff61hw4bpiy++0H333WdaLgAAbudzeZ53N8gffvhBktS2bVujo20BAPAUn8vzBADA15HnCZQR53kClRN5ni4gzxMA4G5cGB4AAEPkeQIAYKjK/2xLnifgXewbxr2An20BADBE8wQAwJDbm2dpgdjbt2+XzWYr8vb1118XWl5mZqZq167NBRIAAD7D7c2ztEDs7t27Kzs72+k2YsQItWzZUg8++KDTsnJzczVw4ED16tXL3WUCAOAy4wOGSgvEHjRokNP45ORkpaam6uDBg4qLi1NAQIDsdrvj+dzcXK1bt07jxo0rFKg9bdo0tWnTRnFxcdq1a1eJdd28edNxDV1J5HkCADzGpS3P8gRi3239+vW6cOGCXnjhBafHC9JVFi1aVKaayPMEAHiLS82zPIHYd0tNTVViYqKaN2/ueOzChdvKr+4AABSjSURBVAsaOnSoli1bVuarA5HnCQDwFpeaZ3kCse/03XffadOmTRo+fLjT4yNHjtSgQYP0s5/9rMw1kecJAPAWj5yqUlIg9p2WLl2qBg0aqG/fvk6Pb926VfPnz3cEYg8fPlyXL1+Wn5+flixZ4omSAQAoM5euMFSeQOwClmVp6dKlGjJkiPz9/Z2eS0tLc2zFStK6dev01ltvadeuXWrWrJkrJQMA4DYuNc/yBGIX2Lp1q06ePKkRI0YUWn7btm2d7n/zzTeqVq2a4yhfAAAqkkvNs7yB2NJPBwp1795dbdq0ccuKAADgLUZh2FLlCcT2RPgpAKDy8UQ/4Nq2AAAYonkCAGDIeJ9nZQvEJs8TnkBmJXBvY8sTAABDNE8AAAy5tXm6K8vTsizNnz9frVu3VmBgoJo1a6Y333zTnaUCAOAyl87zLE5BlmdkZKRycnKUkpKihIQEZWZmqlGjRo4szztNnz5dn3/+uVOW54QJE/TZZ59p/vz5io6O1sWLF3Xx4kV3lgoAgMuMzvMsLcvzbgXn1mzZskVxcXGFns/NzVWzZs00btw4TZ8+XZJ07NgxtW/fXocPH1ZUVJSr6+V477CJqzhgCG7HAUNA5eET53l6Osvzk08+UWRkpDZs2KCWLVsqIiJCI0aMKHXL8+bNm7py5YrTDQAATzBunp7O8vzHP/6hrKwsrV69Wu+//76WLVumvXv36qmnniqxLsKwAQDeYtw8PZ3lWZDA8v7776tXr16KjY1Vamqqtm3bpoyMjGLrIgwbAOAtbj9VpbxZnqGhofLz81Pr1q0djxWkrJw+fbrY9yUMGwDgLcbN09NZnj169NDt27d14sQJx2PffvutJCk8PNy0XAAA3M64eRZkeWZkZGjFihVauHChJkyYoGvXrunXv/61du/eraysLO3du1fDhg0zzvKMj49X586dNWzYMO3bt0979+7Viy++qH//93932hoFAKCiGJ/n6eksz2rVqumTTz7RuHHj9LOf/Uy1atVSnz59tGDBAtfXEgAANzI+z7MyZHlK5HkCAH7iE+d5AgBwr6N5AgBgyGifZ2XL8gQAwBPY8gQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM0TwBADBE8wQAwBDNEwAAQzRPAAAM+VV0AZ5iWZYk6cqVKxVcCQCgIhX0gYK+4A5VtnleuHBBkhQWFlbBlQAAfMHVq1cVEhLilmVV2eZZv359SdLp06fdNlnedOXKFYWFhel///d/VadOnYouxxj1V5zKXLtE/RWtKtZvWZauXr2qpk2buu19qmzzrFbtp925ISEhlfIDUKBOnTrUX4Eqc/2VuXaJ+itaVavf3RtRHDAEAIAhmicAAIaqz5w5c2ZFF+Ep1atXV2xsrPz8Kuev09RfsSpz/ZW5don6Kxr1l85mufPYXQAA7gH8bAsAgCGaJwAAhmieAAAYonkCAGCoUjfPRYsWKSIiQkFBQYqJidFXX31V4vjVq1erTZs2CgoKUnR0tD799FMvVeps7ty56tq1q2rXrq3GjRurf//+ysjIKPE1y5Ytk81mc7oFBQV5qWJnM2fOLFRLmzZtSnyNr8y9JEVERBSq32azacyYMUWOr+i5/5//+R/94he/UNOmTWWz2fTxxx87PW9Zll5//XWFhoaqRo0aio+P1/Hjx0tdrun3x9215+bmaurUqYqOjlatWrXUtGlTDRkyRGfOnClxma58/jxRvyQNHTq0UC2PPvpoqcv1xtyXpf6ivgc2m01JSUnFLtNb81+Wv5M3btzQmDFj1KBBAwUHB+vJJ5/UuXPnSlyuq9+Xu1Xa5rly5UpNnjxZM2bMUHp6ujp06KDExER9//33RY7ftWuXBg4cqOHDh2vfvn3q37+/+vfvr8OHD3u5cmnHjh0aM2aMdu/erc2bNys3N1cJCQm6du1aia+rU6eOsrOzHbesrCwvVVzY/fff71TLl19+WexYX5p7Sfr666+dat+8ebMk6T/+4z+KfU1Fzv21a9fUoUMHLVq0qMjnf/vb3+rdd9/V73//e+3Zs0e1atVSYmKibty4UewyTb8/nqj9+vXrSk9P1/Tp05Wenq61a9cqIyNDffv2LXW5Jp+/8iht7iXp0UcfdaplxYoVJS7TW3MvlV7/nXVnZ2dryZIlstlsevLJJ0tcrjfmvyx/JydNmqRPPvlEq1ev1o4dO3TmzBk98cQTJS7Xle9LkaxK6qGHHrLGjBnjuJ+Xl2c1bdrUmjt3bpHjn376aevxxx93eiwmJsZ68cUXPVpnWXz//feWJGvHjh3Fjlm6dKkVEhLixaqKN2PGDKtDhw5lHu/Lc29ZljVhwgTrvvvus/Lz84t83pfmXpL10UcfOe7n5+dbdrvdSkpKcjx26dIlKzAw0FqxYkWxyzH9/rjD3bUX5auvvrIkWVlZWcWOMf38uUtR9T///PNWv379jJZTEXNvWWWb/379+lm9e/cucUxFzf/dfycvXbpk+fv7W6tXr3aMOXbsmCXJSktLK3IZrn5filIptzxv3bqlvXv3Kj4+3vFYtWrVFB8fr7S0tCJfk5aW5jRekhITE4sd702XL1+W9K+L2Rfnxx9/VHh4uMLCwtSvXz8dOXLEG+UV6fjx42ratKkiIyP17LPP6vTp08WO9eW5v3XrlpYvX65hw4bJZrMVO86X5v5OJ0+e1NmzZ53mNyQkRDExMcXOryvfH2+5fPmybDab6tatW+I4k8+fp23fvl2NGzdWVFSURo8e7Uh0Koovz/25c+e0ceNGDR8+vNSxFTH/d/+d3Lt3r3Jzc53msk2bNmrRokWxc+nK96U4lbJ5/vDDD8rLy1OTJk2cHm/SpInOnj1b5GvOnj1rNN5b8vPzNXHiRPXo0UMPPPBAseOioqK0ZMkSrVu3TsuXL1d+fr66d++u7777zovV/iQmJkbLli3T3//+dy1evFgnT55Ur169dPXq1SLH++rcS9LHH3+sS5cuaejQocWO8aW5v1vBHJrMryvfH2+4ceOGpk6dqoEDB5Z4QXLTz58nPfroo3r//ff1+eef66233tKOHTvUp08f5eXlFTneV+dekt577z3Vrl271J89K2L+i/o7efbsWQUEBBT6j1ZpfaBgTFlfU5zKee2lKmTMmDE6fPhwqfsMunXrpm7dujnud+/eXW3bttUf/vAHzZ4929NlOunTp4/j3+3bt1dMTIzCw8O1atWqMv2v1ZekpqaqT58+JUYV+dLcV1W5ubl6+umnZVmWFi9eXOJYX/r8PfPMM45/R0dHq3379rrvvvu0fft2xcXFebWW8lqyZImeffbZUg+Gq4j5L+vfSW+qlFueDRs2VPXq1QsdVXXu3DnZ7fYiX2O3243Ge8PYsWO1YcMGbdu2Tc2bNzd6rb+/vzp16qTMzEwPVVd2devWVevWrYutxRfnXpKysrK0ZcsWjRgxwuh1vjT3BXNoMr+ufH88qaBxZmVlafPmzcYxWKV9/rwpMjJSDRs2LLYWX5v7Al988YUyMjKMvwuS5+e/uL+Tdrtdt27d0qVLl5zGl9YHCsaU9TXFqZTNMyAgQF26dNHnn3/ueCw/P1+ff/650xbCnbp16+Y0XpI2b95c7HhPsixLY8eO1UcffaStW7eqZcuWxsvIy8vToUOHFBoa6oEKzfz44486ceJEsbX40tzfaenSpWrcuLEef/xxo9f50ty3bNlSdrvdaX6vXLmiPXv2FDu/rnx/PKWgcR4/flxbtmxRgwYNjJdR2ufPm7777jtduHCh2Fp8ae7vlJqaqi5duqhDhw7Gr/XU/Jf2d7JLly7y9/d3msuMjAydPn262Ll05ftSUoGV0ocffmgFBgZay5Yts44ePWqNGjXKqlu3rnX27FnLsixr8ODB1quvvuoYv3PnTsvPz8+aP3++dezYMWvGjBmWv7+/dejQIa/XPnr0aCskJMTavn27lZ2d7bhdv37dMebu+t944w1r06ZN1okTJ6y9e/dazzzzjBUUFGQdOXLE6/X/6le/srZv326dPHnS2rlzpxUfH281bNjQ+v7774us3ZfmvkBeXp7VokULa+rUqYWe87W5v3r1qrVv3z5r3759liQrOTnZ2rdvn+OI1Hnz5ll169a11q1bZx08eNDq16+f1bJlSysnJ8exjN69e1sLFy503C/t++ON2m/dumX17dvXat68ubV//36n78LNmzeLrb20z5+36r969ao1ZcoUKy0tzTp58qS1ZcsWq3PnzlarVq2sGzduFFu/t+a+tPoLXL582apZs6a1ePHiIpdRUfNflr+TL730ktWiRQtr69at1jfffGN169bN6tatm9NyoqKirLVr1zrul+X7UhaVtnlalmUtXLjQatGihRUQEGA99NBD1u7dux3PPfzww9bzzz/vNH7VqlVW69atrYCAAOv++++3Nm7c6OWKfyKpyNvSpUsdY+6uf+LEiY51bdKkifXYY49Z6enp3i/esqwBAwZYoaGhVkBAgNWsWTNrwIABVmZmpuN5X577Aps2bbIkWRkZGYWe87W537ZtW5Gfl4Ia8/PzrenTp1tNmjSxAgMDrbi4uELrFR4ebs2YMcPpsZK+P96o/eTJk8V+F7Zt21Zs7aV9/rxV//Xr162EhASrUaNGlr+/vxUeHm6NHDmyUBOsqLkvrf4Cf/jDH6waNWpYly5dKnIZFTX/Zfk7mZOTY7388stWvXr1rJo1a1q//OUvrezs7ELLufM1Zfm+lAWRZAAAGKqU+zwBAKhINE8AAAzRPAEAMETzBADAEM0TAABDNE8AAAzRPAEAMETzBADAEM0TqCCxsbGaOHFihbz39u3bZbPZCl1UG0DZ0DyBKq6oJt29e3dlZ2crJCSkgqoCKjeaJ1BJ5ebmuvzagIAA2e122Ww2N1YE3DtonoAXXLt2TUOGDFFwcLBCQ0O1YMECp+dtNps+/vhjp8fq1q2rZcuWSZJOnTolm82mlStX6uGHH1ZQUJA++OADXbhwQQMHDlSzZs1Us2ZNRUdHa8WKFY5lDB06VDt27NA777wjm80mm82mU6dOFfmz7Zo1a3T//fcrMDBQERERhWqMiIjQnDlzNGzYMNWuXVstWrTQf//3fzuev3XrlsaOHavQ0FAFBQUpPDxcc+fOddcUAj6F5gl4wSuvvKIdO3Zo3bp1+uyzz7R9+3alp6cbL+fVV1/VhAkTdOzYMSUmJurGjRvq0qWLNm7cqMOHD2vUqFEaPHiwvvrqK0nSO++8o27dumnkyJHKzs5Wdna2wsLCCi137969evrpp/XMM8/o0KFDmjlzpqZPn+5o3gUWLFigBx98UPv27dPLL7+s0aNHKyMjQ5L07rvvav369Vq1apUyMjL0wQcfKCIiwngdgcrAr6ILAKq6H3/8UampqVq+fLni4uIkSe+9956aN29uvKyJEyfqiSeecHpsypQpjn+PGzdOmzZt0qpVq/TQQw8pJCREAQEBqlmzpux2e7HLTU5OVlxcnKZPny5Jat26tY4ePaqkpCQNHTrUMe6xxx7Tyy+/LEmaOnWqUlJStG3bNkVFRen06dNq1aqVevbsKZvNpvDwcOP1AyoLtjwBDztx4oRu3bqlmJgYx2P169dXVFSU8bIefPBBp/t5eXmaPXu2oqOjVb9+fQUHB2vTpk06ffq00XKPHTumHj16OD3Wo0cPHT9+XHl5eY7H2rdv7/i3zWaT3W7X999/L+mnn4j379+vqKgojR8/Xp999pnp6gGVBs0T8AE2m013R+sWdUBQrVq1nO4nJSXpnXfe0dSpU7Vt2zbt379fiYmJunXrlkfq9Pf3d7pvs9mUn58vSercubNOnjyp2bNnKycnR08//bSeeuopj9QBVDSaJ+Bh9913n/z9/bVnzx7HY//85z/17bffOu43atRI2dnZjvvHjx/X9evXS132zp071a9fPz333HPq0KGDIiMjnZYr/XRk7Z1bj0Vp27atdu7cWWjZrVu3VvXq1Uuto0CdOnU0YMAA/fGPf9TKlSu1Zs0aXbx4scyvByoL9nkCHhYcHKzhw4frlVdeUYMGDdS4cWP95je/UbVq//q/a+/evfW73/1O3bp1U15enqZOnVpoK68orVq10l//+lft2rVL9erVU3Jyss6dO6d27do5xkRERGjPnj06deqUgoODVb9+/ULL+dWvfqWuXbtq9uzZGjBggNLS0vS73/1O//Vf/1Xm9UxOTlZoaKg6deqkatWqafXq1bLb7apbt26ZlwFUFmx5Al6QlJSkXr166Re/+IXi4+PVs2dPdenSxfH8ggULFBYWpl69emnQoEGaMmWKatasWepyp02bps6dOysxMVGxsbGy2+3q37+/05gpU6aoevXqateunRo1alTk/tDOnTtr1apV+vDDD/XAAw/o9ddf16xZs5wOFipN7dq19dvf/lYPPvigunbtqlOnTunTTz91+k8CUFXYrLt3tAAAgBLxX0IAAAzRPAEAMETzBADAEM0TAABDNE8AAAzRPAEAMETzBADAEM0TAABDNE8AAAzRPAEAMETzBADA0P8DI8lgq7Q/IEMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 500x2500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from synthesizer.hparams import hparams\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "spk2utts = {}\n",
    "with open('datasets/vctk/synthesizer/train.txt') as metadata_file:\n",
    "    for line in metadata_file:\n",
    "        items = line.strip().split(\"|\")\n",
    "        spk = items[0].replace('audio-', '').split('_')[0]\n",
    "        if spk not in spk2utts.keys():\n",
    "            spk2utts[spk] = []\n",
    "        spk2utts[spk].append(int(items[3]))\n",
    "\n",
    "spks = list(spk2utts.keys())\n",
    "len_in_min = [sum(spk2utts[i]) / hparams.sample_rate / 60 for i in spks]\n",
    "print('%d speakers are used for training and testing' % len(spks))\n",
    "plt.rcdefaults()\n",
    "fig, ax = plt.subplots(figsize=(5, 25))\n",
    "\n",
    "y_pos = np.arange(len(spks))\n",
    "\n",
    "\n",
    "ax.barh(y_pos, len_in_min, align='center')\n",
    "ax.set_yticks(y_pos)\n",
    "ax.set_yticklabels(spks)\n",
    "ax.invert_yaxis() \n",
    "ax.set_xlabel('durations')\n",
    "ax.set_title('The duration of each speaker from VCTK set')\n",
    "\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# split the data into training testing and validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "! mv datasets/vctk/synthesizer/train.txt datasets/vctk/synthesizer/all.txt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "utts = [line.strip() for line in open('datasets/vctk/synthesizer/all.txt', 'r').readlines()]\n",
    "spk_dict = {}\n",
    "\n",
    "for u in utts:\n",
    "    spk = u.split('-')[1].split('_')[0]\n",
    "    assert spk[0] == 'p'\n",
    "    if spk not in spk_dict.keys():\n",
    "        spk_dict[spk] = []\n",
    "    spk_dict[spk].append(u)\n",
    "\n",
    "spk2num = [(spk, len(spk_dict[spk])) for spk in spk_dict.keys()]\n",
    "spk2num.sort(key=lambda x:x[1])\n",
    "test_set = spk2num[:8]\n",
    "training_set = spk2num[8:]\n",
    "\n",
    "train_wf = open('datasets/vctk/synthesizer/train.txt', 'w')\n",
    "test_wf = open('datasets/vctk/synthesizer/test.txt', 'w')\n",
    "val_wf = open('datasets/vctk/synthesizer/val.txt', 'w')\n",
    "\n",
    "for spk, _ in training_set:\n",
    "    for utt in spk_dict[spk][:8]:\n",
    "        val_wf.write(utt + '\\n')\n",
    "    for utt in spk_dict[spk][8:]:\n",
    "        train_wf.write(utt + '\\n')\n",
    "        \n",
    "for spk, _ in test_set:\n",
    "    for utt in spk_dict[spk]:\n",
    "        test_wf.write(utt + '\\n')\n",
    "        \n",
    "train_wf.close()\n",
    "test_wf.close()\n",
    "val_wf.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Next step, speaker embedding extraction, use deep_speaker/get_gvector.ipynb to obtain speaker embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (rtvc)",
   "language": "python",
   "name": "rtvc"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}