{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import librosa\n",
    "import io\n",
    "import soundfile as sf\n",
    "from moviepy.editor import VideoFileClip\n",
    "from tqdm import tqdm\n",
    "import pickle as pk\n",
    "import os\n",
    "import tensorflow as tf\n",
    "# from tensorflow.keras.saving import register_keras_serializable\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "from tensorflow.keras import layers, models\n",
    "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# real_audio_dir = (\n",
    "#     r\"H:\\.shortcut-targets-by-id\\1jH_pc6mMj0Iu8wLS1r0vggMWpVElJvOU\\SIH2024_DATASET\\REAL\"\n",
    "# )\n",
    "# fake_audio_dir = (\n",
    "#     r\"H:\\.shortcut-targets-by-id\\1jH_pc6mMj0Iu8wLS1r0vggMWpVElJvOU\\SIH2024_DATASET\\FAKE\"\n",
    "# )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# with open(\n",
    "#     r\"H:\\.shortcut-targets-by-id\\1jH_pc6mMj0Iu8wLS1r0vggMWpVElJvOU\\SIH2024_DATASET\\real_files.pkl\",\n",
    "#     \"rb\",\n",
    "# ) as f:\n",
    "#     real_files = pk.load(f)\n",
    "\n",
    "# with open(\n",
    "#     r\"H:\\.shortcut-targets-by-id\\1jH_pc6mMj0Iu8wLS1r0vggMWpVElJvOU\\SIH2024_DATASET\\fake_files.pkl\",\n",
    "#     \"rb\",\n",
    "# ) as f:\n",
    "#     fake_files = pk.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# len(real_files), len(fake_files)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# real_files = real_files[:2000]\n",
    "# fake_files = fake_files[:2000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# fake_files = fake_files[: len(real_files)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# len(real_files), len(fake_files)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def extract_features(file_path):\n",
    "#     try:\n",
    "#         # Load the video file\n",
    "#         video_clip = VideoFileClip(file_path)\n",
    "#         audio = video_clip.audio\n",
    "#         fps = audio.fps\n",
    "#         audio_samples = np.array(\n",
    "#             list(audio.iter_frames(fps=fps, dtype=\"float32\"))\n",
    "#         ).flatten()\n",
    "#         buffer = io.BytesIO()\n",
    "#         sf.write(buffer, audio_samples, fps, format=\"wav\")\n",
    "#         buffer.seek(0)\n",
    "#         x, sr = librosa.load(buffer, sr=None)\n",
    "#         mfccs = librosa.feature.mfcc(y=x, sr=sr, n_mfcc=20)\n",
    "\n",
    "#         return mfccs\n",
    "\n",
    "#     except Exception as e:\n",
    "#         print(f\"Error encountered while parsing file: {file_path}, {e}\")\n",
    "#         return None\n",
    "\n",
    "\n",
    "# def load_data(real_dir, fake_dir):\n",
    "#     labels = []\n",
    "#     features = []\n",
    "\n",
    "#     # Load real audios\n",
    "#     for file_name in real_files:\n",
    "#         file_path = os.path.join(real_dir, file_name)\n",
    "#         mfccs = extract_features(file_path)\n",
    "#         if mfccs is not None:\n",
    "#             features.append(mfccs)\n",
    "#             labels.append(0)  # 0 for REAL\n",
    "\n",
    "#     # Load fake audios\n",
    "#     for file_name in fake_files:\n",
    "#         file_path = os.path.join(fake_dir, file_name)\n",
    "#         mfccs = extract_features(file_path)\n",
    "#         if mfccs is not None:\n",
    "#             features.append(mfccs)\n",
    "#             labels.append(1)  # 1 for FAKE\n",
    "\n",
    "#     return np.array(features), np.array(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def extract_frame_features(file_path, frame_duration=1.0):\n",
    "#     try:\n",
    "#         video_clip = VideoFileClip(file_path)\n",
    "#         audio = video_clip.audio\n",
    "#         fps = audio.fps\n",
    "#         audio_samples = np.array(\n",
    "#             list(audio.iter_frames(fps=fps, dtype=\"float32\"))\n",
    "#         ).flatten()\n",
    "#         buffer = io.BytesIO()\n",
    "#         sf.write(buffer, audio_samples, fps, format=\"wav\")\n",
    "#         buffer.seek(0)\n",
    "#         x, sr = librosa.load(buffer, sr=None)\n",
    "\n",
    "#         # Split audio into frames of 'frame_duration' seconds\n",
    "#         frame_length = int(frame_duration * sr)\n",
    "#         frames = [\n",
    "#             librosa.feature.mfcc(y=x[i : i + frame_length], sr=sr, n_mfcc=20)\n",
    "#             for i in range(0, len(x), frame_length)\n",
    "#             if i + frame_length <= len(x)\n",
    "#         ]\n",
    "\n",
    "#         return frames  # Returns list of MFCCs for each frame\n",
    "\n",
    "#     except Exception as e:\n",
    "#         print(f\"Error processing file {file_path}: {e}\")\n",
    "#         return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_frame_features(file_path, frame_duration=1.0):\n",
    "    video_clip = VideoFileClip(file_path)\n",
    "    audio = video_clip.audio\n",
    "    fps = audio.fps\n",
    "    audio_samples = np.array(\n",
    "        list(audio.iter_frames(fps=fps, dtype=\"float32\"))\n",
    "    ).flatten()\n",
    "    buffer = io.BytesIO()\n",
    "    sf.write(buffer, audio_samples, fps, format=\"wav\")\n",
    "    buffer.seek(0)\n",
    "    x, sr = librosa.load(buffer, sr=None)\n",
    "\n",
    "    # Split audio into frames of 'frame_duration' seconds\n",
    "    frame_length = int(frame_duration * sr)\n",
    "    frames = []\n",
    "    timestamps = []\n",
    "\n",
    "    for i in range(0, len(x), frame_length):\n",
    "        if i + frame_length <= len(x):\n",
    "            # Extract MFCCs for each frame and store the timestamp\n",
    "            frame_mfcc = librosa.feature.mfcc(y=x[i: i + frame_length], sr=sr, n_mfcc=20)\n",
    "            frames.append(frame_mfcc)\n",
    "            timestamp = i / sr  # Convert index to seconds\n",
    "            timestamps.append(timestamp)\n",
    "\n",
    "    return frames, timestamps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def load_data(real_dir, fake_dir, real_files, fake_files):\n",
    "#     labels, features = [], []\n",
    "\n",
    "#     # Load real audio frames with progress bar\n",
    "#     print(\"Loading real audio files:\")\n",
    "#     for file_name in tqdm(real_files, desc=\"Processing Real Files\"):\n",
    "#         file_path = os.path.join(real_dir, file_name)\n",
    "#         frame_features, timestamps = extract_frame_features(file_path)\n",
    "#         if frame_features:\n",
    "#             features.extend(frame_features)\n",
    "#             labels.extend([0] * len(frame_features))  # Label 0 for REAL\n",
    "\n",
    "#     # Load fake audio frames with progress bar\n",
    "#     print(\"Loading fake audio files:\")\n",
    "#     for file_name in tqdm(fake_files, desc=\"Processing Fake Files\"):\n",
    "#         file_path = os.path.join(fake_dir, file_name)\n",
    "#         frame_features = extract_frame_features(file_path)\n",
    "#         if frame_features:\n",
    "#             features.extend(frame_features)\n",
    "#             labels.extend([1] * len(frame_features))  # Label 1 for FAKE\n",
    "\n",
    "#     # Convert to numpy arrays\n",
    "#     features = np.array(features)\n",
    "#     labels = np.array(labels)\n",
    "\n",
    "#     # Shuffle the data\n",
    "#     indices = np.arange(len(features))\n",
    "#     np.random.shuffle(indices)\n",
    "#     features = features[indices]\n",
    "#     labels = labels[indices]\n",
    "\n",
    "#     return features, labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# X, y = load_data(real_audio_dir, fake_audio_dir, real_files, fake_files)\n",
    "# X = X[..., np.newaxis]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# with open(\"X_for_dl_2000.pkl\", \"wb\") as f:\n",
    "#     pk.dump(X, f)\n",
    "# with open(\"y_for_dl_2000.pkl\", \"wb\") as f:\n",
    "#     pk.dump(y, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"X_for_dl_2000.pkl\", \"rb\") as f:\n",
    "    X = pk.load(f)\n",
    "with open(\"y_for_dl_2000.pkl\", \"rb\") as f:\n",
    "    y = pk.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y, test_size=0.2, random_state=30\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TCN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model = models.Sequential(\n",
    "#     [\n",
    "#         layers.Conv1D(\n",
    "#             64,\n",
    "#             kernel_size=3,\n",
    "#             dilation_rate=1,\n",
    "#             padding=\"causal\",\n",
    "#             activation=\"relu\",\n",
    "#             input_shape=(X.shape[1], X.shape[2]),\n",
    "#         ),\n",
    "#         layers.Conv1D(\n",
    "#             128, kernel_size=3, dilation_rate=2, padding=\"causal\", activation=\"relu\"\n",
    "#         ),\n",
    "#         layers.Conv1D(\n",
    "#             256, kernel_size=3, dilation_rate=4, padding=\"causal\", activation=\"relu\"\n",
    "#         ),\n",
    "#         layers.GlobalAveragePooling1D(),\n",
    "#         layers.Dropout(0.5),\n",
    "#         layers.Dense(64, activation=\"relu\"),\n",
    "#         layers.Dense(2, activation=\"softmax\"),\n",
    "#     ]\n",
    "# )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras import models, layers\n",
    "\n",
    "model = models.Sequential(\n",
    "    [\n",
    "        layers.Conv1D(\n",
    "            64,\n",
    "            kernel_size=3,\n",
    "            dilation_rate=1,\n",
    "            padding=\"causal\",\n",
    "            activation=\"relu\",\n",
    "            input_shape=(X.shape[1], X.shape[2]),\n",
    "        ),\n",
    "        layers.BatchNormalization(),\n",
    "        layers.Conv1D(\n",
    "            128, kernel_size=3, dilation_rate=2, padding=\"causal\", activation=\"relu\"\n",
    "        ),\n",
    "        layers.BatchNormalization(),\n",
    "        layers.Conv1D(\n",
    "            256, kernel_size=3, dilation_rate=4, padding=\"causal\", activation=\"relu\"\n",
    "        ),\n",
    "        layers.BatchNormalization(),\n",
    "        layers.GlobalAveragePooling1D(),\n",
    "        layers.Dropout(0.5),\n",
    "        layers.Dense(128, activation=\"relu\"),\n",
    "        layers.Dropout(0.3),\n",
    "        layers.Dense(2, activation=\"softmax\"),\n",
    "    ]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    optimizer=\"adam\", loss=\"sparse_categorical_crossentropy\", metrics=[\"accuracy\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "checkpoint = ModelCheckpoint(\n",
    "    \"model/best_model.keras\", monitor=\"val_loss\", save_best_only=True\n",
    ")\n",
    "early_stopping = EarlyStopping(monitor=\"val_loss\", patience=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\">Model: \"sequential\"</span>\n",
       "</pre>\n"
      ],
      "text/plain": [
       "\u001b[1mModel: \"sequential\"\u001b[0m\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\">┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
       "┃<span style=\"font-weight: bold\"> Layer (type)                    </span>┃<span style=\"font-weight: bold\"> Output Shape           </span>┃<span style=\"font-weight: bold\">       Param # </span>┃\n",
       "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
       "│ conv1d (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Conv1D</span>)                 │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">20</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>)         │        <span style=\"color: #00af00; text-decoration-color: #00af00\">16,768</span> │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ batch_normalization             │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">20</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>)         │           <span style=\"color: #00af00; text-decoration-color: #00af00\">256</span> │\n",
       "│ (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">BatchNormalization</span>)            │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ conv1d_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Conv1D</span>)               │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">20</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)        │        <span style=\"color: #00af00; text-decoration-color: #00af00\">24,704</span> │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ batch_normalization_1           │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">20</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)        │           <span style=\"color: #00af00; text-decoration-color: #00af00\">512</span> │\n",
       "│ (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">BatchNormalization</span>)            │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ conv1d_2 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Conv1D</span>)               │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">20</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">256</span>)        │        <span style=\"color: #00af00; text-decoration-color: #00af00\">98,560</span> │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ batch_normalization_2           │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">20</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">256</span>)        │         <span style=\"color: #00af00; text-decoration-color: #00af00\">1,024</span> │\n",
       "│ (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">BatchNormalization</span>)            │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ global_average_pooling1d        │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">256</span>)            │             <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
       "│ (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">GlobalAveragePooling1D</span>)        │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dropout (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dropout</span>)               │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">256</span>)            │             <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dense (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dense</span>)                   │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)            │        <span style=\"color: #00af00; text-decoration-color: #00af00\">32,896</span> │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dropout_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dropout</span>)             │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)            │             <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dense_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dense</span>)                 │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">2</span>)              │           <span style=\"color: #00af00; text-decoration-color: #00af00\">258</span> │\n",
       "└─────────────────────────────────┴────────────────────────┴───────────────┘\n",
       "</pre>\n"
      ],
      "text/plain": [
       "┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
       "┃\u001b[1m \u001b[0m\u001b[1mLayer (type)                   \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape          \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m      Param #\u001b[0m\u001b[1m \u001b[0m┃\n",
       "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
       "│ conv1d (\u001b[38;5;33mConv1D\u001b[0m)                 │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m20\u001b[0m, \u001b[38;5;34m64\u001b[0m)         │        \u001b[38;5;34m16,768\u001b[0m │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ batch_normalization             │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m20\u001b[0m, \u001b[38;5;34m64\u001b[0m)         │           \u001b[38;5;34m256\u001b[0m │\n",
       "│ (\u001b[38;5;33mBatchNormalization\u001b[0m)            │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ conv1d_1 (\u001b[38;5;33mConv1D\u001b[0m)               │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m20\u001b[0m, \u001b[38;5;34m128\u001b[0m)        │        \u001b[38;5;34m24,704\u001b[0m │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ batch_normalization_1           │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m20\u001b[0m, \u001b[38;5;34m128\u001b[0m)        │           \u001b[38;5;34m512\u001b[0m │\n",
       "│ (\u001b[38;5;33mBatchNormalization\u001b[0m)            │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ conv1d_2 (\u001b[38;5;33mConv1D\u001b[0m)               │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m20\u001b[0m, \u001b[38;5;34m256\u001b[0m)        │        \u001b[38;5;34m98,560\u001b[0m │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ batch_normalization_2           │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m20\u001b[0m, \u001b[38;5;34m256\u001b[0m)        │         \u001b[38;5;34m1,024\u001b[0m │\n",
       "│ (\u001b[38;5;33mBatchNormalization\u001b[0m)            │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ global_average_pooling1d        │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m256\u001b[0m)            │             \u001b[38;5;34m0\u001b[0m │\n",
       "│ (\u001b[38;5;33mGlobalAveragePooling1D\u001b[0m)        │                        │               │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dropout (\u001b[38;5;33mDropout\u001b[0m)               │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m256\u001b[0m)            │             \u001b[38;5;34m0\u001b[0m │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dense (\u001b[38;5;33mDense\u001b[0m)                   │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m)            │        \u001b[38;5;34m32,896\u001b[0m │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dropout_1 (\u001b[38;5;33mDropout\u001b[0m)             │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m)            │             \u001b[38;5;34m0\u001b[0m │\n",
       "├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
       "│ dense_1 (\u001b[38;5;33mDense\u001b[0m)                 │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m2\u001b[0m)              │           \u001b[38;5;34m258\u001b[0m │\n",
       "└─────────────────────────────────┴────────────────────────┴───────────────┘\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Total params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">174,978</span> (683.51 KB)\n",
       "</pre>\n"
      ],
      "text/plain": [
       "\u001b[1m Total params: \u001b[0m\u001b[38;5;34m174,978\u001b[0m (683.51 KB)\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Trainable params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">174,082</span> (680.01 KB)\n",
       "</pre>\n"
      ],
      "text/plain": [
       "\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m174,082\u001b[0m (680.01 KB)\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Non-trainable params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">896</span> (3.50 KB)\n",
       "</pre>\n"
      ],
      "text/plain": [
       "\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m896\u001b[0m (3.50 KB)\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 4ms/step - accuracy: 0.6385 - loss: 0.6350 - val_accuracy: 0.6710 - val_loss: 0.6011\n",
      "Epoch 2/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m17s\u001b[0m 4ms/step - accuracy: 0.6680 - loss: 0.6062 - val_accuracy: 0.6838 - val_loss: 0.5800\n",
      "Epoch 3/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m16s\u001b[0m 4ms/step - accuracy: 0.6856 - loss: 0.5882 - val_accuracy: 0.7069 - val_loss: 0.5591\n",
      "Epoch 4/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.6969 - loss: 0.5731 - val_accuracy: 0.7187 - val_loss: 0.5497\n",
      "Epoch 5/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7038 - loss: 0.5649 - val_accuracy: 0.7303 - val_loss: 0.5353\n",
      "Epoch 6/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7127 - loss: 0.5569 - val_accuracy: 0.7343 - val_loss: 0.5330\n",
      "Epoch 7/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7198 - loss: 0.5478 - val_accuracy: 0.7102 - val_loss: 0.5598\n",
      "Epoch 8/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7239 - loss: 0.5452 - val_accuracy: 0.7404 - val_loss: 0.5247\n",
      "Epoch 9/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7271 - loss: 0.5389 - val_accuracy: 0.7310 - val_loss: 0.5310\n",
      "Epoch 10/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7301 - loss: 0.5323 - val_accuracy: 0.7369 - val_loss: 0.5335\n",
      "Epoch 11/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7358 - loss: 0.5272 - val_accuracy: 0.7529 - val_loss: 0.5058\n",
      "Epoch 12/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7363 - loss: 0.5263 - val_accuracy: 0.7451 - val_loss: 0.5065\n",
      "Epoch 13/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 4ms/step - accuracy: 0.7379 - loss: 0.5212 - val_accuracy: 0.7451 - val_loss: 0.5055\n",
      "Epoch 14/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7454 - loss: 0.5105 - val_accuracy: 0.7447 - val_loss: 0.5048\n",
      "Epoch 15/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7478 - loss: 0.5100 - val_accuracy: 0.7554 - val_loss: 0.4946\n",
      "Epoch 16/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7442 - loss: 0.5087 - val_accuracy: 0.7533 - val_loss: 0.5004\n",
      "Epoch 17/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 4ms/step - accuracy: 0.7513 - loss: 0.5005 - val_accuracy: 0.7469 - val_loss: 0.5045\n",
      "Epoch 18/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 4ms/step - accuracy: 0.7507 - loss: 0.4992 - val_accuracy: 0.7519 - val_loss: 0.4980\n",
      "Epoch 19/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 4ms/step - accuracy: 0.7528 - loss: 0.4976 - val_accuracy: 0.7553 - val_loss: 0.4930\n",
      "Epoch 20/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7582 - loss: 0.4947 - val_accuracy: 0.7637 - val_loss: 0.4833\n",
      "Epoch 21/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7561 - loss: 0.4986 - val_accuracy: 0.7668 - val_loss: 0.4831\n",
      "Epoch 22/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7593 - loss: 0.4891 - val_accuracy: 0.7671 - val_loss: 0.4819\n",
      "Epoch 23/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 4ms/step - accuracy: 0.7578 - loss: 0.4900 - val_accuracy: 0.7671 - val_loss: 0.4808\n",
      "Epoch 24/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7628 - loss: 0.4851 - val_accuracy: 0.7586 - val_loss: 0.5014\n",
      "Epoch 25/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 4ms/step - accuracy: 0.7609 - loss: 0.4850 - val_accuracy: 0.7563 - val_loss: 0.4884\n",
      "Epoch 26/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7647 - loss: 0.4826 - val_accuracy: 0.7679 - val_loss: 0.4788\n",
      "Epoch 27/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7623 - loss: 0.4848 - val_accuracy: 0.7476 - val_loss: 0.5020\n",
      "Epoch 28/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7665 - loss: 0.4792 - val_accuracy: 0.7659 - val_loss: 0.4835\n",
      "Epoch 29/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7658 - loss: 0.4796 - val_accuracy: 0.7688 - val_loss: 0.4923\n",
      "Epoch 30/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7688 - loss: 0.4759 - val_accuracy: 0.7709 - val_loss: 0.4781\n",
      "Epoch 31/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7702 - loss: 0.4755 - val_accuracy: 0.7553 - val_loss: 0.4968\n",
      "Epoch 32/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7703 - loss: 0.4728 - val_accuracy: 0.7692 - val_loss: 0.4744\n",
      "Epoch 33/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7703 - loss: 0.4716 - val_accuracy: 0.7613 - val_loss: 0.4869\n",
      "Epoch 34/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7728 - loss: 0.4707 - val_accuracy: 0.7648 - val_loss: 0.4952\n",
      "Epoch 35/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7699 - loss: 0.4720 - val_accuracy: 0.7648 - val_loss: 0.4968\n",
      "Epoch 36/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7727 - loss: 0.4688 - val_accuracy: 0.7643 - val_loss: 0.5095\n",
      "Epoch 37/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7730 - loss: 0.4670 - val_accuracy: 0.7674 - val_loss: 0.4827\n",
      "Epoch 38/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7749 - loss: 0.4659 - val_accuracy: 0.7728 - val_loss: 0.4697\n",
      "Epoch 39/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m16s\u001b[0m 4ms/step - accuracy: 0.7772 - loss: 0.4618 - val_accuracy: 0.7753 - val_loss: 0.4774\n",
      "Epoch 40/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m16s\u001b[0m 4ms/step - accuracy: 0.7795 - loss: 0.4587 - val_accuracy: 0.7663 - val_loss: 0.4824\n",
      "Epoch 41/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 5ms/step - accuracy: 0.7765 - loss: 0.4638 - val_accuracy: 0.7561 - val_loss: 0.4910\n",
      "Epoch 42/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 4ms/step - accuracy: 0.7768 - loss: 0.4616 - val_accuracy: 0.7749 - val_loss: 0.4737\n",
      "Epoch 43/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 5ms/step - accuracy: 0.7800 - loss: 0.4554 - val_accuracy: 0.7698 - val_loss: 0.4747\n",
      "Epoch 44/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 5ms/step - accuracy: 0.7816 - loss: 0.4528 - val_accuracy: 0.7476 - val_loss: 0.4988\n",
      "Epoch 45/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 5ms/step - accuracy: 0.7819 - loss: 0.4553 - val_accuracy: 0.7630 - val_loss: 0.4820\n",
      "Epoch 46/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 5ms/step - accuracy: 0.7780 - loss: 0.4587 - val_accuracy: 0.7554 - val_loss: 0.4887\n",
      "Epoch 47/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 5ms/step - accuracy: 0.7832 - loss: 0.4555 - val_accuracy: 0.7773 - val_loss: 0.4709\n",
      "Epoch 48/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 5ms/step - accuracy: 0.7831 - loss: 0.4511 - val_accuracy: 0.7667 - val_loss: 0.4760\n",
      "Epoch 49/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m17s\u001b[0m 4ms/step - accuracy: 0.7831 - loss: 0.4513 - val_accuracy: 0.7731 - val_loss: 0.4812\n",
      "Epoch 50/50\n",
      "\u001b[1m3998/3998\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m15s\u001b[0m 4ms/step - accuracy: 0.7837 - loss: 0.4550 - val_accuracy: 0.7775 - val_loss: 0.4859\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(\n",
    "    X_train,\n",
    "    y_train,\n",
    "    epochs=50,\n",
    "    batch_size=16,\n",
    "    validation_data=(X_test, y_test),\n",
    "    callbacks=[checkpoint],\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tf.keras.models.load_model(\"model/TCN.keras\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m500/500\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        REAL       0.89      0.62      0.73      7920\n",
      "        FAKE       0.71      0.92      0.80      8072\n",
      "\n",
      "    accuracy                           0.77     15992\n",
      "   macro avg       0.80      0.77      0.77     15992\n",
      "weighted avg       0.80      0.77      0.77     15992\n",
      "\n",
      "[[4910 3010]\n",
      " [ 623 7449]]\n"
     ]
    }
   ],
   "source": [
    "# model = tf.keras.models.load_model(\"model/best_model.keras\")\n",
    "y_pred = model.predict(X_test)\n",
    "y_pred_labels = np.argmax(y_pred, axis=1)\n",
    "\n",
    "# Print classification report\n",
    "print(classification_report(y_test, y_pred_labels, target_names=[\"REAL\", \"FAKE\"]))\n",
    "print(confusion_matrix(y_test, y_pred_labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m1/1\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 169ms/step\n",
      "Found 8 deepfake frames:\n",
      "Frame 1 at 0.00s: FAKE\n",
      "Frame 2 at 1.00s: FAKE\n",
      "Frame 3 at 2.00s: FAKE\n",
      "Frame 4 at 3.00s: FAKE\n",
      "Frame 5 at 4.00s: FAKE\n",
      "Frame 8 at 7.00s: FAKE\n",
      "Frame 11 at 10.00s: FAKE\n",
      "Frame 15 at 14.00s: FAKE\n"
     ]
    }
   ],
   "source": [
    "def test_on_video(file_path, frame_duration=1.0):\n",
    "    # Load the trained model\n",
    "    model = tf.keras.models.load_model(\"model/TCN.keras\")\n",
    "\n",
    "    # Extract features and timestamps for each frame in the new video\n",
    "    frames, timestamps = extract_frame_features(file_path, frame_duration)\n",
    "\n",
    "    if frames is None or timestamps is None:\n",
    "        print(\"No frames extracted.\")\n",
    "        return\n",
    "\n",
    "    # Reshape frames for model input\n",
    "    frames = np.array(frames)[..., np.newaxis]\n",
    "\n",
    "    # Predict on each frame\n",
    "    predictions = model.predict(frames)\n",
    "    pred_labels = np.argmax(predictions, axis=1)\n",
    "\n",
    "    # Store deepfake frames, their timestamps, and frame indices\n",
    "    deepfake_frames = []\n",
    "    deepfake_timestamps = []\n",
    "    deepfake_indices = []\n",
    "\n",
    "    # Identify deepfake frames\n",
    "    for i, label in enumerate(pred_labels):\n",
    "        if label == 1:  # If the label is FAKE\n",
    "            deepfake_frames.append(frames[i])\n",
    "            deepfake_timestamps.append(timestamps[i])\n",
    "            deepfake_indices.append(i)\n",
    "\n",
    "    if not deepfake_frames:\n",
    "        print(\"No deepfake frames detected in the video.\")\n",
    "        return\n",
    "\n",
    "    # Analyze deepfake frames\n",
    "    print(f\"Found {len(deepfake_frames)} deepfake frames:\")\n",
    "    for i, (timestamp, index) in enumerate(zip(deepfake_timestamps, deepfake_indices)):\n",
    "        print(f\"Frame {index + 1} at {timestamp:.2f}s: FAKE\")\n",
    "\n",
    "\n",
    "# Example usage\n",
    "test_video_path = r\"REAL\\ajqslcypsw.mp4\"  # Replace with your test video path\n",
    "test_on_video(test_video_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def test_on_video(file_path, frame_duration=1.0):\n",
    "#     # Load the trained model\n",
    "#     model = tf.keras.models.load_model(\"model/best_model.keras\")\n",
    "\n",
    "#     # Extract features for each frame in the new video\n",
    "#     frames = extract_frame_features(file_path, frame_duration)\n",
    "\n",
    "#     if frames is None:\n",
    "#         print(\"No frames extracted.\")\n",
    "#         return\n",
    "\n",
    "#     # Reshape frames for model input\n",
    "#     frames = np.array(frames)[..., np.newaxis]\n",
    "\n",
    "#     # Predict on each frame\n",
    "#     predictions = model.predict(frames)\n",
    "#     pred_labels = np.argmax(predictions, axis=1)\n",
    "\n",
    "#     # Output results for each frame\n",
    "#     for i, label in enumerate(pred_labels):\n",
    "#         status = \"REAL\" if label == 0 else \"FAKE\"\n",
    "#         print(f\"Frame {i+1}: {status}\")\n",
    "\n",
    "\n",
    "# # Example usage\n",
    "# test_video_path = r\"REAL\\bddjdhzfze.mp4\"  # Replace with your test video path\n",
    "# test_on_video(test_video_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# @register_keras_serializable()\n",
    "# class AudioModel(tf.keras.Model):\n",
    "#     def __init__(self, input_shape):\n",
    "#         super(AudioModel, self).__init__()\n",
    "#         self.input_shape = input_shape  # Store the input shape\n",
    "#         # Define the model layers\n",
    "#         self.conv1 = layers.Conv2D(\n",
    "#             32, kernel_size=(3, 3), activation=\"relu\", input_shape=input_shape\n",
    "#         )\n",
    "#         self.conv2 = layers.Conv2D(64, kernel_size=(3, 3), activation=\"relu\")\n",
    "#         self.pool = layers.MaxPooling2D(pool_size=(2, 2))\n",
    "#         self.dropout1 = layers.Dropout(0.25)\n",
    "\n",
    "#         self.reshape = layers.Reshape((64, -1))\n",
    "#         self.gru = layers.Bidirectional(layers.GRU(128, return_sequences=False))\n",
    "\n",
    "#         self.dense1 = layers.Dense(128, activation=\"relu\")\n",
    "#         self.dropout2 = layers.Dropout(0.5)\n",
    "#         self.dense2 = layers.Dense(2, activation=\"softmax\")\n",
    "\n",
    "#     def call(self, inputs):\n",
    "#         # Forward pass through the layers\n",
    "#         x = self.conv1(inputs)\n",
    "#         x = self.conv2(x)\n",
    "#         x = self.pool(x)\n",
    "#         x = self.dropout1(x)\n",
    "\n",
    "#         x = self.reshape(x)\n",
    "#         x = self.gru(x)\n",
    "\n",
    "#         x = self.dense1(x)\n",
    "#         x = self.dropout2(x)\n",
    "#         return self.dense2(x)\n",
    "\n",
    "#     def get_config(self):\n",
    "#         config = super(AudioModel, self).get_config()\n",
    "#         config.update(\n",
    "#             {\"input_shape\": self.input_shape}  # Include input shape in config\n",
    "#         )\n",
    "#         return config\n",
    "\n",
    "#     @classmethod\n",
    "#     def from_config(cls, config):\n",
    "#         # Create a model instance from the config\n",
    "#         input_shape = config.pop(\"input_shape\")  # Extract input_shape from config\n",
    "#         return cls(input_shape)  # Create an instance of the model\n",
    "\n",
    "\n",
    "# # Function to create and compile the model\n",
    "# def create_model(input_shape):\n",
    "#     model = AudioModel(input_shape)\n",
    "#     model.compile(\n",
    "#         optimizer=\"adam\", loss=\"sparse_categorical_crossentropy\", metrics=[\"accuracy\"]\n",
    "#     )\n",
    "#     return model\n",
    "\n",
    "\n",
    "# # Example usage\n",
    "# input_shape = (\n",
    "#     64,\n",
    "#     40,\n",
    "#     1,\n",
    "# )  # Adjust based on your data (e.g., (n_mfccs, time_steps, channels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model = create_model(input_shape)\n",
    "# model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# checkpoint = ModelCheckpoint(r\"models/dl_model.keras\", monitor=\"val_loss\", save_best_only=True, verbose=1)\n",
    "# early_stopping = EarlyStopping(monitor=\"val_loss\", patience=5, verbose=1)\n",
    "\n",
    "# history = model.fit(\n",
    "#     X_train, y_train, epochs=10, batch_size=16, validation_data=(X_test, y_test), callbacks=[checkpoint, early_stopping]\n",
    "# )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.save(r\"models/dl_model.keras\", overwrite=True)\n",
    "# print(\"Model saved successfully.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Ensure to import keras properly\n",
    "# import tensorflow as tf\n",
    "# from tensorflow import keras\n",
    "\n",
    "\n",
    "# # Function to load the model\n",
    "# def load_model(model_path):\n",
    "#     try:\n",
    "#         # Load the model from the specified path\n",
    "#         model = keras.models.load_model(model_path)\n",
    "#         print(\"Model loaded successfully.\")\n",
    "#         return model\n",
    "#     except Exception as e:\n",
    "#         print(f\"Error loading model: {e}\")\n",
    "#         return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model_path = r\"models/dl_model.keras\"\n",
    "\n",
    "# # Load the model\n",
    "# loaded_model = load_model(model_path)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}