Deep Fake Voice Recognition.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "32a64ee8-4432-46b9-aa30-c36bdefc4eb8",
   "metadata": {},
   "source": [
    "# Import Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9d741976-92ae-42af-90e6-3edce859ac86",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "import librosa\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from tqdm import tqdm\n",
    "import IPython\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from imblearn.over_sampling import RandomOverSampler\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Activation, Dropout, Conv2D, MaxPool2D, Flatten\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from tensorflow.keras.callbacks import EarlyStopping"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5395f80-b176-4131-85b4-b86fa9a12439",
   "metadata": {},
   "source": [
    "# Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5a5de763-45e7-4f81-9459-163282c6d026",
   "metadata": {},
   "outputs": [],
   "source": [
    "real_audio = \"/mnt/hdd/Datasets/deep-voice/DEMONSTRATION/DEMONSTRATION/linus-original-DEMO.mp3\"\n",
    "fake_audio = \"/mnt/hdd/Datasets/deep-voice/DEMONSTRATION/DEMONSTRATION/linus-to-musk-DEMO.mp3\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6026553f-eca6-4521-a200-36df71db6d55",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Real Audio:\")\n",
    "IPython.display.Audio(real_audio)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "77a8b1c5-e5a7-49ea-888c-09211666d3ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Fake Audio:\")\n",
    "IPython.display.Audio(fake_audio)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2cf9f079-4ca5-4f7f-b6a6-c0b61cf6a814",
   "metadata": {},
   "outputs": [],
   "source": [
    "real_ad, real_sr = librosa.load(real_audio)\n",
    "plt.figure(figsize=(12, 4))\n",
    "plt.plot(real_ad)\n",
    "plt.title(\"Real Audio Data\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "026e80de-5ce3-4734-a16b-1a6dbd2ad3cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "real_spec = np.abs(librosa.stft(real_ad))\n",
    "real_spec = librosa.amplitude_to_db(real_spec, ref=np.max)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(real_spec, sr=real_sr, x_axis=\"time\", y_axis=\"log\")\n",
    "plt.colorbar(format=\"%+2.0f dB\")\n",
    "plt.title(\"Real Audio Spectogram\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6596ce0b-8424-42d1-a949-3d7f62334252",
   "metadata": {},
   "outputs": [],
   "source": [
    "real_mel_spect = librosa.feature.melspectrogram(y=real_ad, sr=real_sr)\n",
    "real_mel_spect = librosa.power_to_db(real_mel_spect, ref=np.max)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(real_mel_spect, y_axis=\"mel\", x_axis=\"time\")\n",
    "plt.title(\"Real Audio Mel Spectogram\")\n",
    "plt.colorbar(format=\"%+2.0f dB\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7689ff78-9dbf-464a-ae51-3cb6c871d5b0",
   "metadata": {},
   "outputs": [],
   "source": [
    "real_chroma = librosa.feature.chroma_cqt(y=real_ad, sr=real_sr, bins_per_octave=36)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(real_chroma, sr=real_sr, x_axis=\"time\", y_axis=\"chroma\", vmin=0, vmax=1)\n",
    "plt.colorbar()\n",
    "plt.title(\"Real Audio Chromagram\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e90f20a9-2f81-452b-908e-199706d40de1",
   "metadata": {},
   "outputs": [],
   "source": [
    "real_mfccs = librosa.feature.mfcc(y=real_ad, sr=real_sr)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(real_mfccs, sr=real_sr, x_axis=\"time\")\n",
    "plt.colorbar()\n",
    "plt.title(\"Real Audio Mel-Frequency Cepstral Coefficients (MFCCs)\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5ff1182f-ebf8-4373-93a0-792b832888a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "fake_ad, fake_sr = librosa.load(fake_audio)\n",
    "plt.figure(figsize=(12, 4))\n",
    "plt.plot(fake_ad)\n",
    "plt.title(\"Fake Audio Data\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c48a4cf8-69dd-44c9-b0ed-7eeb8a9e070d",
   "metadata": {},
   "outputs": [],
   "source": [
    "fake_spec = np.abs(librosa.stft(fake_ad))\n",
    "fake_spec = librosa.amplitude_to_db(fake_spec, ref=np.max)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(fake_spec, sr=fake_sr, x_axis=\"time\", y_axis=\"log\")\n",
    "plt.colorbar(format=\"%+2.0f dB\")\n",
    "plt.title(\"Fake Audio Spectogram\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5adc2950-1985-4225-b7bc-d8823499526f",
   "metadata": {},
   "outputs": [],
   "source": [
    "fake_mel_spect = librosa.feature.melspectrogram(y=fake_ad, sr=fake_sr)\n",
    "fake_mel_spect = librosa.power_to_db(fake_mel_spect, ref=np.max)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(fake_mel_spect, y_axis=\"mel\", x_axis=\"time\")\n",
    "plt.title(\"Fake Audio Mel Spectogram\")\n",
    "plt.colorbar(format=\"%+2.0f dB\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7a052ee8-7204-49e5-b049-6302b4a11f86",
   "metadata": {},
   "outputs": [],
   "source": [
    "fake_chroma = librosa.feature.chroma_cqt(y=fake_ad, sr=fake_sr, bins_per_octave=36)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(fake_chroma, sr=fake_sr, x_axis=\"time\", y_axis=\"chroma\", vmin=0, vmax=1)\n",
    "plt.colorbar()\n",
    "plt.title(\"Fake Audio Chromagram\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bb7d1a38-c90f-4b67-a01c-e3afd248406e",
   "metadata": {},
   "outputs": [],
   "source": [
    "fake_mfccs = librosa.feature.mfcc(y=fake_ad, sr=fake_sr)\n",
    "\n",
    "plt.figure(figsize=(14, 5))\n",
    "librosa.display.specshow(fake_mfccs, sr=fake_sr, x_axis=\"time\")\n",
    "plt.colorbar()\n",
    "plt.title(\"Fake Audio Mel-Frequency Cepstral Coefficients (MFCCs)\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b26d17db-b554-4a11-9cc9-578cf3a5e1c3",
   "metadata": {},
   "outputs": [],
   "source": [
    "audio_files_path = \"/mnt/hdd/Datasets/deep-voice/KAGGLE/AUDIO\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "89f80c1c-0ac1-4f88-8c2f-e7b9c763b3c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "folders = os.listdir(audio_files_path)\n",
    "folders"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "62e9e13d-633d-4bbd-a85e-3d18017292d7",
   "metadata": {},
   "source": [
    "# Preprocess"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9bf47c86-5a43-4c5e-9031-7f9fff8d2055",
   "metadata": {},
   "outputs": [],
   "source": [
    "data = []\n",
    "labels = []\n",
    "\n",
    "for folder in folders:\n",
    "    files = os.listdir(os.path.join(audio_files_path, folder))\n",
    "    for file in tqdm(files):\n",
    "        file_path = os.path.join(audio_files_path, folder, file)\n",
    "        audio, sample_rate = librosa.load(file_path, res_type=\"kaiser_fast\")\n",
    "        mfccs_features = librosa.feature.mfcc(y=audio, sr=sample_rate, n_mfcc=40)\n",
    "        mfccs_features_scaled = np.mean(mfccs_features.T, axis=0)\n",
    "        data.append(mfccs_features_scaled)\n",
    "        labels.append(folder)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ed887837-319a-4a2f-8b63-e63f0f1c2c6f",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_df = pd.DataFrame({\"features\": data, \"class\": labels})\n",
    "feature_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5a343d7c-aa77-4ec9-a200-d5fe259749ed",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_df[\"class\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "527d0aaa-7c90-4236-afae-6e2ad60a40d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "def label_encoder(column):\n",
    "    le = LabelEncoder().fit(column)\n",
    "    print(column.name, le.classes_)\n",
    "    return le.transform(column)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "852b2260-5657-4b8e-84ae-e404cd4d3ce2",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_df[\"class\"] = label_encoder(feature_df[\"class\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c6222b46-de27-47d3-8205-cd289a8ef1d5",
   "metadata": {},
   "source": [
    "# Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2aa42eed-e7f5-4cc7-8103-40d891de4a63",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = np.array(feature_df[\"features\"].tolist())\n",
    "y = np.array(feature_df[\"class\"].tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8def9808-06e9-46e0-8bf4-98d44b772ece",
   "metadata": {},
   "outputs": [],
   "source": [
    "ros = RandomOverSampler(random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "df3a375f-d32c-492a-ab9d-0bd46d6fff8a",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_resampled, y_resampled = ros.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7839fda8-cf28-4b8f-93c3-a6eb05cfb52e",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_resampled = to_categorical(y_resampled)\n",
    "#y = to_categorical(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d4f7c017-010e-439b-a55b-02f2b2e594f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X_resampled, y_resampled, test_size=0.2, random_state=42)\n",
    "#X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5d7a5200-c087-4cd2-b7f9-d85eb9d667df",
   "metadata": {},
   "outputs": [],
   "source": [
    "num_labels = len(feature_df[\"class\"].unique())\n",
    "num_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "821d626d-0957-4e10-bf0c-3984c96e2c97",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_shape = feature_df[\"features\"][0].shape\n",
    "input_shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2059bcb1-71d7-4879-b99d-69d0d6b0da81",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(128, input_shape=input_shape))\n",
    "model.add(Activation(activation=\"relu\"))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(256))\n",
    "model.add(Activation(activation=\"relu\"))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(128))\n",
    "model.add(Activation(activation=\"relu\"))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(num_labels))\n",
    "model.add(Activation(activation=\"softmax\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fb6254fb-c497-4f8a-90ca-6a986a69053c",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c9d4e82a-bf46-4333-8b37-9c2455ed558c",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "153cf41a-bccc-485b-bdf8-d384326daa2a",
   "metadata": {},
   "outputs": [],
   "source": [
    "early = EarlyStopping(monitor=\"val_loss\", patience=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d8549b9f-7f6e-4e29-a88c-b48f84754601",
   "metadata": {},
   "outputs": [],
   "source": [
    "history = model.fit(X_train, y_train, validation_data=(X_test, y_test), batch_size=2, epochs=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0e984f68-4c33-49b3-a853-184fe2d57750",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save(\"deepfake.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5fcb9a91-a575-407c-b4c1-7975dafbde7b",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_loss, test_acc = model.evaluate(X_test, y_test)\n",
    "print(\"Test Loss:\", test_loss)\n",
    "print(\"Test Accuracy:\", test_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e71f7a68-4001-4476-9f74-2f213fd30006",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "plt.title(\"Model Accuracy\")\n",
    "plt.plot(history.history[\"accuracy\"], label=\"train\")\n",
    "plt.plot(history.history[\"val_accuracy\"], label=\"validation\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8e869e82-a8dd-4449-ba6d-753e93aa9281",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "plt.title(\"Model Loss\")\n",
    "plt.plot(history.history[\"loss\"], label=\"train\")\n",
    "plt.plot(history.history[\"val_loss\"], label=\"validation\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc40dfdd-b10d-4177-a1b0-efd0995f8335",
   "metadata": {},
   "source": [
    "# Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2e45aaf8-9372-41c9-a668-95ae58472d64",
   "metadata": {},
   "outputs": [],
   "source": [
    "def detect_fake(filename):\n",
    "    sound_signal, sample_rate = librosa.load(filename, res_type=\"kaiser_fast\")\n",
    "    mfcc_features = librosa.feature.mfcc(y=sound_signal, sr=sample_rate, n_mfcc=40)\n",
    "    mfccs_features_scaled = np.mean(mfcc_features.T, axis=0)\n",
    "    mfccs_features_scaled = mfccs_features_scaled.reshape(1, -1)\n",
    "    result_array = model.predict(mfccs_features_scaled)\n",
    "    print(result_array)\n",
    "    result_classes = [\"FAKE\", \"REAL\"]\n",
    "    result = np.argmax(result_array[0])\n",
    "    print(\"Result:\", result_classes[result])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2736de0b-8406-4275-9942-c84227d9f8b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_real = \"/mnt/hdd/Datasets/deep-voice/DEMONSTRATION/DEMONSTRATION/linus-original-DEMO.mp3\"\n",
    "test_fake = \"/mnt/hdd/Datasets/deep-voice/DEMONSTRATION/DEMONSTRATION/linus-to-musk-DEMO.mp3\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "92faad52-976b-4e17-bf88-568fff77a1cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "detect_fake(test_real)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e732ae95-647e-42d0-a885-025ee9176ab4",
   "metadata": {},
   "outputs": [],
   "source": [
    "detect_fake(test_fake)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5c8938da-c0b1-4dd2-aef6-55979686e520",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}