diff --git a/.gitignore b/.gitignore
index 1dce118..42026b8 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,3 +4,10 @@ Data/
 .idea/
 .venv/
 */spectogram_model.h5
+spectrogram_classifier.h5
+build/
+dist/
+installer_output/
+*.spec
+hook_tensorflow.py
+hook_runtime.py
diff --git a/README.md b/README.md
index ea72563..8148f1a 100644
--- a/README.md
+++ b/README.md
@@ -1,7 +1,23 @@
-## Instructions
+## Running from Source
 
-1. Install dependencies by running `pip install -r requirements.txt`
-2. Then run the gui by using `python3 gui.py`
+1. Install dependencies: `pip install -r requirements.txt`
+2. Launch the app: `python gui.py`
+
+## Building the Executable
+
+Requires PyInstaller (`pip install pyinstaller`):
+
+```bash
+.venv/Scripts/pyinstaller --name "MelodiProof-AI" --noconsole --collect-all onnxruntime --add-data "App_Spectrograms;App_Spectrograms" --add-data "App_Covers;App_Covers" --add-data "App_Audio;App_Audio" --add-data "spectrogram_classifier.onnx;." --hidden-import="PIL._tkinter_finder" --runtime-hook="hook_runtime.py" -y gui.py
+```
+
+The executable will be in `dist/MelodiProof-AI/`.
+
+## Building the Installer
+
+1. Install [Inno Setup](https://jrsoftware.org/isinfo.php)
+2. Open `installer.iss` in Inno Setup and click **Compile**
+3. The installer (`MelodiProof-AI-Setup.exe`) will be output to `installer_output/`
 
 ## Datasets
 
diff --git a/gui.py b/gui.py
index aebce84..9e7b390 100644
--- a/gui.py
+++ b/gui.py
@@ -1,21 +1,25 @@
 import os
+import sys
 import tkinter as tk
 import numpy as np
 from tkinter import ttk
 from PIL import Image, ImageTk
-from tensorflow.keras.models import load_model
-from tensorflow.keras.preprocessing.image import load_img, img_to_array
+import onnxruntime as ort
 import pygame  # For audio playback
 
+# Resolve base path for bundled resources (PyInstaller) or source directory
+BASE_DIR = getattr(sys, '_MEIPASS', os.path.dirname(os.path.abspath(__file__)))
+
 # Constants
 IMG_HEIGHT, IMG_WIDTH = 128, 128
-SPECTROGRAM_DIR = "App_Spectrograms"
-COVERS_DIR = "App_Covers"
-MP3_DIR = "App_Audio"
-MODEL_PATH = "spectrogram_classifier.h5"
+SPECTROGRAM_DIR = os.path.join(BASE_DIR, "App_Spectrograms")
+COVERS_DIR = os.path.join(BASE_DIR, "App_Covers")
+MP3_DIR = os.path.join(BASE_DIR, "App_Audio")
+MODEL_PATH = os.path.join(BASE_DIR, "spectrogram_classifier.onnx")
 
 # Load the trained model
-model = load_model(MODEL_PATH)
+session = ort.InferenceSession(MODEL_PATH)
+input_name = session.get_inputs()[0].name
 
 # Initialize pygame mixer for audio playback
 pygame.mixer.init()
@@ -26,13 +30,10 @@
 
 
 def predict_spectrogram(image_path):
-    """
-    Predict whether a spectrogram belongs to AI-Generated or Not.
-    """
-    img = load_img(image_path, target_size=(IMG_HEIGHT, IMG_WIDTH))
-    img_array = img_to_array(img) / 255.0
-    img_array = img_array[np.newaxis, ...]  # Add batch dimension
-    prediction = model.predict(img_array)
+    img = Image.open(image_path).convert("RGB").resize((IMG_WIDTH, IMG_HEIGHT))
+    img_array = np.array(img, dtype=np.float32) / 255.0
+    img_array = img_array[np.newaxis, ...]
+    prediction = session.run(None, {input_name: img_array})[0]
     return "AI-Generated" if prediction[0] < 0.5 else "Not AI-Generated"
 
 
@@ -44,7 +45,7 @@ def get_cover_image(song_name, size=(50, 50)):
     if os.path.exists(cover_path):
         cover_image = Image.open(cover_path).resize(size)
     else:
-        cover_image = Image.open("App_Covers/placeholder.jpeg").resize(size)
+        cover_image = Image.open(os.path.join(COVERS_DIR, "placeholder.jpeg")).resize(size)
     return ImageTk.PhotoImage(cover_image)
 
 
diff --git a/installer.iss b/installer.iss
new file mode 100644
index 0000000..d26479a
--- /dev/null
+++ b/installer.iss
@@ -0,0 +1,28 @@
+[Setup]
+AppName=MelodiProof AI
+AppVersion=1.0
+AppPublisher=MelodiProof
+DefaultDirName={autopf}\MelodiProof-AI
+DefaultGroupName=MelodiProof AI
+OutputDir=installer_output
+OutputBaseFilename=MelodiProof-AI-Setup
+Compression=lzma
+SolidCompression=yes
+WizardStyle=modern
+
+[Languages]
+Name: "english"; MessagesFile: "compiler:Default.isl"
+
+[Tasks]
+Name: "desktopicon"; Description: "{cm:CreateDesktopIcon}"; GroupDescription: "{cm:AdditionalIcons}"; Flags: unchecked
+
+[Files]
+Source: "dist\MelodiProof-AI\*"; DestDir: "{app}"; Flags: ignoreversion recursesubdirs createallsubdirs
+
+[Icons]
+Name: "{group}\MelodiProof AI"; Filename: "{app}\MelodiProof-AI.exe"
+Name: "{group}\{cm:UninstallProgram,MelodiProof AI}"; Filename: "{uninstallexe}"
+Name: "{autodesktop}\MelodiProof AI"; Filename: "{app}\MelodiProof-AI.exe"; Tasks: desktopicon
+
+[Run]
+Filename: "{app}\MelodiProof-AI.exe"; Description: "{cm:LaunchProgram,MelodiProof AI}"; Flags: nowait postinstall skipifsilent
diff --git a/requirements.txt b/requirements.txt
index 2aa02b3..325f2bd 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -2,8 +2,7 @@ pydub~=0.25.1
 numpy~=1.26.4
 matplotlib~=3.9.2
 scipy~=1.14.1
-tensorflow~=2.18.0
-scikit-learn~=1.5.1
 pillow~=10.4.0
 librosa~=0.10.2.post1
-pygame~=2.6.1
\ No newline at end of file
+pygame~=2.6.1
+onnxruntime~=1.25.0
\ No newline at end of file
diff --git a/spectrogram_classifier.onnx b/spectrogram_classifier.onnx
new file mode 100644
index 0000000..aa4e9d0
Binary files /dev/null and b/spectrogram_classifier.onnx differ