tahamajs/ML/Files/CAs/Project - Storage Bucket

Final Project: Voice Authentication and Gender Classification using Machine Learning

Overview

This project focuses on utilizing machine learning algorithms to perform voice authentication and gender classification using audio features. The aim is to design a model capable of distinguishing between male and female speakers and authenticating the identity of the speaker using their voice, based on various audio features.

Objectives

1. Voice Authentication : The model will authenticate a speaker's identity based on their voice using closed-set and open-set authentication techniques.
2. Gender Classification : The model will classify the gender of the speaker from audio features.
3. Feature Extraction : Extract relevant features from audio signals, such as Mel-frequency cepstral coefficients (MFCC) , spectral centroid , log Mel spectrogram , and more.
4. Data Preprocessing : Implement noise reduction, normalization, and windowing techniques to enhance the quality of the input data.

Project Structure

The project is divided into several phases:

Phase 1: Introduction to Audio Signal Processing

• Data Preprocessing : Learn and implement the basic signal processing techniques, such as:
• Noise reduction (e.g., using spectral subtraction or bandpass filtering).
• Normalization and windowing techniques.
• Feature Extraction : Apply various algorithms to extract features from audio signals. Focus will be on:
• MFCC (Mel Frequency Cepstral Coefficients)
• Spectral Centroid
• Log Mel Spectrogram
• Zero-Crossing Rate
• Chroma Features
• Spectral Contrast

Phase 2: Practical Implementation with Dataset

• Dataset : The dataset consists of audio files in MP3 format, named with the pattern HWx_Qy_Student-ID_gender.mp3.
• Tasks :
• Gender Classification : Use machine learning models to classify the gender of the speaker.
• Voice Authentication : Use models to authenticate the voice based on prior recordings from the same individual.
• Steps :

1. Data Loading : Load the raw audio files.
2. Data Preprocessing : Clean and prepare the data by removing noise and resampling.
3. Feature Extraction : Apply feature extraction techniques.
4. Model Training : Train models using extracted features for both tasks (gender classification and voice authentication).
5. Model Evaluation : Evaluate the models using appropriate metrics (e.g., accuracy, precision, recall, F1-score, confusion matrix).

Phase 3: Report Writing and Analysis

• Initial Report : Prepare a group report covering:
• Voice authentication methods (closed-set and open-set).
• Data preprocessing challenges and solutions.
• Comparison of feature extraction techniques.
• Final Report : The final report will include:
• A detailed explanation of the models used, the results, and the challenges faced.
• A comparative analysis of different classification techniques such as Logistic Regression, KNN, SVM, MLP, and XGBoost.
• Evaluation using performance metrics and visualizations (e.g., ROC Curve, confusion matrix).
• Clustering Analysis : Apply clustering algorithms to identify patterns within the audio data and visualize the results.

Phase 4: Grading Criteria

The grading will be based on the following:

• Initial Report (20%)
• Preprocessing and Feature Extraction (30%)
• Classification (30%)
• Clustering (20%)

Phase 5: Final Considerations

• Plagiarism Policy : Any form of plagiarism or copying from other students will result in severe penalties.
• Collaboration : The project can be done individually or in a group of up to 4 people.

Required Libraries and Tools

• Python : Programming language used for model development.
• Librosa : Library for audio processing.
• Scikit-learn : For machine learning models.
• TensorFlow / Keras : For deep learning models (if applicable).
• Matplotlib : For visualization.

How to Run the Project

1. Clone the Repository :

   git clone https://github.com/your-repository/voice-authentication-project.git

1. Install Dependencies :

   pip install -r requirements.txt

1. Data Preparation : Download the dataset from Google Drive and place it in the appropriate folder.
2. Run the Code : Execute the scripts for data preprocessing, feature extraction, and model training.

---