Persian Language Speech Recognition System

Overview

This project implements a speech recognition system for Persian language commands using a Convolutional Neural Network (CNN). The system is developed in Python, leveraging TensorFlow and Keras, and it classifies spoken commands based on an audio dataset.

Features

• CNN Model: Trained using a dataset of Persian commands.
• Dataset: Custom Persian voice commands in WAV format.
• Example Audio: Includes a sample audio file (roshan.wav) for testing.

File Structure

• CI_Lab_Project.py: Python script that builds and trains the CNN model.
• classification_model35.h5: Pretrained CNN model for classification.
• Voices-wav.zip: Compressed dataset containing voice recordings.
• roshan.wav: Sample audio file used for testing the model.

---

Installation

To run this project, you'll need Python and several libraries. Follow these steps:

Prerequisites

• Python 3.x
• TensorFlow
• Keras
• NumPy
• LibROSA

Steps

1. Clone the repository

   git clone https://github.com/AminLari/Persian-Language-Speech-Recognition-System.git
   

2. Install the dependencies

   pip install tensorflow librosa keras numpy
   

Usage

1. Extract the dataset: After downloading the repository, you need to extract the voice command dataset.

   unzip Voices-wav.zip
   

2. Run the training and evaluation script: To train the CNN model on the dataset, run the following Python script. The script will load the dataset, preprocess the audio data, and train the model.

   python CI_Lab_Project.py
   

3. Test the model with a sample audio: After training the model, you can test it with a sample audio file. The provided sample file is roshan.wav. Modify the script to classify your own audio files if needed.

   # Modify script to test a different file
   test_audio = 'path/to/your/audio.wav'
   prediction = model.predict(test_audio)
   print(f'Predicted Class: {prediction}')
   

Results

The developed framework extracts time-frequency domain features from audio signals using MFCC.

The model can be evaluated using the provided sample audio file (roshan.wav).

• Sample Input: roshan.wav
• Predicted Output: The model predicts the class label for the input audio file.

📞 Contact

If you have any questions, feedback, or suggestions regarding this project, feel free to reach out:

• Name: Mohammadamin Lari
• Email: mohammadamin.lari@gmail.com
• GitHub: AminLari

You are welcome to create issues or pull requests to improve the project. Contributions are highly appreciated!