Documentation

Kannada Speech Synthesis Application (GUI)

v3.0 Stable Release

Kannada Speech Synthesis project (Core)

Kannada Speech Synthesis is submitted as the final year project which is a part of the curriculum specified by the Department of Electronics and Communication of M S Ramaiah Institute of Technology, Bangalore. All the dependencies and frameworks which are used in the core project are mentioned in the About page of the application. Contact Developer to obtain the source and trained model of the core project.

Developer

Shashank Sharma (Click to Expand)

MAIL

Core Project 
- Preprocessing
- Training
- Pitch Shift using soundtretch
- Testing on 
	- Ubuntu 18.04 on Windows Subsystem for Linux(WSL)  [Operational as Expected]
	- Ubuntu 18.04 on Google Cloud [Operational as Expected]
	- CentOS7 on Google Cloud [Not Operational due to Missing Libraries]

Application
- Handle Numbers in input
- Complete User Interface (Main, About, Table and Analysis Window)
- Plots (Wave, Spectrum, Spectrogram, MFCC) and Text analysis
- Kannada Version Build
- SQLite Database Implementation ( Synthesis and Translation)
- Media Player Devt.
- Database Search (With Autocomplete)
- Production Code and Documentation
- Application Themes
- Testset Integration with App

Recent Updates

Recent Updates on Core Project

1. The core project is now tested on Ubuntu18.04 WSL (Windows Subsystem for Linux)
2. Noise removal using SOX
3. Pitch Shifting using soundstretch

Recent Updates on GUI Application

1. Handles Number input (Currently supporting 0-1 Lakh)
2. Play unprocessed audio from the misc menu
3. Open Terminal from the Application (Alt + T/P/G )
4. Added MFCC colormap plot
5. The application is now available in Kannada Version (./run.sh -kan or Change language within the application)
6. The audio if in testset can now be played in the misc. menu.

Features of GUI Application

1. SQLite Based Database
2. In-App Media Player
3. Easy to use UI
4. Database View for easy selection with search feature (Synthesis and Translation)
5. Audio Analysis for Speech Enthusiasts
6. Themes that you can choose

Screenshots

Check the Application Screenshots in the Screenshots Directory.

Logo

Designed by Shashank Sharma on Canva

Building and Starting the Application

Request for Trained Model.

Request Developer for the trained model of synthesizing Kannada Text. Setup Backend for your machine. The backend will be hosted on Web Server in future releases. Check the file Core/README.md for installation instructions.

Building the Application

1. Open App/Main.py and checkout setEnv() function. Change the directory path as per your project.

2. Run the command below

    chmod 755 ./run.sh
   

Start the Application by running run.sh

./run.sh

License

GNU GPL v3.0

WaveCLI

A collection of Wave Plot, Process and Manipulating utilities available in command line interface scripts and is developed by Shashank Sharma for this project. Have a look at the project here. The same project in modular version is used for analysis window in Kannada TTS Application.