Final Product Report - Team WannaBe

Abstract

Do you wish to express yourself artistically? We believe that becoming an engineer, doctor, or lawyer is relatively simple, yet they argue that artists are born, not produced. Here is an AI-powered program that can help you become a singer or a painter! You may have seen other efforts aimed at certain fields like healthcare, hardware design, and so on. We stopped at nothing in the project to investigate if AI can assist us with modest everyday routine activities. We investigated the use of AI/ML technologies to benefit our everyday lives. We have devised strategies to clarify your voice (transform your bathroom singing to studio-recorded audio quality) and reduce background noise from home-recorded meetings (COVID effect!). It may assist you in transforming your sloppy artwork into a gorgeous painting. Of course, this isn't only for artists; it can be used as a 'filter' for Instagram or Snapchat, and it may be quite useful for police agencies when producing drawings from scribbles. The trick is training the model on a large dataset and utilizing a Generative Adversarial Network (the greatest invention in AI), which results in mapping and producing pictures with a high match score based on the training dataset. We have developed a web application that has two segments : i) Wanna be a Painter - This segment allows users to upload grayscale images/sketches or hand drawings to convert them into beautiful paintings. ii) Wanna be a Singer - This segment allows users to upload audio files into the web application and download the augmented audio with crisp and clear voice.

Gitlab link - https://code.vt.edu/wannabe

Description
Keywords
Painter, Singer, Deep Learning, GAN, NodeJS, React, Flask, Neural Networks, PyTorch, Tensorflow, Audio, Image
Citation