Akash-Apturkar/Sentiment-Analysis-of-speech-using-NLP-with-Android-Connect-feature-and-web-scraping

We aim to develop a ‘Smart Speech Ecosystem’ that takes audio input, converts it to text and performs various analytical functions on it. Our project focuses on performing Sentiment Analysis on the input speech, extracting relevant data from the speech using NLTK and also, online opinion analysis of news headlines regarding a topic using 'Beautifulsoup'. First, we test the ‘SentimentIntensityAnalyzer’ model of ‘nltk.vader’ for accuracy of polarity scores Our project functions in 3 modes: 1.Speech Mode: takes speech input from the user and performs the following analytical functions using NLTK : Tokenization, PoS tagging, Word Lemmatization, Name entity tagging. Extracting emotions from speech and plotting the weightage of emotions using matplotlib. Constructing the sentence structure tree. Extracting person names from the speech. Web scraping Google news headlines for these names to determine online opinion about the person. 2.Opinion Mode: takes a topic name from the user and runs Web scraping of Google news headlines for this topic to determine online opinion about the topic. 3.Remote Mode: takes the speech input remotely from a Kivy based android app and runs the same analysis as speech mode. The audio input file is stored on an AWS S3 bucket by the app and is fetched by the python script.