Sentiment analysis of speech with application to various languages

Abstract

In this paper we aim to explore and implement a modern speech recognition system using Natural Language Processing (NLP) and sentiment analysis that can be applied in the area of audio and text archive investigation from various languages. To that end we developed a project that can be used to convert speech to text and perform various analysis on a converted text. Furthermore, we focused on recognizing different information such as names, emotions and also determine the overall sentiment of a script. Furthermore, we perform web scraping for names and organizations of importance used in speech. To achieve this, we used Python with various specialized modules. In order to simplify the task of collecting and storing audio inputs for processing we have developed an Android app with connection to a cloud database. This methodology can easily be applied for the purposes of digital presentation and preservation of cultural and scientific heritage.