Voice Language Prediction Using Machine Learning

Abstract

When a speaker cannot be positively recognized, speech-language identification may be used to determine the language they are speaking. Some basic machine-learning methods will also be covered. The ability to recognize human speech may be learned. First and foremost, we must establish search criteria. Differentiating languages by their spoken characteristics may help to derive a feature. After collecting all the necessary audio data from a variety of sources, we compress the resulting audio file. Methods of identifying languages have been used (LID). With the highest possible F1 score, Machine Learning is the most successful tactic. Therefore, we used machine learning classifiers in our studies. Our work focuses on predicting spoken language. To get the highest possible F1 score, we will employ a Machine Learning classifier in situations when identifying the other person's language is straightforward. Five algorithms were employed to get to this point. This work's Decision tree method has the highest F1 score and highest accuracy. Others have been more accurate, although they have been overfitting. Therefore, the decision tree approach proved effective for our project.