English Accent Trainer for non-native speakers

Project Title

English Accent Trainer for non-native speakers

Project Area of Specialization Artificial IntelligenceProject Summary

English Accent Trainer is an android application that will be used to help specifically those who want to give linguistic exams or are going abroad for a job or some other purposes and generally for all who just want to improve a particular English accent. This app will provide a passage to the user to read it and the app will provide similarity index of the accent of user. This similarity index will provide information about the accent of the user to which native English accent it resembles and how much improvements are required. There will be training exercises for the user to improve his/her accent in a particular native accent.

Project Objectives

The objective we want to achieve is that this app should provide the similarity between the user's accent to any of the native accents and provide training exercises so the user will be able to improve his accent by using this accent.

By this app, we want to achieve the purpose that students can practice for the linguistic exams and can learn the native accents to do jobs abroad.

Project Implementation Method

This project basically works on Natural Language Processing and Machine Learning. NLP is involved in this way that when the user will read the given passage, the app will detect the speech activity and will process the speech signal. In the speech signal processing phase the first task is to remove the noise from the speech signal. For recognition of the accent, we need features which are specific to the accent like (MFCC) so then feature extraction will be performed on the speech signal.

After extracting the specific features then the process of feature selection will be performed which will further refine our set of features. After all, this classification modeling will be performed. There are many classification models that can be used for classification e.g, HMM, RNN and SVM, etc.

After classification then training of the model will be performed so that the similarity index task can achieve its maximum accuracy.

Benefits of the Project

This project will be very useful for students who want to study abroad but their way of speaking English is a big hurdle for them. Instead of paying a lot to coaching classes, they can just download this app and can learn different accents on their own phase. Its benefit is that the user is not time-specific, he is free to use it according to his timetable.

Secondly, there are many people who want to do the job abroad or in an organization where the English accent matters a lot and this is a very big hurdle for them that they can not give additional time to coaching classes from their busy schedule. By the use of this app they can manage their progress report and can easily it in their free time.

Technical Details of Final Deliverable

Technical Details of the deliverable includes that to provide them a good user interface to user so that he can use the app easily we are using the android studio to develope a friendly user interface.

The task to process the speech will be done by Natural Language Processing methodologies.

Feature Extraction task will be done by techniques like MFCC, HMM, etc.

Model Classification algorithms like HMM , SVM, RNN. one of them will be used for classification.

Machine learning will be used to classify accents and to train the model.

Final Deliverable of the Project Software SystemCore Industry EducationOther Industries IT , Media Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Quality Education, Partnerships to achieve the GoalRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	60000
Paid Google Colab for training model	Equipment	1	30000	30000
GPU	Equipment	1	20000	20000
Stationary	Miscellaneous	1	5000	5000
Play Store App Upload	Miscellaneous	1	5000	5000