Automatic classification of pronunciation errors

<div>With increasing globalization need for learning foreign languages grows rapidly too.&nbsp;</div><div>Study of any foreign language is closely related to the interaction of the student and the teacher,&nbsp;</div><div>because of without assessing, especially setting the correct pronunciation, training does not&nbsp;</div><div>acquire the necessary efficiency. For various reasons, many students do not have the opportunity&nbsp;</div><div>to engage with the teacher. By following needs of modern society, in the last decade has been&nbsp;</div><div>developing various direction such as automatic teaching pronunciation, uniting researchers of&nbsp;</div><div>different disciplines: linguistics, psychology, education, speech recognition, etc.</div><div>The aim of this study is the application and improvement of methods for automatic&nbsp;</div><div>evaluation of pronunciation for studying English (L2) and the development of methods for&nbsp;</div><div>assessing the quality of the content, depending on the native language (L1) of a learner.</div><div>As a framework for speech recognition Sphinx 4 will be used.</div><div>As the basic methods for automatic evaluation of pronunciation used statistical&nbsp;</div><div>techniques and methods of machine learning, this requires a training sample. As such sampling&nbsp;</div><div>at the first stage is used a sample with a limited vocabulary which is applied for testing and&nbsp;</div><div>prototyping training system. To expand of training sample website provides an interface for&nbsp;</div><div>adding new pronunciation examples by users and ability for experts to annotate these examples.&nbsp;</div><div>Annotation is done with separation different types of errors on the same segment of speech for&nbsp;</div><div>the purpose of obtaining the best possible correction informative evaluation of pronunciation.&nbsp;</div><div>To improve the material supply training workouts for specific problem sounds are available.&nbsp;</div><div>When implementing an application considered by previous developments such systems in order&nbsp;</div><div>to optimize the learning process and user interaction with the system applied to the interface&nbsp;</div><div>structure and content of the training lessons.</div>