Multitask Speech Recognition Model

Student: Aleksandr Kutsakov

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Final Grade: 10

Year of Graduation: 2024

Automatic Speech Recognition (ASR) plays a significant role in modern artificial intelligence development. Popularity of such technologies has grown exponentially during last years. Therefore, the research community have faced the challenge of recognition more specific languages. The main problem is an insufficient amount of data for these domains, caused by lower popularity of datasets with new languages. Additionally, while training model for another domain, the first one also must be supported to work with multilingual cases. We propose a multitask training pipeline to tackle both these problems and provide comparable quality for new data. Moreover, we research other regularization approaches to achieve better performance on the small domain.

Full text (added May 19, 2024)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses