Design of Text Independent Automatic Speaker Recognition Systems

Student: Drozdova Kseniia

Supervisor: Anastasiya A. Bonch-Osmolovskaya

Educational Programme: Language Theory and Computational Linguistics (Master)

Final Grade: 10

Year of Graduation: 2017

This paper presents an overview of a state-of-the-art text-independent speaker verification and identification systems. The work contains description of all stages of building such systems: feature extraction, building a voice model, final decision rules. All sections contain a theoretical description of the methods and ways of implementing them in Python. We present four voice source modelling techniques: Gaussian mixture models (GMM) approach, GMM with the Universal Background Model (UBM), Support Vector Machine (SVM) and stacking algorithms GMM and SVM. The performance of all algorithms is evaluated using a large telephone-speech database. Best quality for verification system is achieved by the SVM and is equal to 96.6 %. The best identification results obtained using GMM and is equal to 95.8%.

Full text (added June 1, 2017)

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