Speaker recognition with short utterances: A literature analysis

Presenter Information

Darion CarrilFollow

Panel Name

Emerging Technologies in Homeland Security and Emergency Preparedness

Location

Lecture Center Concourse

Start Date

3-5-2019 3:00 PM

End Date

3-5-2019 5:00 PM

Presentation Type

Poster Session

Academic Major

Business

Abstract

Voice-based speaker recognition is currently one of the fastest growing technologies and supports important applications such as smart homes and bio-metric security. To improve efficiency and ease of use, the desire to enable accurate voice recognition methods using short utterances has increased dramatically. Because short utterances provide less data, clustering of speech units obtained from short utterances is often needed to provide high accuracy speaker recognition, which presents additional challenges for data processing. The project will analyze a plethora of related papers on the topic of short utterances in speaker-based voice recognition and conduct a comparative study. More specifically, this project will attempt to aggregate and identify the current state of voice-based speaker recognition using short utterances, how the use of short utterances challenges traditional speaker recognition methods, and future avenues for improvement.

Select Where This Work Originated From

Course assignment/project

First Faculty Advisor

Liyue Fan

First Advisor Email

liyuefan@albany.edu

The work you will be presenting can best be described as

Finished or mostly finished by conference date

This document is currently not available here.

Share

COinS
 
May 3rd, 3:00 PM May 3rd, 5:00 PM

Speaker recognition with short utterances: A literature analysis

Lecture Center Concourse

Voice-based speaker recognition is currently one of the fastest growing technologies and supports important applications such as smart homes and bio-metric security. To improve efficiency and ease of use, the desire to enable accurate voice recognition methods using short utterances has increased dramatically. Because short utterances provide less data, clustering of speech units obtained from short utterances is often needed to provide high accuracy speaker recognition, which presents additional challenges for data processing. The project will analyze a plethora of related papers on the topic of short utterances in speaker-based voice recognition and conduct a comparative study. More specifically, this project will attempt to aggregate and identify the current state of voice-based speaker recognition using short utterances, how the use of short utterances challenges traditional speaker recognition methods, and future avenues for improvement.