Faculty Mentor(s)
Joseph Makarewicz
Project Type
Honors Program project
Scholarship Domain(s)
Scholarship of Discovery
Presentation Type
Presentation
Abstract
A method of vocal signal processing was examined to determine if principal component analysis of spectral data may be used to characterize differences between speakers and if these differences may be used to separate mixtures of vocal signals. Processing was done on a set of voice recordings from 30 different speakers in order to create a projection matrix which could be used by an algorithm to identify the source of an unknown recording from one of the 30 speakers. Two different identification algorithms were tested, both of which were generally unable to correctly identify the source of a single vocal signal. However, one principal component derived from the processing provided a notable distinction between values for male and female speakers. Because of the lack of success in identifying single speakers, the method was unable to be used to separate mixtures of vocal signals. A possible cause of the lack of success could be rooted in the processing methodology’s lack of differentiation between time segments of speech and segments of silence. The incorporation of this distinction into the signal processing method was recommended as a topic for future study.
Permission Type
This work is licensed under a Creative Commons Attribution 4.0 License.
Included in
Vocal Processing with Spectral Analysis
Reed 330
A method of vocal signal processing was examined to determine if principal component analysis of spectral data may be used to characterize differences between speakers and if these differences may be used to separate mixtures of vocal signals. Processing was done on a set of voice recordings from 30 different speakers in order to create a projection matrix which could be used by an algorithm to identify the source of an unknown recording from one of the 30 speakers. Two different identification algorithms were tested, both of which were generally unable to correctly identify the source of a single vocal signal. However, one principal component derived from the processing provided a notable distinction between values for male and female speakers. Because of the lack of success in identifying single speakers, the method was unable to be used to separate mixtures of vocal signals. A possible cause of the lack of success could be rooted in the processing methodology’s lack of differentiation between time segments of speech and segments of silence. The incorporation of this distinction into the signal processing method was recommended as a topic for future study.