DocumentCode
2917361
Title
Automated evaluation of singers´ vibrato through time and frequency analysis of the pitch contour using the DSK6713
Author
Amir, Noam ; Erlich, Tom ; Grabstein, Nitzan ; Fainguelernt, Jacob
Author_Institution
Dept. for Commun. Disorders, Tel Aviv Univ., Tel Aviv, Israel
fYear
2009
fDate
5-7 July 2009
Firstpage
1
Lastpage
5
Abstract
Vibrato is a musical effect produced in singing and in playing of musical instruments, serving to enrich the musical sound. It is defined as a periodic change of the fundamental frequency (the perceptual correlate, "pitch", is often used interchangeably), and is used to add expression and vocal-like qualities to instrumental music. Empirically, vibrato commonly has a periodicity of 4-8 Hz. Attaining good vibrato in the singing voice requires prolonged training by professional singing coaches. Moreover, while prominent singers\´ vibrato is easily distinguishable, this is not the case for beginners. Indeed, the main algorithmic challenge in evaluating singing qualities regards the fact that judging vibrato quality and even presence is rather subjective, and depends to some extent on the listener. This paper presents a tool to analyze the vibrato of singing students, assisting singing practice and self study. Experimental results and a real-time DSP based implementation of the algorithm will be presented.
Keywords
acoustic signal processing; musical instruments; frequency 4 Hz to 8 Hz; musical instrument; musical sound; pitch contour; real-time DSP; singer vibrato; time-frequency analysis; vibrato quality; Acoustic signal processing; Acoustical engineering; Biomedical signal processing; Digital signal processing; Frequency; Instruments; Jacobian matrices; Multiple signal classification; Signal processing algorithms; Vibration measurement; Music; Pattern Analysis; Real-Time; Signal Processing;
fLanguage
English
Publisher
ieee
Conference_Titel
Digital Signal Processing, 2009 16th International Conference on
Conference_Location
Santorini-Hellas
Print_ISBN
978-1-4244-3297-4
Electronic_ISBN
978-1-4244-3298-1
Type
conf
DOI
10.1109/ICDSP.2009.5201121
Filename
5201121
Link To Document