This shows you the differences between two versions of the page.

Link to this comparison view

mathematical_analysis_of_different_natural_voice_types [2014/04/01 20:36] (current)
bziolko created
Line 1: Line 1:
 +Felix De Jong 1,2, H. Lycke 3, W. Decoster 3, A. Ivanova 4, M.M. Van Hulle 5 (Belgium & The Netherlands)
 +**Mathematical analysis of different natural voice types**
 +1. Department of ENT – Head and Neck Surgery, University Hospitals KU Leuven, Leuven, Belgium.\\
 +2. Department of ENT, Bernhoven Hospital, Uden, The Netherlands\\
 +3. Research Group ExpORL, Department of Neuroscience, KU Leuven, Leuven, Belgium\\
 +4. Leuven Statistics Research Centre, KU Leuven, Leuven, Belgium\\
 +5. Laboratory of Neuro- and Psychophysiology, Department of Neuroscience KU Leuven, Leuven, Belgium\\
 +The aim of this study was to verify the existence of Voice Range Profile parameters with which the data can be partitioned into a number of clearly separated clusters. A data-driven approach was adopted as it imposes minimal assumptions on the nature of the data, what elements to use for its analysis, and even the very existence of voice clusters. The data from 206 female conservatory singing students (18 - 25 years) and 256 male subjects (18 – 52 years), consisting of 9  young singing students, 17 professional singers, 61 professional choir singers and 169 with and without singing experience, was investigated. Parameters derived from the geometry of the VRP; the register transition zone; the geometry of the chest/head voice parts and the linear characteristics of the minimum and maximum intensity curves were used. Additionally, a number of voice frequency and intensity ratios and differences were defined based on some of the above parameters. For the clustering analysis Ward’s minimum method and K-means clustering were used. For both the male and female voices Ward’s procedure indicated that there could be three or four clusters in the data. Based on the migration index the three-cluster solution turned out to be the most consistent one. For the male voices the frequency of the register dip was the parameter that led to the best three-cluster separation. For the female voices this was the ratio of the perimeter length of the chest voice part of the voice range profile versus the total perimeter length. This study demonstrates that parameter combinations of the VRP exist that generate a clear separation of voice clusters. A second salient result of this study is the finding that each of these features has to do with register transition, which is an important aspect in voice classification in singing practise. 
Copyright © XXII PVC Organizing Committee 2013. All Rights Reserved.