RT info:eu-repo/semantics/conferenceObject
T1 Speaker's Gender Detection from Glottal Biometry
T2 IV Jornadas de Reconocimiento Biométrico de Personas
A1 Gómez Vilda, Pedro
A1 Fernández Baillo, Roberto
A1 Álvarez Marquina, Agustín
A1 Mazaira Fernández, Luis Miguel
A1 Martínez Olalla, Rafael
A1 Rodellar Biarge, Victoria
AB Through the present work a biometric signature of a speaker’s voiceis proposed for the detection of the speaker’s gender. The estimation methodrelies on the extraction of the glottal flow derivative from voice after removingthe vocal tract transfer function by inverse filtering. This spectral density isrelated to the vocal fold cover biomechanics, and it is well known that certainspeaker’s features as gender, age or pathologic condition are present in it. Forsuch a database of 100 pathology-free speakers equally balanced in gender andage is used as an experimental framework to draft the results exposed in thework. As the estimated biometric parameters show a certain degree of crosscorrelationPrincipal Component Analysis (PCA) is used to reduce parameterdimension. The principal components are used in unsupervised k-meansclustering of speakers (unsupervised gender detection). The outcome groupingshows an almost complete separation of speakers by gender in terms of themost relevant parameters derived from a statistical dispersion study. Possibleapplications of the study can be found in forensic acoustics as well as in speakeridentification and verification tasks.
PB Universidad de Valladolid. Escuela Técnica Superior de Ingeniería Informática
SN 978-84-691-5008-5
YR 2008
FD 2008
LK http://uvadoc.uva.es/handle/10324/26994
UL http://uvadoc.uva.es/handle/10324/26994
LA eng
NO González Ferreras, César, Cardeñoso Payo, Valentín y Vivaracho Pascual, Carlos. IV Jornadas de Reconocimiento Biométrico de Personas. Universidad de Valladolid. E.T.S. de Ingeniería Informática, 2008.
NO Producción Científica
DS UVaDOC
RD 25-abr-2024