Fifth International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2007)

Florence, Italy
December 13-15, 2007

Towards a Cry Classification Based on Articulated Signal Processing

Sergio Daniel Cano Ortiz (1), Israel Suaste (2), Daniel Escobedo (1), Taco Ekkel (3), Carlos Alberto Reyes García (2)

(1) CENPIS, University of Oriente, Santiago de Cuba, Cuba
(2) INAOE, Carretera de Cholula s-n, Puebla, Mexico,
(3) Faculty of Informatics, University of Twente, The Netherlands

The paper assumes the implementation of a cry-based classifier for neonatal diagnosis. The main contribution is concerned with the articulated processing of cry signals, which includes two kinds of approaches: a threshold-based classification and ANN-based classification. Every one of those approaches makes its own contributions to the cry classification, both are adequately combined in a classifier of two-class (pathological and normal). Moreover the use of cry unit as a primary data was also an interesting aspect held by the authors. This articulated cry processing is the main body of a new cry-based methodology for neonatal diagnosis, which will be presented in a few months by the Group of Speech Processing in Cuba.
Index Terms. cry analysis, neural networks

Full Paper (reprinted with permission from Firenze University Press)

Bibliographic reference.  Cano Ortiz, Sergio Daniel / Suaste, Israel / Escobedo, Daniel / Ekkel, Taco / Reyes García, Carlos Alberto (2007): "Towards a cry classification based on articulated signal processing", In MAVEBA-2007, 219-222.