Portal de Periódicos da CAPES

Improving Baby Caring with Automatic Infant Cry Recognition

2006; Springer Science+Business Media; Linguagem: Inglês

10.1007/11788713_101

1611-3349

Sandra E. Barajas-Montiel, Carlos A. Reyes-García, Emilio Arch-Tirado, Mario Mandujano,

Advanced Chemical Sensor Technologies

Babies are human beings who cannot satisfy their necessities by themselves, they completely depend of cares and attentions by adults. The cry is the natural media babies use to express their needs. Several studies have demonstrated that cry is a useful tool to determine the different emotional and physiological states from an infant, and in addition to make medical diagnoses of diseases related to the central nervous system. This work presents the analysis and extraction of characteristics from infant crying for its automatic classification with Support Vector Machines. Several classification tasks were done, working in the identification of pain, hunger, and deafness levels with results of up to 96 % of correct classification. Besides some results, we show the implementation and experimentation done.