Decoding baby talk: A novel approach for normal infant cry signal classification

This paper describes a novel approach to identify a baby physiological state and its needs. In this work normal infant cry signal of ages 1day to six months old is used. In particular there are fixed cry attributes for a healthy infant cry, which can be classified into five groups such as: Neh, Eh, Owh, Eairh and Heh. The infant cry signal is segmented by using Pitch frequency and the features like Short-time energy, Harmonicity Factor (HF) and Harmonic-to-Average Power (HAPR) are extracted and MFC (mel-frequency cepstrum) coefficients is computed over MATLAB. KNN classifier using Pitch, Short-time energy, Harmonicity Factor (HF) and Harmonic-to-Average Power (HAPR) are used to classify the normal infant cry signal. Percentages of results obtained are Neh 80%, Eh 90%, Owh 80%, Eairh 90%, and Heh 90% respectively. Decoding baby talk supports the mother´s built-in intuition about knowing and responding to their baby´s needs, and physician to treat infant early.