Skip to main content

Table 2 Significant works on infant cry reason classification

From: A review of infant cry analysis and classification

First author

Dataset

Features

Classifiers

Best performance

Felipe [43] (2019)

iCOPE (pain vs. no pain)

Mel Scale (MS), MFCC,

SVM

71.68%

  

Constant-Q Chromagram (CQC)

  
  

Local Binary Pattern (LBP),

  
  

Local Phase Quantization (LPQ),

  
  

Robust Local Binary Pattern (RLBP)

  
  

extracted from spectrogram

  

Sharma [39] (2019)

Donate A Cry (hungry, burp needed,

Mean frequency; standard deviation;

K-means clustering,

81.27%

 

belly pain, discomfort, tired, lonely,

median frequency; third quartile range;

hierarchical clustering, Gaussian

 
 

feeling cold/hot, is scared, unidentified)

spectral entropy; kurtosis, skewness,

mixture models clustering

 
  

spectral flatness, etc.

  

Maghfira [36] (2019)

Dunstan Baby Database (pain, hunger,

Spectrogram

CNN-RNN

94.97%

 

discomfort, need to burp, belly pain)

   

Franti [9] (2018)

Dunstan Baby Database (pain, hunger,

Spectrogram

CNN

89%

 

discomfort, need to burp, belly pain)

   

Liu [13] (2018)

NICU recorded (draw attention cry,

LPC, LPCC, MFCC, BFCC

Nearest Neighbor

76.4%

 

diaper change needed cry, and hungry)

 

Artificial Neural Network

 

Turan [41] (2018)

CRIED

Spectrogram

Capsule Network

86.1%

Osmani [67] (2017)

Dunstan Baby Database (hunger,

spectrum, pitch, zero-crossing rate,

SVM, Bagging Decision Tree,

N/A

 

pain, tiredness, belly pain, need burp)

root mean square, intensity, energy along

and Boosted Trees,

 
  

with their calculated statistics (mean,

  
  

variance, skewness, etc)

  

Chang [50] (2016)

Collected from National Taiwan

Spectrogram

CNN

78.5%

 

University Hospital (hungry, pain, and sleep)

   

Bano [20] (2015)

Self-recorded (hungry, need to

Pitch short-time energy MFCC

KNN

86%

 

burp, sleepy, pain, discomfort)

Statistical properties of MFCC

  

Orlandi [21] (2015)

Self-recorded (full term vs. preterm)

CU length, F0 median, F0 mean,

Logistic regression,

87%

  

F0 standard deviation, F0 minimum,

Multilayer perceptron NN,

 
  

F0 maximum, number of estimated

Support Vector Machine,

 
  

F0 values, F123 median,

and Random Forest

 
  

F123 mean, F123 standard deviation,

  
  

F123 minimum, F123 maximum

  

Bhagatpatil [22] (2015)

Self-recorded (pain, hunger, discomfort,

LFCC, MFCC

K-mean clustering, KNN

91.58%

 

need to burp, belly pain)

   

Rosales-Pérez [71] (2014)

Baby Chillanto (hungry vs. pain)

MFCC, LPC

Fuzzy model

97.96%

Yamamoto [23] (2013)

Self-recorded (discomfort, hungry, sleepy)

FFT

Nearest neighbor

62.1%