|
| Ref. No. | Dataset | Methodology | Research challenges |
|
| [16] | DA-IICT infant cry and baby Chillanto | Convolutional restricted Boltzmann machine (ConvRBM) model | (i) Model was not implemented in any real-time environment
(ii) Model was not evaluated against any other model apart from MFCC |
| [17] | Self-recorded datasets, baby Chillanto, National Taiwan University Hospital, Dunstan baby, iCOPE, University of Milano Bicocca | MFCC, KNN, SVM, GMM, CNN and RNN | (i) Scalability of the dataset
(ii) Unavailability of skilled labors to collect data
(iiii) Lack of collaboration between medical professionals and researchers |
| [18] | Sainte-Justine Hospital (Montreal, Canada), Al-Sahel Hospital (Lebanon), Al-Raee Hospital (Lebanon) | DFNN | (i) Exclusion of various topologies and transfer functions
(ii) exclusion of deep features and nonlinear statistical features |
| [19] | Audio recordings from free sound, BigSoundBank, sound archive, ZapSplat, SoundBible and sound jay | CNN | (i) Model was not evaluated against any other models |
| [20] | Self-recorded audio recordings | SVC and RBF | (i) Exclusion of more features or categories in the dataset |
| [21] | Donate-a-cry corpus | SVM, random forest, logistic regression, KNN | (i) Exclusion of a more extensive dataset for enhanced justification of the model |
| [22] | Datasets from various online resources with infant cry clips | Random forest classification model | (i) Use of smart phones other than Motorola G5/G6 not included for evaluation |
| [23] | Donate-a-cry corpus | KNN and SVM | (i) Exclusion of an extensive dataset with more infant cry categories |
|
