The Identification of Infant Cries by Using Codebook as Feature Matching, and MFCC as Feature Extraction
Identifikasi Jenis Tangis Bayi menggunakan Codebook untuk Pengenal Pola dan MFCC untuk Ekstraksi Ciri
Date
2013Author
Renanti, Medhanita Dewi
Buono, Agus
Kusuma, Eng Wisnu Ananta
Metadata
Show full item recordAbstract
In this paper, we focused on automation of Dunstan Baby Language. This software uses MFCC as feature extraction and codebook as feature matching. The codebook of clusters is made from the proceeds of all the baby’s cries data, by using the k-means clustering. The scope of this research are: 1) the infant cries classification used is the version of the Dunstan Baby Language, 2) this software is used to identify the meaning of 0-3 month old infant cries. The methodology of this research consists of several stages of process: data collection, preprocessing, codebook modeling of infant cries, testing and analysis, and interface manufacturing. The data is taken from Dunstan Baby Language videos that has been processed. The data is divided into two, training data and testing data. There are 140 training data, each of which represents the 28 hungry infant cries, 28 sleepy infant cries, 28 wanted to burp infant cries, 28 in pain infant cries, and 28 uncomfortable infant cries (could be because his diaper is wet/too hot/cold air or anything else). The testing data is 35, respectively 7 infant cries for each type of infant cry. Silence cutting is in the preprocessing stage and the feature extraction uses MFCC method. The interface making of the infant cries identification is made based on the training data that produces the highest accuracy. The making of this research is using Matlab R2010b version 7.11.0.584 software. The research varying frame length: 25 ms/frame length = 275, 40 ms/frame length = 440, 60 ms/ frame length = 660; overlap frame: 0%, 25%, 40%; the number of codewords: 1 to 18, except for frame length 275 and overlap frame = 0% using 1 to 29 clusters. The identification of this type of infant cries uses the minimum distance of euclidean and mahalanobis distance. Accuracy value using euclidean distance is between 37% and 94%. Whereas, accuracy value using mahalanobis distance is between 9% and 83%. Codebook model and MFCC with the higher accuracy is: frame length = 440, overlap frame = 0.4, k = 18. Eventhough the distance using that produce the higher accuracy is euclidean distance. That model can produce accuracy recognition of infant cries with the higher about 94%. Sound ‘eh’ is the most familiar, whereas sound ‘owh’ is always missunderstood and generally it is known as ‘neh’ and ‘eairh’. The weakness point of this research is the silence is only be cut at the beginning and at the end of speech signal. Hopefully, in the next research, the silence can be cut in each sound segment so that it can produce more specific sound. It has impact on the bigger accuracy as well.