The crying of infants aged 0-3 months can be classified according to their needs, as identified by Dunstan Baby Language, which consists of specific sounds denoting different needs. These sounds include "eairh" for discomfort caused by fart, "neh" indicating hunger, "heh" representing general discomfort, "owh" signaling tiredness or sleepiness, and "eh" expressing the need to burp. The baby crying sound data was obtained from the Dunstan Baby Language (DBL) database, which includes educational videos about infants and a collection of babies crying sounds. These sounds were converted into *.wav audio format and divided into 5-second segments. A total of 188 audio data segments were collected. The research employed the Artificial Neural Network (ANN) classification method and the Mel-Frequency Cepstral Coefficient (MFCC) feature extraction method. The collected data underwent feature extraction, aiming to identify distinctive characteristics using the librosa library in the Python programming language. This process allowed us to obtain specific information from the acquired sound data. The results of this study achieved an accuracy level of 90%. This research contributes to the understanding and classification of infant crying based on the Dunstan Baby Language, offering insights into their various needs. The implementation of ANN and MFCC techniques showcases the effectiveness of this approach in accurately classifying infant cries and provides a foundation for further research in the field of infant communication.