A Low-Cost Intelligent Hardware System for Real-Time Infant Cry Detection and Classification
Pradeep Pathirana1, Sagara Sumathipala2
1Pradeep Pathirana, Department of Computational Mathematics, Faculty of Information Technology, University of Moratuwa, Sri Lanka.
2Sagara Sumathipala, Department of Computational Mathematics, Faculty of Information Technology, University of Moratuwa, Sri Lanka.
Manuscript received on 06 December 2019 | Revised Manuscript received on 20 December 2019 | Manuscript Published on 31 December 2019 | PP: 18-22 | Volume-8 Issue-12S2 October 2019 | Retrieval Number: L100410812S219/2019©BEIESP | DOI: 10.35940/ijitee.L1004.10812S219
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Cry of an infant serves as the main communication language to seek the attention of their caretakers. Acoustic characteristics of cries provide insights into the physiological and psychological states of the infants. To understand the reason behind the cries, caretakers pay attention to acoustic characteristics like tone, pitch, and loudness, etc. Infant cry classification has gained importance in both commercial and medical fields due to its applications such as baby monitoring and non-invasive diagnosis of health conditions of newborns in early stages. This paper discusses the implementation of a low-cost hardware device for real-time cry classification. Further, this presents the use of time and frequency domain features to detect cry words and identify the reasons. The proposed solution covers the use of classification techniques like artificial neural networks and k nearest neighbors to gain accuracy figures over 90% and 70% for cry detection and classification respectively while maintaining the resource utilization at a minimum level to keep hardware solution simple and low cost.
Keywords: Cry classification, Voice Activity Detection, Short-Time Energy, Short-Time Zero-Crossings, Mel-frequency Cepstral Coefficients, Artificial Neural Networks, K-Nearest Neighbors.
Scope of the Article: Real-Time Information Systems