EKSTRAKSI FITUR BUNYI KETUKAN BUAH KELAPA BERBASIS POWER-NORMALIZED CEPSTRAL COEFFICIENTS (PNCC)
DOI:
https://doi.org/10.21009/03.1201.FA06Abstract
Abstrak
Bunyi ketukan buah kelapa bervariasi bergantung pada tingkat kematangan buah kelapa. Penentuan tingkat kematangan buah kelapa secara manual memiliki beberapa kendala yang perlu diatasi. Proses ini cenderung subjektif dan rentan terhadap ketidakkonsistenan, karena tergantung pada pengalaman dan penilaian individu. Oleh karena itu, penelitian ini bertujuan untuk memaparkan hasil eksperimen ekstraksi ciri bunyi kelapa menggunakan Power-Normalized Cepstral Coefficients (PNCC). Melalui pendekatan ini diperoleh gambaran yang lebih objektif dan komprehensif tentang perbedaan karakteristik bunyi ketukan pada tingkat kematangan yang berbeda. PNCC digunakan untuk mengekstraksi fitur bunyi ketukan buah kelapa yang muda, matang, dan tua. Data yang digunakan dalam penelitian ini diakuisisi menggunakan mikrofon dua arah yang ditempatkan pada kotak tertutup. Fitur diekstrak dengan membagi rekaman bunyi ketukan berdurasi satu detik. Fitur PNCC diekstrak dari bunyi ketukan buah kelapa dan dilakukan reduksi dimensi menggunakan Principal Component Analysis (PCA) dengan tiga jenis kernel, yaitu Linear, RBF, dan Sigmoid. Evaluasi perbedaan karakteristik bunyi dilakukan dengan menghitung Silhouette Score pada setiap kernel PCA. Hasil penelitian menunjukkan bahwa pada kernel Linear, RBF, dan Sigmoid, diperoleh Silhouette Score berturut-turut sebesar 0.205446, 0.179289, dan 0.194963. Temuan ini memberikan pemahaman lebih dalam tentang perbedaan karakteristik bunyi ketukan pada tingkat kematangan buah kelapa yang berbeda dan dapat menjadi dasar untuk pengembangan metode non-invasif yang efisien dalam menentukan tingkat kematangan buah kelapa secara akurat.
Kata-kata kunci: analisis bunyi, sinyal akustik, buah kelapa, PNCC, ekstraksi fitur
Abstract
The tapping sound of a coconut varies depending on the ripeness level of the coconut. Manually determining the ripeness level of a coconut has several obstacles that need to be overcome. This process tends to be subjective and prone to inconsistency, as it depends on individual experience and judgment. Therefore, this study aims to present the experimental results of coconut sound feature extraction using Power-Normalized Cepstral Coefficients (PNCC). Through this approach, a more objective and comprehensive picture of the differences in beat characteristics at different maturity levels is obtained. PNCC was used to extract the beat sound features of young, mature and old coconuts. The data used in this study was acquired using a two-way microphone placed in a closed box. Features were extracted by splitting a one-second recording of the tapping sound. PNCC features were extracted from the coconut tapping sound and dimension reduction was performed using Principal Component Analysis (PCA) with three types of kernels, namely Linear, RBF, and Sigmoid. Evaluation of differences in sound characteristics was carried out by calculating the Silhouette Score on each PCA kernel. The results showed that for Linear, RBF, and Sigmoid kernels, the Silhouette Score was 0.205446, 0.179289, and 0.194963, respectively. These findings provide a deeper understanding of the differences in the characteristics of tapping sounds at different maturity levels of coconut fruits and can be the basis for the development of efficient non-invasive methods to accurately determine the maturity level of coconut fruits.
Keywords: sound analysis, acoustic signal, coconut fruit, PNCC, feature extraction
References
[2] Donis-González et al., “Assessment of chestnut (Castanea spp.) slice quality using color images,” Journal of Food Engineering, vol. 115, no. 3, pp. 407-414, 2013.
[3] Bhargava, Anuja, Atul Bansal, “Fruits and vegetables quality evaluation using computer vision: A review,” Journal of King Saud University-Computer and Information Sciences, vol. 33, no. 3, pp. 243-257, 2021.
[4] Mohammadi et al., Vahid, “Detecting maturity of persimmon fruit based on image processing technique,” Scientia Horticulturae, vol. 184, pp. 123-128, 2015.
[5] Gatchalian et al., “Measurement of young coconut (Cocos nucifera, L.) maturity by sound waves,” Journal of food engineering, vol. 23, no. 3, pp. 253-276, 1994.
[6] Kim, Chanwoo, R. M. Stern, “Power-normalized cepstral coefficients (PNCC) for robust speech recognition,” IEEE/ACM Transactions on audio, speech, and language processing, vol. 24, no. 7, pp. 1315-1329, 2016.
[7] Pearson, Karl, “LIII. On lines and planes of closest fit to systems of points in space,” The London, Edinburgh and Dublin philosophical magazine and journal of science, vol. 2, no. 11, pp. 559-572, 1901.
[8] Rousseeuw, J. Peter, “Silhouettes: a graphical aid to the interpretation and validation of cluster analysis,” Journal of computational and applied mathematics, vol. 20, pp. 53-65, 1987.
[9] De Amorim, Renato Cordeiro, Christian Hennig, “Recovering the number of clusters in data sets with noise features using feature rescaling factors,” Information sciences, vol. 324, pp. 126-145, 2015.
