Karateristik Akustik Vokalisasi Sapi Bali Betina
Heny Vensye Saiya, Prof. Dr. Ir. Sigit Bintara, M.Si, IPU., ASEAN Eng. ; Susilohadi, S.Si., M.Si., Ph.D. ; Prof. Ir. Prof. Ir. Diah Tri Widayati, M.P., Ph.D., IPU. ASEAN Eng.
2026 | Disertasi | S3 Ilmu Peternakan
Vokalisasi merupakan sinyal komunikasi penting yang dapat merefleksikan kebutuhan, kondisi fisiologis, dan respons perilaku ternak. Penelitian ini bertujuan mengidentifikasi karakteristik akustik vokalisasi sapi bali pada kondisi pascapemisahan dari pedet, antisipasi dan penundaan akses makan, serta kondisi normal dan estrus. Penelitian dilaksanakan dalam tiga tahap. Tahap pertama menganalisis 21 parameter akustik menggunakan pustaka Parselmouth pada Python, analisis matriks korelasi, Principal Component Analysis (PCA), dan General Linear Model (GLM) untuk mengetahui menganalisis karakteristik akustik dan mengidentifikasi pola vokalisasi sapi bali pascapemisahan dari pedet. Hasilnya menunjukkan bahwa karakteristik fitur akustik sapi bali pascapemisahan dari pedet menunjukkan 3 pola vokal yaitu vokal tinggi, vokal dalam dan vokal rendah. Ketiga pola vokal dapat dibedakan dengan fitur akustik. Fitur akustik duration, f0 mean, f0 extent dan jitter secara praktis sudah cukup memisahkan ketiga pola suara sapi bali. Tahap kedua untuk mengidentifikasi, membandingkan dan menentukan karakteristik akustik vokalisasi sapi bali selama kondisi antisipasi makan sebagai representasi valensi positif dan valensi negatif menggunakan analisis ANOVA, uji Tukey, dan Decision Tree Classification. Durasi dan rasio voicing menjadi karakteristik vokal pada kondisi normal dan antisipasi makan, sedangkan laju amplitudo, frekuensi minimum, dan formant 2 mencirikan vokal selama penundaan akses makan. Fitur yang paling berkontribusi terhadap klasifikasi adalah formant 1, formant 3, formant 2, dan intensitas. Tahap ketiga mengidentifikasi dan menentukan alterasi fitur akustik vokal normal dan vokal estrus sapi bali melalui uji statistik, koreksi Bonferroni, ukuran efek, serta Linear Discriminant Analysis (LDA). Fitur akustik yang mengalami perubahan antara vokalisasi normal dan vokalisasi estrus pada sapi bali meliputi jitter, shimmer, intensity, dan durasi. Vokalisasi estrus dicirikan oleh peningkatan nilai jitter dan shimmer, serta penurunan intensitas dan durasi dibandingkan dengan vokalisasi normal. Secara keseluruhan, vokalisasi sapi bali menyimpan informasi akustik yang berpotensi digunakan sebagai penanda noninvasif kondisi fisiologis dan perilaku ternak.
Vocalization is an important communication signal that can reflect the needs, physiological state, and behavioral responses of livestock. This study aimed to identify the acoustic characteristics of bali cattle vocalizations under several conditions: post-separation from calves, anticipation and delay of feed access, and normal versus estrus states. The research was conducted in three stages. The first stage analyzed 21 acoustic parameters using the Parselmouth library in Python, correlation matrix analysis, Principal Component Analysis (PCA), and a General Linear Model (GLM) to characterize acoustic features and identify vocalization patterns following calf separation. The results revealed three distinct vocal patterns: high-pitched, deep, and low-pitched calls, which could be differentiated based on acoustic features, with duration, mean f0, f0 extent, and jitter being practically sufficient to separate the three vocal patterns. The second stage identified, compared, and determined the acoustic characteristics of vocalizations during feed anticipation (representing positive valence) and feed-access delay (representing negative valence), using ANOVA, Tukey's test, and Decision Tree Classification. Duration and voicing ratio characterized vocalizations under normal and feed-anticipation conditions, whereas amplitude rate, minimum frequency, and formant 2 characterized vocalizations during feed-access delay, with formant 1, formant 3, formant 2, and intensity contributing most to classification accuracy. The third stage identified and determined alterations in acoustic features between normal and estrus vocalizations through statistical testing, Bonferroni correction, effect size analysis, and Linear Discriminant Analysis (LDA). The acoustic features that differed significantly between normal and estrus vocalizations were jitter, shimmer, intensity, and duration, with estrus vocalizations characterized by increased jitter and shimmer alongside decreased intensity and duration relative to normal vocalizations. Collectively, these findings demonstrate that bali cattle vocalizations encode acoustic information with strong potential as a non-invasive indicator of physiological state and behavioral response in livestock management.
Vocalization is an important communication signal that can reflect the needs, physiological state, and behavioral responses of livestock. This study aimed to identify the acoustic characteristics of bali cattle vocalizations under several conditions: post-separation from calves, anticipation and delay of feed access, and normal versus estrus states. The research was conducted in three stages. The first stage analyzed 21 acoustic parameters using the Parselmouth library in Python, correlation matrix analysis, Principal Component Analysis (PCA), and a General Linear Model (GLM) to characterize acoustic features and identify vocalization patterns following calf separation. The results revealed three distinct vocal patterns: high-pitched, deep, and low-pitched calls, which could be differentiated based on acoustic features, with duration, mean f0, f0 extent, and jitter being practically sufficient to separate the three vocal patterns. The second stage identified, compared, and determined the acoustic characteristics of vocalizations during feed anticipation (representing positive valence) and feed-access delay (representing negative valence), using ANOVA, Tukey's test, and Decision Tree Classification. Duration and voicing ratio characterized vocalizations under normal and feed-anticipation conditions, whereas amplitude rate, minimum frequency, and formant 2 characterized vocalizations during feed-access delay, with formant 1, formant 3, formant 2, and intensity contributing most to classification accuracy. The third stage identified and determined alterations in acoustic features between normal and estrus vocalizations through statistical testing, Bonferroni correction, effect size analysis, and Linear Discriminant Analysis (LDA). The acoustic features that differed significantly between normal and estrus vocalizations were jitter, shimmer, intensity, and duration, with estrus vocalizations characterized by increased jitter and shimmer alongside decreased intensity and duration relative to normal vocalizations. Collectively, these findings demonstrate that bali cattle vocalizations encode acoustic information with strong potential as a non-invasive indicator of physiological state and behavioral response in livestock management.
Kata Kunci : Acoustic features, animal welfare, animal behaviour, vocal communication, emotional expression