Search

• Monitor bioacoustics to detect respiratory problems. • Algorithm to detect sneezing in broiler chickens. • Classification of unbalanced dataset. • The algorithm obtained a sensitivity of 66.7% and a precision of 88.4%. Respiratory diseases are a major health challenge in meat chicken production. As sneezing is a clinical sign of many respiratory diseases, sound has a great potential in monitoring these diseases. This study focussed on the development of an algorithm to monitor chicken sneezing sounds in a situation where multiple birds are active and multiple noise sources are present. An experiment was designed where the sneezing from within a group of 51 chickens was recorded. 763 sneezes were annotated out of 480 min of sound recordings. First, the number of labelled sneezes of adequate quality were investigated. Then raw sound signal was filtered using spectral subtraction and split into short intervals with elevated energy that could be sneezes. This led to a highly unbalanced dataset containing only 0.24% sneezes, from which features characterising the sneezing sounds were calculated. These were then grouped into 8 different features on which the algorithm classified the sound as sneeze or no-sneeze with a sensitivity of 66.7% and a precision of 88.4%. The algorithm enabled the monitoring of the number of sneezes in the experimental group. This work represents the first step towards the development of an automated sound-based monitoring system for poultry health. [ABSTRACT FROM AUTHOR]

Highlights • Queenless colonies can generate slightly different sound patterns. • Bee sound patterns can be affected by biological conditions. • A bee sound classification model must include data of varied colonies to reliable. • The proposal methodology can be applied in the analysis of more states in bee colonies. Abstract This study presents an analysis of a multiclass classification problem to identify queenless states by monitoring bee sound in two possible cases; a strong and healthy colony that lost its queen and a reduced population queenless colony. The sound patterns were compared with patterns of healthy queenright colonies. Five colonies of Carniola honey bee were monitored by using a system based on a Raspberry Pi 2 and omnidirectional microphones placed inside the hives. Feature extraction was carried out by Mel Frequency Cepstral Coefficients (MFCCs) method. A multiclass model with three outcome variables was constructed. For feature selection and regularization, a Lasso logistic Regression model was used along with one vs all strategy. To provide visual evidence and examine the results, data was analyzed by scatter plots of Singular Value Decomposition (SVD). The results show that is possible to detect the queenless state in both cases. Queenless or healthy colonies can generate slightly different patterns and the data clusters of the same condition tend to be close. The proposed methodology can be applied for the analysis of more conditions in bee colonies. [ABSTRACT FROM AUTHOR]

Bovine respiratory disease (BRD) complex in calves impairs health and welfare and causes severe economic losses for the Stockperson. Early recognition of BRD should lead to earlier veterinary (antibiotic/anti-inflammatory) treatment interventions thereby reducing the severity of the disease and associated costs. Coughing is one of the clinical manifestations of BRD. It is believed that by automatically and continuously monitoring the sounds within calf houses, and analysing the coughing frequency, early recognition of BRD in calves is possible. Therefore, the objective of the present study was to develop an automated calf cough monitor and examine its potential as an early warning system for BRD in artificially reared dairy calves. The coughing sounds of 62 calves were continuously recorded by a microphone over a three-month period. A sound analysis algorithm was developed to distinguish calf coughs from other sounds (e.g. mechanical sounds). During the sound recording period the health of the calves was assessed and scored periodically per week by a trained human observer. Calves presenting with BRD received antibiotic and/or anti-inflammatory treatment and the dates of treatment were recorded. This treatment date reference served as a comparison for the investigation of whether an increase in coughing frequency could be related to calves developing BRD. The calf cough detection algorithm achieved 50.3% sensitivity, 99.2% specificity and 87.5% precision. Four out of five periods, where coughing frequency was observed to be increased, coincided with the development of BRD in more than one calf. This period of increased coughing frequency was always observed before the calves were treated. Therefore, the calf cough monitor has the potential to identify early onset of BRD in calves. [ABSTRACT FROM AUTHOR]

In commercial poultry production there are a number of diseases which are of particular importance due to the heavy economic losses that can arise if a flock becomes infected. The development of an automated and rapid disease detection system would therefore be of considerable benefit to both production and animal welfare. This study represents an intelligence device for diagnosing avian diseases by using Data-mining methods and Dempster-Shafer evidence theory (D-S). 14-day-old chickens were divided into four groups. Each group was deliberately infected with a disease: Newcastle Disease (ND), Bronchitis Virus (BV), Avian Influenenza (AI), and the last group was considered as control samples. Fast Fourier Transform (FFT) and Discrete Wavelet Transform (DWT) were used to process the chicken’s sound signals in frequency and time-frequency domains, respectively. In order to achieve information, 25 statistical features from frequency domains, and 75 statistical features from time-frequency domains were extracted. During dimensionality reduction stage, the best features of the sound signals were selected, using improved distance evaluation (IDE) method. The chicken’s sound signals were analyzed in two consecutive days after virus infection. Support vector machine (SVM) was used as the classifier in this study. The first classification was done with SVM and based on sound features in frequency and time-frequency domains with accuracy of 41.35 and 83.33%, respectively. The accuracy of the method based on D-S infusion of sound data reached 91.15%. The developed model based on achievement result could diagnose Newcastle Disease, Bronchitis Virus and Avian Influenza from sound signals. [ABSTRACT FROM AUTHOR]

In modern farming there is a growing demand for innovative tools gathering and analysing information concerning the herd, as well as individual animals. In Precision Livestock Farming (PLF), technology continuously measures various variables as activity, food intake or oestrus activity, thereby supporting farmers in monitoring his livestock. Sound analysis has shown to be useful as an early warning tool in pigs and it is unknown whether sound analysis can also be applied in cattle. Goal of this research was to determine whether a correlation can be found between cattle vocalisation and cattle behaviour. The vocalisations and behaviour of Holstein Friesian cattle were observed using audio and video recordings. Four cameras and four microphones were installed at a high production dairy farm in Herwijnen, the Netherlands. Three sets (a set consisting of both a camera and a microphone) recorded dairy cattle between two and fourteen years of age, one set recorded heifers between four and ten months of age. Recordings were made for fifteen days in three consecutive weeks, ten hours per day. Calls of cattle were traced to an individual cow and, if possible, linked with simultaneously expressed behaviour. The used ethogram consisted of six behavioural groups: lying & ruminating, feeding related behaviour, social interaction, sexual related behaviour, stress related behaviour and remaining behaviour. Lying & ruminating was a separate class since this behaviour expresses the needs of a cow. The maximum frequency in Hertz (Hz) of each call was determined. Statistical analysis showed a significant difference between the mean maximum frequency (Hz) of calls during lying & ruminating and calls recorded during other behaviours (83 ± 4.3 Hz versus 298 ± 8.0 Hz; p < 0.05). Calls by adult dairy cattle had a significantly lower maximum frequency (Hz) than calls by heifers. (332.6 ± 0.2 Hz versus 218.5 Hz ± 0.3 Hz; p < 0.05). This study may provide a foothold towards the use of sound analysis as a tool for dairy cattle management. If calls by cattle can be used to monitor welfare, dairy farmers can be alerted when cattle welfare is decreasing. [ABSTRACT FROM AUTHOR]

This paper extends existing research on the feed intake of broiler chickens and describes an advanced monitoring system to accurately measure the feed intake of broilers at group level by a real-time sound processing technology. In this research, the pecking sounds of 10 male, 39-day-old, broiler chickens were recorded by a microphone that was attached around the feeder. At the same time, the appearance of chickens around feeder was recorded by a camera that was positioned on the top of the feeding pen. Simultaneously, a weighing system was used to automatically record feed uptake of broilers as a reference method. An existing algorithm was further developed to detect the pecking sounds of 10 broiler chickens while the birds were all eating together. The feed intake of broiler chickens was obtained by pecking sound analysis. The results of the algorithm were compared to the reference of the feed uptake values recorded by the weighing system. The relationship between feed intake obtained from the algorithm and feed intake recorded by a weighing scale was investigated and a strong relation was found between these two variables. The Pearson Product Moment Correlation Coefficient test (PPMCC) was performed to define the correlation between these two variables which resulted in R 2 = 0.994. In addition to this high relationship, 86% of feed intake was correctly monitored using sound analysis. Although the accuracy of the proposed system remained at 86%, the correlation between the feed intake obtained from the algorithm and feed intake recorded by a weighing scale was very high ( R 2 = 0.994), the results suggest that this continuous pecking sound detection system has the potential to be used as a tool to assess the feed intake and feeding behaviour of multiple chickens around a feeder. The advantage of this system is that measurements can be made continuously throughout the life span of a flock, in a fully automated, completely non-invasive way. [ABSTRACT FROM AUTHOR]

Highlights: [•] An automatic sound detection system that uses microphone attached in feeder is proposed. [•] Algorithm to detect the pecking sounds of broiler chickens is presented. [•] Results of the algorithm are compared to reference methods. [•] The correlation between number of peckings and feed intake was in R 2 =0.985. [•] Pecking sound detection system has potential to be used as a tool to monitor feed intake of chickens. [Copyright &y& Elsevier]

Abstract: The objective of this paper is to develop an algorithm that could be used in order to reduce the spread of chicken hatching in industrial incubators for chicken eggs. The approach that is used is based on frequency analysis of sounds recorded inside the industrial incubator and aims at identifying the time at which all the eggs inside the incubator have reached the internal pipping stage. The developed algorithm is able to be calibrated automatically in order to adjust for sounds around the incubator and the acoustics of every incubator. The algorithm has been implemented in a Digital Signal Processor and applied in real-time in an industrial environment. It is shown that the algorithm can correctly identify the time at which 93–98% of the eggs have had been in the internal pipping stage. This level of accuracy is considered adequate for a practical application focusing on reduction of the hatching window. [Copyright &y& Elsevier]

Abstract: The objective of this paper is to assess if the dynamics in the energy envelope of pig cough sounds are related to pathological conditions of the respiratory system. Two groups of pigs are compared. The first group, the sick pigs, is suffering from pneumonia by infection of Pasteurella Multocida. The second group, the control group, consists of healthy pigs which produced induced coughs by nebulisation of citric acid. The cough sounds of both groups were used to calculate the energy envelope, after which two signals are derived for further modelling. The first signal is an artificial step input, the second signal is part of the energy envelope of the cough signal that starts at the maximum level and decays in time. Using an autoregressive model estimation technique, the decay of the energy envelope is modelled as an input–output system. Based on the Young Identification Criterion (YIC) and R 2, the optimal model is proven to be a first order model with a first order denominator. Using this first order transfer function structure to model all cough sounds, the time constant of the simulated output is estimated based on the model parameters. The time constant shows significant higher values for the decay of the cough signals from pigs which are infected with Pasteurella Multocida compared to non-infected pigs (P <0.0001). The results are in accordance with previous work in which pathological conditions have shown to influence cough sound duration. This modelling technique gives more insight in the effect of changes in lung condition on cough sound generation and might be incorporated in techniques for automatic cough evaluation. [Copyright &y& Elsevier]

Abstract: In the course of evolution, nature evolved manifold means of communication. Sound is one of the most important means to convey information and to express emotional states and conditions. The acoustic monitoring of farm animals may serve as an efficient management tool to enhance animal health, welfare, and farm efficiency. The final goal of this work is a Call-Recogniser, a device that identifies the meaning of vocal utterances of cows and presents the meaning to the farmer. Such a call-recogniser must be able to recognise the meaning of species-specific calls, independent from the individual animal and the probably more or less noisy environment. As in speech recognition, the call recognition of animals can be regarded as a statistical paradigm. During the learning or training phase, feature vectors from known calls are calculated. From the feature vectors of calls with the same meaning, reference patterns are built and stored. For recognition, the feature vectors from an unknown call are calculated in the same way, and the system then determines the reference pattern that is most similar to the feature vector to be recognised, and outputs its meaning. Despite the vocabulary size and complexity of human speech, which is unique in the animal realm, sound production and reception in vertebrates have much in common. This encourages, adaptation of methods and experiences from speech recognition to recognise animal calls. The problem of animal independent call recognition is comparable to speaker independent word spotting in speech recognition. In speech recognition, double stochastic processes, such as hidden Markov models (HMMs), have proved very efficient. They are applied here to recognise animal calls, using utterances of cows as an example. The results reveal that HMMs are well suited for animal call recognition. [Copyright &y& Elsevier]