AN EXECUTABLE METHOD FOR AN INTELLIGENT SPEECH AND CALL RECOGNITION SYSTEM USING A MACHINE LEARNING-BASED APPROACH

This paper describes a novel call recognizer system based on the machine learning approach. Current trends, intelligence, emotional recognition and other factors are important challenges in the real world. The proposed system provides robustness with high accuracy and adequate response time for the human–computer interaction. Intelligence and emotion recognition from the speech of human–computer interfaces are simulated via multiple classifier systems (MCSs). At a higher-level stage, the acoustic stream phase extracts certain acoustic features based on the pitch and energy of the signal. Here, the feature space is labeled with various emotional types in the training phase. Emotional categories are trained in the acoustic feature space. The semantic stream process converts speech into text in the input speech signal. Text classification algorithms are applied subsequently. The clustering and classification process is performed via a [Formula: see text]-means algorithm. The detection of the Tone of Voice of call recognition system is achieved with the XGBoost model for feature extraction and detection of a particular phrase in the client call phase. Speech expressions are used for understanding the human emotion. The algorithms are tested and demonstrate good performance in the simulation environment.

Precision as a metric for acoustic survey design using occupancy or spatial capture-recapture

Passive acoustic surveys provide a convenient and cost-effective way to monitor animal populations. Methods for conducting and analysing such surveys, especially for performing automated call recognition from sound recordings, are undergoing rapid development. However, no standard metric exists to evaluate the proposed changes. Furthermore, most metrics that are currently used are specific to a single stage of the survey workflow, and therefore may not reflect the overall effects of a design choice.Here, we attempt to define and evaluate the effectiveness of surveys conducted in two common frameworks of population inference – occupancy modelling and spatially explicit capture-recapture (SCR). Specifically, we investigate precision (standard error of the final estimate) as a possible metric of survey performance, but we show that it does not lead to generally optimal designs in occupancy modelling. In contrast, precision of the SCR density estimate can be optimised with fewer experiment-specific parameters. We illustrate these issues using simulations.We further demonstrate how SCR precision can be used to evaluate design choices on a field survey of little spotted kiwi (Apteryx owenii). We show that precision correctly measures tradeoffs involving sampling effort. As a case study, we compare automated call recognition software with human annotations. The proposed metric captured the tradeoff between missed calls (8% loss of precision when using the software) and faster data through-put (60% gain), while common metrics based on per-second agreement failed to identify optimal improvements and could be inflated by deleting data.Due to the flexibility of SCR framework, the approach presented here can be applied to a wide range of different survey designs. As the precision is directly related to the power of detecting temporal trends or other effects in the subsequent inference, this metric evaluates design choices at the application level, and can capture tradeoffs that are missed by stage-specific metrics, thus enabling reliable comparison between different experimental designs and analysis methods.

Automated Fraudulent Phone Call Recognition through Deep Learning

Several studies have shown that the phone number and call behavior generated by a phone call reveal the type of phone call. By analyzing the phone number rules and call behavior patterns, we can recognize the fraudulent phone call. The success of this recognition heavily depends on the particular set of features that are used to construct the classifier. Since these features are human-labor engineered, any change introduced to the telephone fraud can render these carefully constructed features ineffective. In this paper, we show that we can automate the feature engineering process and, thus, automatically recognize the fraudulent phone call by applying our proposed novel approach based on deep learning. We design and construct a new classifier based on Call Detail Records (CDR) for fraudulent phone call recognition and find that the performance achieved by our deep learning-based approach outperforms competing methods. Experimental results demonstrate the effectiveness of the proposed approach. Specifically, in our accuracy evaluation, the obtained accuracy exceeds 99%, and the most performant deep learning model is 4.7% more accurate than the state-of-the-art recognition model on average. Furthermore, we show that our deep learning approach is very stable in real-world environments, and the implicit features automatically learned by our approach are far more resilient to dynamic changes of a fraudulent phone number and its call behavior over time. We conclude that the ability to automatically construct the most relevant phone number features and call behavior features and perform accurate fraudulent phone call recognition makes our deep learning-based approach a precise, efficient, and robust technique for fraudulent phone call recognition.

Heterospecific alarm-call recognition in two warbler hosts of common cuckoos

Species facing similar selection pressures should recognize heterospecific alarm signals. However, no study has so far examined heterospecific alarm-call recognition in response to parasitism by cuckoos. In this study, we tested whether two sympatric host species of the common cuckoo Cuculus canorus, Oriental reed warbler Acrocephalus orientalis (ORW, main host), and black-browed reed warbler Acrocephalus bistrigiceps (BRW, rare host), could recognize each other’s alarm calls in response to cuckoos. Dummies of common cuckoo (parasite) and Eurasian sparrowhawk Accipiter nisus (predator) were used to induce and record alarm calls of the two warbler species, respectively. In the conspecific alarm-call playback experiments, ORW responded more strongly to cuckoo alarm calls than to sparrowhawk alarm calls, while BRW responded less strongly to cuckoo alarm calls than to sparrowhawk alarm calls. In the heterospecific alarm-call playback experiments, both ORW and BRW responded less strongly to cuckoo alarm calls than sparrowhawk alarm calls. BRW seemed to learn the association between parasite-related alarm calls of the ORW and the cuckoo by observing the process of ORW attacking cuckoos. In contrast, alarm calls of BRW to cuckoos were rarely recorded in most cases. BRW with low parasite pressure still developed recognition of heterospecific parasite-related alarm call. Unintended receivers in the same community should recognize heterospecific alarm calls precisely to extract valuable information.