The creation of new approaches to the design and configuration of smart buildings relies heavily on AI tools and Machine Learning (ML) algorithms, particularly optimization techniques. The widespread use of electronic devices has sparked a strong desire to incorporate the Internet of Things (IoT) into houses, leading to the development of smart homes. As networked gadgets proliferate rapidly, this phenomenon is characterized by rapid proliferation. In smart buildings, smart cities, smart grids, and smart homes, interconnected electronic devices are becoming more popular. The objective of this paper is to enhance the functionality of home automation systems through the performance of speech recognition using the Bat-Salp Swarm Optimization (BSSO). This paper investigates the notion of (BSSO), a data analysis methodology that facilitates the automated construction of analytical models. The implementation of BSSO provides an enhancement to the feature selection process in speech recognition, providing an approximation solution that improves the accuracy of system decisions. The use of the BSSO technique improves the precision of the voice recognition system and also incorporates an Artificial Neural Network (ANN) for the classification part. The findings substantiated the efficacy of the employed methodology.
Modern home automation technology is moving us towards the perfect smart home setup. Artificial intelligence (AI) and deep learning are effective in the IoT. In addition, search algorithms rely on metaheuristics, which are essential optimization methods, to solve complex AI problems. Speech recognition and control, data input, voice user interface, and telephone operator automation are the applications. Metaheuristic approaches can mimic physical, biological, or natural principles that exhibit search or optimization traits and predictability [15]. Smart buildings, cities, grids, and homes are just a few examples of numerous environments where networked devices are becoming increasingly common. Smart buildings, towns, electricity systems, and residences are using more networked electronics. This paper introduces Bat-Salp Swarm Optimization (BSSO), a data analysis method that automatically creates ∗ Corresponding author. † These authors contributed equally. analytical models. BSSO uses approximation to improve voice recognition feature selection and system decisions. Moreover, the application of the BSSO technique improves the precision of the voice recognition system. The findings substantiated the efficacy of the employed methodologies. The rest of our paper is organized as follows: Section. 2 gives a brief description of related studies. Section.3 illustrates the proposed model. Section.4 shows the experimental results. Finally, Section.5 is the conclusion.
Numerous metaheuristic strategies in various applications have been introduced by authors
recently. Cross-lingual Voice Conversion is the process of modifying the speaker identification
of a voice sample from one speaker to another speaker who speaks a different language from
the source speaker. Converters are frequently employed in the provision of a converters, and
they displayed the fully integrated power with simulation in different cases. Various
methodologies have been devised for voice recognition; however, attaining a high level of
precision continues to pose a substantial obstacle [
Several authors, including [
The paper's structure contains three main elements (see Figure 1). The preprocessing stage is followed by the feature selection stage, which combines two primary optimisation techniques: Bat optimisation and Salp Swarm optimisation approach (BSSO). The last part is the classification stage, which utilises an Artificial Neural Network (ANN). The subsequent analysis will focus on the main discussion of each stage illustrated in the block.
The first block pre-processes. The Discrete Wavelet Transform (DWT) converted the signal from time to frequency after having filtered it. The Discrete Wavelet Transform (DWT) at its maximal decomposition level produces a signal with precise frequency resolution [20]. The MATLAB digital signal processing program is used to apply three stages of Discrete Wavelet Transform (DWT). The DWT divides signals into approximation and detailed sub bands. Most noise affects the high-frequency sub band, while vital information is in the low-frequency sub band. This level uses Daubechies wavelet [20-21].
The major feature section approach is bat optimisation. The Bat Algorithm (BA) mimics bat hunting. BAs operate on this idea. The bats in the swarm are thought to travel randomly at a certain velocity and position .
They search for prey using a steady frequency , variable wavelength , and a specific
loudness 0. By manipulating pulse wavelength and emission rate ∈ [
1) Bats employ echolocation to perceive distance and possess the ability to differentiate between prey and obstacles.
2) To search for prey, bats fly randomly with the velocity vi at position xi with varying frequencies fi (from a minimum frequency to a maximum frequency) or with varying wavelengths and loudness’s A0. Automatically adjusting wavelength (or frequency) and rate r of pulse emission can also be accomplished depending on target proximity.
3) In the third step, assume that the loudness varies from the largest positive value A0 to the minimum constant value Amin. As a result of the above rules, the position vector of the bat represents a solution in the search space. The algorithm randomly initializes the bats since the global optimal position vector is not known a priori [23].
The bat optimization techniques will apply based on the following equation:
Where i th bat, i = 1, 2,3,. . . , n; j = 1, 2,3,. . . , d; n signifies the population size, d signifies
the dimension of the search space. xlb and xub are lower, upper bounds for j th dimension, rand
is a random number between [
However, the SSO is a bio-inspired algorithm that mimics Salp's natural foraging movements. By mimicking Salps' hunting and foraging techniques, the SSO hopes to efficiently handle complex optimization problems [15,23]. Equation (4) is revised in this study using the SSO hunting equation (9). Keep in mind that the value of the variable c1 changes over time according to equation (8), where L is the maximum number of iterations and l is the current iteration value. The BBSO approach, a novel approach for feature selection in acoustic signal analysis, and the bat optimization algorithm are both greatly improved by this version.
In addition, in the classification phase, we employ hybrid learning utilizing the ANN algorithm, which combines the descent and least-squares methods to determine the parameters [24]. The proposed technique aims to enhance the hardware of the system on a chip by optimising its performance to achieve maximum efficiency. This solution will involve implementing a smart design in the future.
1 = 2 −(4 )2
= 1 × ( + −1)
(8) (9) In the following the main algorithm of this paper as follows:
Part 1 : preprocessing an Feature extraction part Start Loop 1) Preprocessing of audio signal. 2) define objective function f(x), x = (x1, . . . , xd ) T, and Set the initial value of population size n, α, γ, and
N.gen 3) Initialize pulse rates ri and loudness Ai and population based on (1) 4) Evaluate and find x∗ where x∗ ∈ {1, 2, . . . , n} while t ≤ N_gen
for i = 1 to n 5) Adjust frequency (Equation (2)) If i≥2 6) Calculate (8) based on current iteration. 7) Select a solution among the best solutions (best feature of audio signal.
8) Generate a local solution around selected best (5). end if 9) Evaluate objective function if (rand < Ai & f(xi ) )
update based on (9) end if end for t = t + 1 .end while Return best selected features Part 2: Classification stage : 10) Rank the bats and find the current best x∗ - To Train ANN Start Loop 1) Select type of optimization (Hybrid) 2) Choose the number of epochs. 3) Compute the error Err= (actual value-Estimated value)/ (actual value)*100% %% using for estimation of prediction End Loop Display control signal
Our main goal is to accurately identify audio signal qualities for automation systems. The goal is to master and use BSSO. The system extracted features from the audio signal and used those elements to provide instructions. However, adding an ANN signal categorization component enhanced the system. Both the feature extraction and classification methods had over 94% accuracy. The BSSO feature extraction method [24] employed 100–500 iterations (Figure 2). The model is being conducted using a Windows 11 personal computer with a 2.5GHz Core i7 CPU, 16GB of RAM, and MATLAB 2023a software. The cumulative duration for the system to complete all assignments is three minutes. Table 1 illustrates the enhancement in the system's accuracy achieved by the Salp algorithm compared to the BAT techniques and the other techniques based on the same data of this paper from audio signals.
Every signal the smart home sensor unit emits is classified to manage equipment. Categorization uses ANN to classify. Figure 3 shows the ANN classification method and results. Figure 4 displays the classification model managed the smart system utilising audio signals with 98% accuracy and an error histogram. This model classifies audio files using ANN and feature extraction. On audio samples from various sensors, the approach works well. The algorithm is characterised by its high speed and ability to extract practical values for feature extraction. Additionally, it enhances the model of speech values. Furthermore, the model has shortcomings in terms of the time required for data collection, particularly for historical data collection.
AI technologies and Machine Learning (ML) algorithms, specifically optimization techniques, are crucial for developing innovative methods for designing and configuring smart buildings. The Digital device use has increased interest in connecting the Internet of Things (IoT) into houses, creating smart homes. Networked devices are becoming more prevalent in many situations such as smart buildings, cities, grids, and homes. There has been a rise in interest in the concept of constructing smart homes by IoT into homes because of the use of digital devices. It involves the fast spread of connected devices. Connected gadgets are becoming more common in smart buildings, towns, grids, and families. Bat Salp Swarm Optimization is used to integrate voice recognition with home automation systems to improve performance. This paper examines (BSSO), a data analysis method that generates analytical models automatically. Decisions made by voice recognition systems are enhanced by BSSO's feature selection approximation solution. The BSSO method enhances voice recognition accuracy and adds the ANN for classification. The findings supported the methods. The method is distinguished by its rapidity and capacity to get useful values for feature extraction. Furthermore, it improves the framework of speech values. Moreover, the model has deficiencies in terms of the duration needed for data collection, especially for the acquisition of historical data.
During the preparation of this work, the author(s) used QuillBot program and wordtune for rephrasing and improving clarity of certain paragraphs, as well as Grammarly for grammar and spelling checks. All content generated or suggested by these tools was critically reviewed and edited by the authors. The author(s) affirm full responsibility for the accuracy, originality, and integrity of the final manuscript. [14] P., J., N., M., S.M., I., A., Y.,and A.. A Hybrid Bat-Crow Search Algorithm Based Home Energy Management in Smart Grid. In: Barolli, L., Javaid, N., Ikeda, M., Takizawa, M. Complex, Intelligent, and Software Intensive Systems. CISIS 2018. Advances in Intelligent Systems and Computing, Springer, Cham, 19 June 2018,vol 772. [15] M., S., "Salp Swarm Algorithm: A bio-inspired optimizer for engineering design problems."
Advances in engineering software, vol.114 ,2017, pp.163-191. [16] KH., S.,. Energy efficient scheduling of smart home. In: Web, Artificial Intelligence and Network Applications: Proceedings of the Workshops of the 33rd International Conference on Advanced Information Networking and Applications. Springer International Publishing, vol.33, 2019, p. 67-79. [17] N., B.. "Improved salp swarm algorithm with mutation schemes for solving global optimization and engineering problems." Engineering with Computers, pp.1-23, 2022. [18] B.Chaabane, S., "Improved salp swarm optimization algorithm: Application in feature weighting for blind modulation identification." Electronics ,pp.10.16 ,2021. [19] V., H., and M.: "Improving speech recognition using bionic wavelet features." AIMS
Electronics and Electrical Engineering, vol.4.2, 2020, pp. 200-215. [20] T., A. M., A. M., and S. Ch. "Naive Bayes-guided bat algorithm for feature selection." The
Scientific World Journal, vol.2013. [21] Li., Y.:. "An improved bat algorithm based on lévy flights and adjustment factors."
Symmetry, vol.11.7,2019,pp.925. [22] A., S.. "Classification Model for Diabetes Mellitus Diagnosis based on K-Means Clustering Algorithm Optimized with Bat Algorithm." IJACSA) International Journal of Advanced Computer Science and Applications, 2023, pp.14.1. [23] D., M.. "Dry bean cultivars classification using deep CNN features and Salp swarm algorithm based extreme learning machine." Computers and Electronics in Agriculture,2023,pp.204. [24] G., A., G.B, and P.V. Ch.. "Electrocardiogram signal classification using VGGNet: a neural network based classification model." International Journal of Information Technology, vol.15.1,2023, pp. 119-128. [25] R.H.M., K.H.R, and H.FA. "Brain tumors diagnosis and prediction based on applying the learning metaheuristic optimization techniques of particle swarm, ant colony and bee colony." Procedia Computer Science, vol.163, 2019, pp.165-179. [26] R.H.M., K.H.R, and H.FA, and Hesham FA Hamed. "Brain cancer diagnosis and prediction based on neural gas network and adaptive neuro fuzzy." Procedia Computer Science, vol.163 ,2019, pp.518-526.