Fusion: Practice and Applications

Journal DOI


Submit Your Paper

2692-4048ISSN (Online) 2770-0070ISSN (Print)

Vocal Analysis and Sentiment Discernment using AI

Praveen Singh , Preeti Nagrath

One of the major factors for personal development and growth is understanding human emotions, and therefore it plays an important role in imitating human intelligence. Vocal and Sentiment analysis are the major focus points for advancement in Artificial Intelligence (AI). Sentiment analysis provides major help to data analysts of big enterprises to measure public opinion, conducting market research, understanding customers experience and viewing brand and product reputation. Emotion recognition provides an opportunity to grasp the general people’s sentiments about social events, marketing strategies, political views and product liking. In this paper, we have used various AI models on a variety of audio datasets to recognise and analyse the sentiments of the speaker. Our dataset includes some audio songs sung by some singers and some audio clips of few actors. We trained CNN and LSTM models to analyse our dataset and predict their accuracy. The ever-growing need of sentiment analysis coincides greatly with the extension of social media such as forum discussions, social networks like Facebook, Twitter, Instagram and many other similar platforms.

Read More

Doi: https://doi.org/10.54216/FPA.070204

Vol. 7 Issue. 2 PP. 100-109, (2022)

Integrated Decision Making Aided Model to Estimate the Risks of the Excavation System

Lobna Osman

For the last years, a bibliometric examination of risk evaluation approaches for excavating systems has been presented in this publication. To develop an early warning system, it's essential to compile a list of possible dangers that can arise during excavation. Failure Mode and Effects Analysis (FMEA) is a useful approach. Traditional risk assessment techniques have been criticized for a variety of reasons, including a lack of correlation between risk variables, difficult arithmetic operations, and a lack of correctness and preciseness in the evaluations. A unique method of risk analysis in FMEA that uses digraphs and matrix approaches underneath the Pythagorean fuzzy scenario is presented in this research. To get started, we'll defy Pythagorean fuzzy numbers in a triangle form. Both language terminology and risk factor data and information are expressed using them (inclusive of occurrence, severity, and detection). The Pythagorean fuzzy digraph thus captures the interrelationships between the risk variables and the relative importance of each one, as seen in the figure. After that, we create a Pythagorean fuzzy test indicated for each identified failure mode and compute risk priority indexes to determine risk priorities. Using a metro station excavation as a case study, the accuracy of risk assessments in excavation is improved.

Read More

Doi: https://doi.org/10.54216/FPA.070205

Vol. 7 Issue. 2 PP. 110-123, (2022)

Detection of Covid-19 using Cough Sounds

Harsh Taneja , Abhinav , Apoorv , Himanshu Mangal , Naman Agarwal

Coronavirus, the pandemic due to which about 4 million have lost their lives and counting, is still on. Many scientists and researchers are trying to find ways to detect coronavirus as soon as possible in the human body so that they can start their medication and precaution as soon as possible. Still, due to lack of lab facilities, the RT-PCR is taking more than three days to give the report, and in the meanwhile, patients get serious and life in danger. So in this paper, we proposed an audio-based coronavirus detection technique in which we can get results in minutes. Coronavirus is a respiratory disease, and the sound produced while breathing can tell us about the presence of coronavirus. Audio-based detection was already used for the detection of asthma, pneumonia. So, in this paper, we implemented a combination of machine learning and deep learning techniques to find the presence of Covid-19, and the model has an accuracy of 78% and an f1 score of 74%. This technique can be used as a starting point for just audio data to diagnose diseases and save lives.

Read More

Doi: https://doi.org/10.54216/FPA.070202

Vol. 7 Issue. 2 PP. 79-90, (2022)

An Intelligent Optimization Model for Fragment Assembly Problem

Osama Maher

problem as it requires rebuilding the original sequence of DNA from a massive number of fragments. This paper introduces an efficient hybridization between Harris Hawks Optimization (HHO) and Problem Aware Local Search (PALS) to be utilized in solving DNA Fragment Assembly Problem (FAP). The efficiency of the proposed hybrid algor In computational molecular biology, the sequencing of Deoxyribonucleic Acid (DNA) is a very challenging ithm (PHHO) is compared with the original PALS, Firefly Algorithm (FA), Genetic Algorithm (GA), Artificial Bee Colony (ABC), and GAG50. The experimental results show the efficiency of the proposed algorithm compared to other approaches.

Read More

Doi: https://doi.org/10.54216/FPA.070201

Vol. 7 Issue. 2 PP. 66-78, (2022)

Driver Drowsiness Detection in Real-time

Daksh Khetan , Arun Nawani , Anshul Aggarwal , Ms. Surinder Kaur

In modern life, drowsiness is one of the major causes of road accidents, many of which are fatal. Analyzing statistics, it can be assumed that most road accidents occur as a result of drowsiness leading to serious injury and death. For this reason, various studies have been done on designing programs that can detect driver fatigue and alert them before a serious error occurs. This prevents them from falling asleep and having an accident. Some of the most common methods use automotive-based methods to design their own system. But these traditional measures were strongly influenced by other factors such as road structure, vehicle type and driver-wheel driveability. Some methods use psychological methods of their system that often provide the most accurate and consistent results in the driver's drowsiness monitoring. However, such techniques are very tedious as the electrodes need to be placed on the head and body. In addition, few studies are available where independent measurements are used as system installation, but such methods can confuse the driver and lead to unintended consequences. In this paper, we have proposed a non-disruptive and real-time program. Our proposed system classifies it as sleep deprivation. The model is fed with a large database of closed eyes and open eyes to produce results. The driver is notified by Buzz every time he is found drowsy. In our model, we use a standard forward-looking smartphone camera and use the information we have gained to produce results on our website. This can be more economical than using additional hardware.

Read More

Doi: https://doi.org/10.54216/FPA.070203

Vol. 7 Issue. 2 PP. 91-99, (2022)