Applications of ML and AI based technologies for real-time signal processing on autonomous system

• Main Authors: Mariappan, Udhaya Sankar Sankara Moorthy, Rajasekar, Roshan Kumaar, Saravanan, Imayavan, Batcha, Maguthu Sahith Afri Sathick
• Format: Conference Proceeding
• Language: English
• Subjects: Acoustics; Automatic speech recognition; Microphones; Pattern recognition; Real time; Signal processing; Sound sources; Visual impairment
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0152529

Picture handling is a significant field of example for acknowledgment techniques. The sound sources emitted in an auditorium as acoustic signals are recorded with microphones, signals are often coupled between source signals present in that room. The source signals are reverberated and so the signals are deformed and less intelligible, when speakers are in a little space from microphones in room. The convolutional network have intelligible speech signals, detrimental effect on the apparent quality due to reverberation in closed room, which damages the effectiveness of automatic speech recognition. Due to the interference of speech signals and the occurrence of reverberations among the signals makes separation of speech signals from the distant microphone turn into the more complex and challenging process. Therefore, it is in need to build up a technique which can recover the source speech signals from reverberated and interfering signals, and that can also reduce the adverse acoustic effect. This article focused on Recuperating and isolating source signals from the caught mouthpiece signals, in particular visually impaired source division and visually impaired dereverberation of the convolutive discourse blends, will be valuable for some wide scopes of sound applications like biomedical sign investigation, different radio wire remote interchanges, acoustic and discourse preparing, and so on. The speculations and techniques it examines have gotten broad consideration in many trains and fields, advancing the improvement of man-made brainpower frameworks and growing the conceivable outcomes of PC applications. The pattern recognition process can be seen as a mapping process from the sample space to the category space an n-dimensional feature space with n features is considered as parameters into different regions in which each region corresponds to a class of pattern classes that provides efficient real-time signaling for blind sources.