A Novel Adaptive ANC Algorithm for Removal of Background Noise in Speech Applications G Vinothkumar, D Manoj Kumar International Arab Journal of Information Technology, 2024 Noise is an unsafe mechanical toxin that causes serious hearing misfortune in the working environment of every nation. The working people in the military, mining, development, printing, and saw factories tend to lose their hearing performance due to the adverse effects of noise generated by the machines. They undergo elevated levels of noise, with various machinery producing greater levels of noise measured in decibels. These sounds may cause major health problems that may not allow the person to work in such conditions. Algorithms like Least Mean Square (LMS), Normalized Least Mean Squared (NLMS), Filtered-x Least Mean Squared (FxLMS) and Filtered-x Normalized Least Mean Squared (FxNLMS) are frequently being used for noise cancellation. Moreover, these filters have instability and poor noise reduction; slow convergence also requires a greater number of filter taps and less performance to identify the unknown system in the Active Noise Canceller (ANC). In this paper, a Précised FxNLMS (P-FxNLMS) algorithm is introduced for an ANC. This algorithm consists of dual adaptive filters, an updated Variable Step Size (VSS), a delay in the primary path, a slight improvement in the on-line secondary path, and a modified filter step size when compared to an existing ANC system, with the purpose of minimizing the demerits of existing algorithms. Initially, the P-FxNLMS algorithm was tested with Additive White Gaussian Noise (AWGN) and later tested with real noises from the NOISEUS dataset to check the noise reduction performance. The increase in Signal to Noise Ratio (SNR) segmentation for P-FxNLMS is around 1.45 dB to 4.07 dB and 38.46 % to 73.68 % of the Mean Square Error (MSE) as compared to the algorithms available for different sounds with different SNR input levels. From the performance results of MSE and SNR improvement (SNRi), we found improvements compared with existing algorithms
Speech Enhancement with Background Noise Suppression in Various Data Corpus Using Bi-LSTM Algorithm Vinothkumar G, Manoj Kumar D International Journal of Electrical and Electronics Research, 2024 Noise reduction is one of the crucial procedures in today’s teleconferencing scenarios. The signal-to-noise ratio (SNR) is a paramount factor considered for reducing the Bit error rate (BER). Minimizing the BER will result in the increase of SNR which improves the reliability and performance of the communication system. The microphone is the primary audio input device that captures the input signal, as the input signal is carried away it gets interfered with white noise and phase noise. Thus, the output signal is the combination of the input signal and reverberation noise. Our idea is to minimize the interfering noise thus improving the SNR. To achieve this, we develop a real-time speech-enhancing method that utilizes an enhanced recurrent neural network with Bidirectional Long Short Term Memory (Bi-LSTM). One LSTM in this sequence processing framework accepts the input in the forward direction, whereas the other LSTM takes it in the opposite direction, making up the Bi-LSTM. Considering Bi-LSTM, it takes fewer tensor operations which makes it quicker and more efficient. The Bi-LSTM is trained in real-time using various noise signals. The trained system is utilized to provide an unaltered signal by reducing the noise signal, thus making the proposed system comparable to other noise-suppressing systems. The STOI and PESQ metrics demonstrate a rise of approximately 0.5% to 14.8% and 1.77% to 29.8%, respectively, in contrast to the existing algorithms across various sound types and different input signal-to-noise ratio (SNR) levels.
Filter performance of sparse noise for controlling the occurrence of noise-induced hearing loss using hybrid algorithm G. Vinothkumar, P. Phani Kumar Polasi Aip Conference Proceedings, 2022 Noise is the unsafe mechanical toxin that causes serious hearing misfortune in labors of each nation on the planet. The labors in enterprises like mining, development, printing, saw factories, smashers are in danger. They are presented to elevated levels of noise all through their lifetime of work, yet there are not very many NIHL concentrates in India to show its reality. The loud noises can be heard in the airports while the flight takes off, at construction sites and many other industries. These sounds may cause irritation, headaches, and other major health problems that may not allow the person to work in such conditions. This biomedical project includes the process of reduction of noises that are being prevailed during the working of the individual and will allow him to differentiate the types of noises around. In this paper we have introduced Hybrid algorithm which is combination of Sign-Sign LMS and MPNLMS to reduce the sparse noise with betterment of speech quality. The increase in SNR segmentation for SS-MPNLMS is around 1.45 dB to 4.07 dB and 35.71 % to 88.68 % of the MSE also 2.47% to 44.4% improvement of PSNR as compared to the algorithms available for different sounds with different SNR input levels. They will be able to enhance the speech quality during work, reduction of noises will be seen and his hearing ability will not get damaged with the help of the software that is being proposed.
A Fuel level indicator using load cell through an audio feed output and location tracker for visually impaired C Aravindan, G Vinoth Kumar, R Arthi, R Abhilash 2022 International Conference on Communication Computing and Internet of Things Ic3iot 2022 Proceedings, 2022 Now a day's utilization of vehicles has been extensively increased everywhere around the world that each one of us own certain type of transportation facility. Even middle class /high class visually impaired people own vehicles equipped with drivers face common problems like unaware of the amount of fuel filled, distance travelled and unaware of places they are travelling. The proposed work suggests a solution to solve these problems by fixing a prototype in any vehicle and feeds the information in an audio form. Therefore, any Visually impaired person can keep track of all this information and these data can also be shared with some trusted person to take care of the visually impaired person's safety. The prototype uses the fuel level indicator using a load cell and the output is processed with the CPU and sent to an audio feed, GSM modem with GPS module and LCD display are used as the location tracker and display.
