This paper intends to provide the best suited noise removal technique for de-noising and retrieving clean speech from a noisy speech signal. The aim is to use different de-noising techniques and compare their performance and arrive at a conclusion regarding which one of them is best suited for enhancing voice signals. The analysis is done by evaluating the performance of different denoising techniques for different types of speech samples. This evaluation is done by adding random noise to speech signal then applying denoising techniques to get denoised speech signal. A parallelism is drawn between original signal and denoised signal through evaluation parameters such as SNR and PSNR. The denoising methods are broadly classified as ‘The Filtering Methods’ and ‘The Neural Network Methods’. Under filtering methods four different denoising methods have been used. The four different denoising methods are – Adaptive Filter based on LMS Algorithm, Weiner Filter, Chebyshev Filter and Kalman Filter. Under neural network methods we use three different denoising methods ‘ADALINE’ and two deep learning methods with ‘Fully Connected’ and ‘Fully Convolutional’ neural networks. The performance estimation is done based on variation of evaluation parameters (SNR and PSNR values) for different denoising techniques.