Apparels are the main end-use of textile fabrics. In addition to general requirements that have to be satisfied by textile fabrics when converted to human apparel, fabrics are expected to meet certain other specific requirements. These are collectively known as textile comfort. Fabric comfort comprises fabric sound which is the sound generated by fabric during wear by users. Depending on the type of garment, fabric sound can be a source of discomfort. Therefore, the amount of sound generated by fabric can be considered as an index of apparel comfort. This index can determine the suitability of fabrics for their intended end-use . In this work sound generated by samples of fabric woven by cotton and cotton-polyester yarns was investigated. Specifications of the compared samples in warp direction were identical. In order to stimulate sound generated by the samples, an apparatus capable of sound induction was designed and developed. The recorded sound signals were analyzed, using Discrete Fourier Transform, together with Discrete Wavelet Transform. Sub-bands energy of FFT and energy coefficients of wavelet transform were calculated. Additionally stepwise multiple regression technique was employed. Results showed that, generated fabric sound is affected by both fabric pick density and weft yarn linear density. Sound energy is increased, when pick density of cotton-polyester is increased. However in case of polyester fabrics, increases in pick density at certain frequencies resulted in reduction of sound energy. It was also found that in case of cotton-polyester fabrics an inverse relation exists between weft yarn linear density and sound energy. This was confirmed by the observed increases in sub-bands energy of the relevant sound frequencies . It was found that, the induced fabric sound was affected by both physical and mechanical properties of the samples. In this respect, tensile, shear, bending and surface properties as well as weft fractional cover of fabric were found to affect the amount of induced sound. The effect of above factors on sound characteristics of samples was predicted using Multilayer Feed forward network with back propagation learning algorithm. The network results showed that, as far as sound volume is concerned surface roughness and drape coefficient of fabric are the most and the least effective parameters respectively. Based on their sound characteristics samples were categorized into five different class, using Kohonen neural network