Data acquisition and data processing are at the core of pavement management systems. Nowadays, traditional methods of data collection are rarely used in developed countries due to the considerable disadvantages of the traditional methods compared to the automated ones, such as the slow pace of data collection, endangering the safety of the human operators collecting the data, considerable cost, collecting data from only a limited section of road networks, and inconsistency among the data collected by different operators. In contrast, automated methods alleviate the majority of these problems. However, the main drawback of the automated methods is the high cost of purchase, implementation, and maintenance of the equipment, which has deterred their use in developing countries with limited financial resources. To address this problem, developing a new automated method that reduces the production costs while keeping the required accuracy and performance seems imperative. The goal of this research is to investigate the use of microphones, as inexpensive equipment with acceptable accuracy, for collecting pavement macrotexture data, which is an input to the pavement management system. The proposed method is based on the tire/road interaction noise. To this end, a review of the previous researches on audio-based monitoring of various pavement features is presented. By considering the results of the related researches and the goals of this work, a new setup for data collection and an accompanying signal processing method is proposed. To develop and evaluate the proposed method, data from six standard road sections of a test field are collected. To process the collected data, PCA, Cepstrum, LPC, LSF, PSD, and Wavelet methods are employed. The SVM and KNN classification methods are used to evaluate the results of the signal processing step, which is performed in various frequency bands. The best results are obtained by using the Cepstrum signal processing method along with the SVM classifier in the 3000–5000 Hz frequency band resulting in an accuracy of 95% on the test data and the precision error of 1%.