A mobile phone, which is a complex device with number of sensors, processors, communication interfaces and other hardware and software components. Low-cost sophisticated mobile phones are very popular, affordable and commonly used in the present era. Mobile phones are useful in many ways at the same time there are misuses also being reported, such as spreading rumors and violence, fraud, information theft and security breaches using mobile phones. Most of those misuses can be controlled by identifying source mobile phone. Also mobile phone identification can be used for authentication and forensic investigation purposes. There are several approaches proposed by researchers to identify the mobile phone using their physical fingerprints of the built-in components, such as microphone, camera and accelerometer. Due to the imperfections in the electronic components, there will be minor variations in the performance of each mobile phone. Those imperfection and performance variations reflect in the outputs from the device. Source mobile phone can be identified by studying the output variations. Speech is the most common way for humans to communicate through mobile and speech signals can be recorded easily by mobile phones. Recorded speech signals may contain variety of additional information, such as details of surroundings, gender of speaker and also the detail of recorder itself. Evaluating audio features for the detection of a mobile phone using recorded audio is the focus of our study. Since, the research is identifying mobile phones using recorded audio, a clear dataset is essential. Fifty mobile phones including all possible popular brands and models were selected for data collection. Different features, such as MFCC, LFCC and frequency responses were studied. SVM based classifier is built to identify the source mobile phone.