A noise robust audio fingerprint extraction technique for mobile devices using gradient histograms

Audio fingerprinting technology is commonly used in the field of automatic content recognition (ACR). Although various audio fingerprinting techniques have previously been proposed, we propose an advanced landmark extraction technique built on gradient-based features. With the widespread use of smart phones, a large number of audio fingerprinting applications are currently running on mobile phones in outdoor environments. As a result, these audio fingerprinting systems suffer from noise interference. To overcome this problem, we propose a novel scheme to effectively extract salient features and identify an audio signal from them. Test results show that our proposed approach can more accurately identify an audio signal in a noisy environment.