Extending Assisted Audio Capabilities of TIAGo Service Robot

Abstract

The goal of this paper is to present three modules that can be used independently or in conjunction for audio signal processing purposes. Here, the modules are illustrated in the context of acoustic analysis for TIAGo service robot (recording audio data stream, obtaining isolated audio events, generating classification model, identifying audio events). For two of them the friendly user interfaces are described. The signals under classification belong to a home environment. As features we propose the liftering Mel Frequency Cepstral Coefficients, while for classification/identification of isolated audio events we propose the k-Nearest Neighbor algorithm. For the proposed real-life scenario, we have obtained an accuracy of 98.55% in the testing phase.