Content based audio content MIR (Music Information Retrieval) face two major challenges: the robustness of the algorithm and the speed of this operation. In this article a model of fast algorithm for the extraction of audio data by the fingerprinting technique is proposed, used to advertising detection, which is implemented on a low cost embedded platform and then parallelized to run on a platform based on graphics hardware (GPU). Test determined a success rate in all cases and response times approximately two times lower than those obtained with a PC workstation, allowing searches of up to 65 real-time commercial