... process real-time audiostreams, with a buffer size of 512 samples, a frame size of 2048 samples, a linear prediction model of the order of 5, and a maximum of 20 peaks per frame for ODFPADFLOPSODFE529,718ODFSD7,587,542ODFCD14,473,789ODFELP734,370ODFSDLP217,179,364ODFCDLP217,709,168ODFPAD9,555,940Glover ... computationalperformance of all of the described techniques, evaluated using Modal, our new open source library for musicalonset detection, which comes with a free database of samples with hand-labelled ... number of correctly detected onsets, fpis the numb er of false positives (detected onsets with nomatching reference onset), and fnis the number of falsenegatives (reference onsets with no...