Knock control systems based on engine block vibrations analysis are widely adopted in passenger car engines, but such approach shows its main limits at high engine speeds, since knock intensity measurement becomes less reliable due to the increased background mechanical noise. For small two wheelers engines, knock has not been historically considered a crucial issue, mainly due to small-sized combustion chambers and mixture enrichment. Due to more stringent emission regulations and in search of reduced CO2 emissions, an effective on-board knock controller acquires today greater importance also for motorcycle applications, since it could protect the engine when different fuel types are used, and it could significantly reduce fuel consumption (by avoiding lambda enrichment and/or allowing higher compression ratios to be adopted). These types of engines typically work at high rotational speeds and the reduced signal to noise ratio makes knock onset difficult to identify. The paper shows how knock-related information can be extracted both from accelerometer and acoustic signals, and how the correlation with in-cylinder pressure based indexes can be optimized using advanced signal processing algorithms and specific calibration methodologies, for a wide engine speed range. An optimization procedure that has involved all the calibration parameters that make up sound and vibration-based knock indexes, has allowed to successfully apply knock detection techniques up to 13,000 rpm. Experimental results obtained on the engine test bench are shown throughout the paper, demonstrating the feasibility of both approaches, which provide similar signal-to-noise ratio levels, and can therefore be considered as possible alternatives.