Babar,

I am not sure of the purpose of your speedometer (I am assuming for tire speed of a car). You could measure the time between pulses (assuming the pulse is just an instantaneous spike). The pulse could also be inverted by a trigger every time a voltage spike is seen (at contstant speed this should produce a wave form with 50% duty cycle). This would also act as a kind of debounce circuit. Depending on the high time vs. low time compared to distance (the circumference of the tire being monitored) a figure could be generated proportional to the actual speed. Perhaps having the pulse generate an timer count until the next pulse would suffice?
I would recommend using a triggered output pulse as mentioned above, where a spike actually changes the logic of the output. This would probably produce the most linearly accurate figure with the least mathematics. Without doing the math... depending on the lag of the circuit being used to do this, the delay should not make a noticable difference in the o/p, especially at vehicle speeds. How accurate does this really need to be???

Ben