With a flywheel or conical pulley, I think it only make sense to measure angular velocity, not linear velocity. I guess it sort of depends on what you are trying to calculate with that velocity measure.
I think it is possible with a rotatory encoder. An option is using Exentrix (from Smartcoach), which measures velocity and power in isoinertial exercises. I´m sure there are cheaper rotatory encoders on the market.
The software of SmartCoach provides data of both mean and peak velocity (in laps / second).If you want data in m/s you can measure the diameter of the axis and calculate the values (it is only an idea). You can also try to standardize and measure the range of motion and calculate the velocity through the displacement and time of each repetition (provided by SmartCoach).