You can make a crude torque/rpm meter. First, you need to make a disk which will clamp to the shaft of the motor. The disk should be just as light as you can make it (lots of holes). Although the scale is wrong for your application, think of a bicycle wheel as the goal. The rim of the disk needs a groove all the way around to guide and accumulate a thread which you will wind up. The thread will go up to a pulley as high as you can mount it with clear access all the way to the floor below. It's important that the shaft is centered in the disk - if not, the rim speed will vary with shaft rotation, and you want to produce a constant speed so no force is being used to accelerate or decelerate the weight.
To operate this, run out the thread from the groove, leaving it attached to one end, run the other end through the pulley and down to the floor. Attach a known weight to the thread. Start the motor. Give the motor time to get up to speed (which may or may not be possible, especially if the disk is heavy), let's say until the weight is 1 meter from the ground. Now measure the time for the weight to rise H (pick something useful like 1 meter or 1 foot). Call this time T. Call the radius of the disk R. Assume the thread has no mass, but the weight of the mass being lifted is M.
The velocity V of the weight is H/T. The force F on the thread is g M (1 + V). The circumference of the disk is 2 pi R, so the rotational speed of the shaft is (H/T)/ 2 pi R in RPS, or multiply by 60 to get RPM. The torque is F R (force times radius).
You'll have to pay attention to keeping your units straight, but that's just bookkeeping.
What you're doing is, rather than trying for fixed RPM and measuring torque, fixing the torque and measuring RPM. There is no guarantee that this will work well. First, it assumes that you can get the motor up to speed in a suitably short time, before the weight runs out of free travel. If the moment of inertia of the disk is too large (and it goes as the square of the radius) the weight will not reach constant velocity in time, and that means that the motor will not reach constant RPM. The same consideration applies to the reflected torsional load of the weight. Sources of friction, especially the pulley, will also affect the result. Finally, of course, you'll need to make fairly precision measurements of the time interval for the H interval. An electronic timer is probably a good idea here. Make several measurements for each weight, throw out any weird measurements and take the average.
Assuming you're using low-torque, high-speed motors, you'll want to keep the disk radius small, in order to get reasonable weight speed. While this has the advantage of allowing heavier weights for a given torque, pay attention to the thread takeup. If you start getting a lot of layers of thread, the effective radius will increase and this may throw off your calculations. You asked for an "accurate" unit, but you didn't specify exactly what accuracy means.