I have not looked at the mouse audacity method.

I made a tester using a photodiode and oscilloscope, after some trials.

1. The light hitting the photodiode must be collimated so that the light received off the groundglass is from a small spot (1.2 mm) and nearly at 90 degrees.

2. The photodiode must be biased so it is always in its active region. It must not be in off state in the dark, and it must not be saturated in the test light.
That requires adjustment of both the bias resistor and the test light intensity.

These are necessary for the photodiode current pulses to be accurate analogs of the light pulse, otherwise it can give pulses longer than the incident light.

Once I had the tester sorted out, I was able to reset the blind tension on the Speed Graphic to the spec in the service manual.
above 1/100,there is a fair amount of penumbra on the focal plane shutter, and some on the lens shutter, so I used a visual average of the pulse width on the oscilloscope.
Here are the measurements from the last test of the focal plane {setting, measured millisecond)
{1/30, 55}, {1/50, 30}, {1/125, 11), {1/250, 5}, (1/500, 2.2}, {1/1000, 1.3}

So the Graflex focal plane shutter is less than +1/3 of a stop in error.

Here are the measurements from the last test of the Graflex Wollensak shutter on the Optar 1/4.5 135mm {setting, measured ms)
{1/1, 900}, {1/2, 400}, {1/5, 180), {1/10, 103}, (1/25, 32}, {1/50, 19}, {1/100, 10.5}

So the Graflex Wollensak shutter is quite accurate, plus 5%, minus 20% in time which is less than +/- 1/3 stop