Besides, considering that both methods (your app and the Audacity method) use the noise produced by the shutter, why is the method of calculation different*?
With audacity you seem to always use "baseline to last peak". With you app you use "peak to peak" first, then "point between peaks to peak" second, then I don't remember the third, but it is not coherent with the other two.
A clear algorithm should be given, such as "baseline to last peak" every time.
This site uses "from baseline to beginning of peak excluded" that might be clearer and give more repeatable results.
And I know that you do and I criticise, and criticizing is easier than doing , I just wanted to say that your video doesn't seem to demonstrate a lot regarding the application. Which doesn't mean your application isn't useful or precise, as it can still be both things.
* Mmmh maybe the audacity method generates an electric noise through light, which is recorded by the sound card of the computer? A proper comparison would better use a proper optical measurement.