search

Eomys / MoSQITo

MoSQITo is a unified and modular development framework of key sound quality metrics favoring reproducible science and efficient shared scripting among engineers, teachers and researchers community.
Apache License 2.0
134 stars 44 forks source link

The equation of the R_spec calculation #57

Closed harmomyshoes closed 1 year ago

harmomyshoes commented 2 years ago

The code on the roughess_dw_main_calc function underneath 

R_spec[0] = gzi[0] * pow(mod_depth[0] * ki[0], 2)
R_spec[1] = gzi[1] * pow(mod_depth[1] * ki[1], 2)
for i in np.arange(2, 45):
    R_spec[i] = gzi[i] * pow(mod_depth[i] * ki[i] * ki[i - 2], 2)
R_spec[45] = gzi[45] * pow(mod_depth[45] * ki[43], 2)
R_spec[46] = gzi[46] * pow(mod_depth[46] * ki[44], 2)

then I check the "Handbook of Signal Processing in Acoustics" by Peter Daniel. image in the book, the gzi seems to need a pow^2,

whether I have some misunderstandings here.

wantysal commented 1 year ago

Hi, sorry for the late reply...

There is indeed a gray area in this part of the calculation since the article by P. Daniel does not provide the coefficients g(zi)... We used the Aures coefficients and then tried to fit the model to the results given by Daniel and Weber. It appeared that the g(zi) coefficients we used were already squared since they provide better results.

Thank you for noticing this, if you have more data about this calculation, feel free to contribute to MOSQITO development ;) Salomé

harmomyshoes commented 1 year ago

many thanks for the explanation, do not have much data to contribute at the moment, hope will contribute to this project in future.