Problems about Composite Objective Quality Measure

[wqr319]

Thanks for this convenient python implementation for quality measures.
However,I am confused about the composite terms. Can you point out where I can get more information about the linear coefficients of CSIG,CBAK and COVL?Because the coefficients do not correspond to those reported in this paper https://ieeexplore.ieee.org/document/4389058 (table 7).

[schmiph2]

Very good point! I used Loizou's book as a reference. More specifically, his Matlab reference implementation in the supporting material of the book (can be found at the bottom of the page in the link above). But it seems that the implemented composite measure is not the same as in the book or the paper you referenced. And its not just the coefficients but also the used predictors. For example for CSIG he uses IS and PESQ as predictors in the paper, in the Matlab implementation LLR,PESQ and WSS. Although he references the paper in the code.
I am not sure where the difference comes from. But I suppose that for better comparability we should stick with the Matlab implementation (since it is widely used in the literature). What do you think about this?

[wqr319]

Sorry for my late reply.
I think you are correct that we should stick with the Matlab implementation.
Moreover,I have tested the paper version of composite measure.To my surprise,they do not correspond to MOS very well.In fact,they give a totally wrong result.I am not sure if I have made some mistakes in my matlab test code.
test.txt
What is your opinion? :)