the results are quite different, why?

[smith24122412]

I run python test_srmr.py. But the results are quite different, why?

test_code:
ratio_norm, avg_energy_norm = srmr(s, fs, fast=False, norm=True, max_cf=30)
#assert np.allclose(ratio_norm, correct_ratios[3], rtol=1e-6, atol=1e-12)
print(ratio, ratio_norm_fast, ratio_slow, ratio_norm )
print(correct_ratios[1], correct_ratios[2], correct_ratios[0], correct_ratios[3] )

print output:
6.736847774405838 3.715814103437327 6.055213264706728 2.3186338803227864
6.062267651334784 3.44954048853273 6.116783821833282 2.300403317120358

the results are quite different, why ?

[jfsantos]

Thanks for your message. I have not tried the tests in a while, and I can reproduce the tests not passing here. The fast implementations depend on the Gammatone package, for which I unfortunately did not tag a version. In general, the absolute scores do not matter that much, as long as correlations between the scores achieved by this implementation and the MATLAB implementation match.

Not closing this but I am not sure I will have time in the near future to have a look, so if you do some investigation, please keep me posted!

[LTuohaha]

I run python test_srmr.py. But the results are quite different, why?

test_code: ratio_norm, avg_energy_norm = srmr(s, fs, fast=False, norm=True, max_cf=30) #assert np.allclose(ratio_norm, correct_ratios[3], rtol=1e-6, atol=1e-12) print(ratio, ratio_norm_fast, ratio_slow, ratio_norm ) print(correct_ratios[1], correct_ratios[2], correct_ratios[0], correct_ratios[3] )

print output: 6.736847774405838 3.715814103437327 6.055213264706728 2.3186338803227864 6.062267651334784 3.44954048853273 6.116783821833282 2.300403317120358

the results are quite different, why ?

It said that the srmr scores are ranging from 0 to 1, but your scores above are larger than 1.( I got the same problems when I ran the srmr.py file). Do you know why?