offline secondary_path_identification issue about rpianc HOT 2 CLOSED

Hello and thank you for your interest in this project!

Small disclaimer at the beginning, I have not worked on ANC nor digital signal processing since finishing this project ~3-4 years ago, so I may be missing something :)

I know this is not clear from the code, but I believe that in the end I used Mains/offline_sec_path_modelling.cpp , Mains/record_noise.cpp or Mains/record_test_data.cpp only to record the white noise, without outputting any correction signal (as you can see in Mains/offline_sec_path_modelling.cpp, the processing buffer is never copied to the playback buffer, as it is done in the Mains/feedforward_anc.cpp). Then I did offline secondary path modelling in octave, using Scripts/OctaveCode/lms_offlince_sec_path_modelling.m or Scripts/OctaveCode/rls_offline_sec_path_modelling.m), where the adaptive filter is supplied with the error signal calculated as you've described.

In a typical setup where an adaptive filter is used for unknown system identification, yes it is true that you have to use something like e(n) = d(n) - y(n); and provide that to the adaptive filter, as shown in the diagram below.

image source

In the case of the active noise control system, the e(n) = d(n) - y(n); part does not happen during digital signals processing, but the acoustic waves that carry the noise signal (reference_generated_for_processing noise in that case) and the output from the filter (emitted by the speaker) interfere and the error microphone captures the resulting e(n) signal.

image source

Maybe it is worth trying to modify this code to actually output the correction signal from the adaptive filter (noise would have to be played by a separate speaker though, as the playback device used here would have to emit only the correction signal) and this way the estimate of the secondary path would be more accurate.

I hope that helps!

from rpianc.

Thanks for your help. I am now start to debug the anc system.

from rpianc.