Voxengo Premium Membership - All Voxengo Plugins For a Fixed One-Time Fee
Forums     Plugins     CurveEQ The fallback speed of CurveEQ

This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.

Me & a friend have compare CurveEQ and its analysis with 2 other prominent match EQs.  We were surprised how different the results were between the 3.

Something dawned on us, and we did a test.  From the test, we came to the conclusion that the differences primarily must come from that the 3 EQs uses very different curve fallback rates (during recording of the curve).  For one of them, we estimated the fallback rate to about 15 db/sec.  For CurveEQ however, the fallback rate seemed to be there, but extremely slow (?) It's difficult to see, due to the layout of the grid.

It seems that CurveEQ records the mostly the peaks, more than the average profile over time.  This might be why the different results and it might also be the reason why the recorded curve of CurveEQ works great for some material but may fall behind the results of other match-EQs with some material.

Is this observation correct, that CurveEQ have little- or no curve fallback?


CurveEQ does not use fallback during matching - it accumulates statistics until you finish the capture/matching process.

Thanks, mr Vaneev, for responding.  Now I understand.

Reading my post from yesterday, I feel my curiosity came off as confrontational.  This was not my intention.  I feel the CurveEQ is one of the best out there, and it is the only one I know of which is completely free from aliasing artifacts.

The test I described above was conducted by first removing all frequencies below 1 kHz from a source material which was 60 secs long, and then adding a 1 sec long narrow 100 Hz tone somewhere along the 60 sec timeline.  CurveEQ recorded this material including the 1 sec low note as if the material consisted of plenty of material below 1 kHz.

May I suggest for future versions, that CurveEQ gets equipped with the ability to collect statistics which takes loudness over time into account; perhaps using RMS, or peak fallback etc - maybe as selectable options alongside 'peak' - to be able to create a curve that is even more corresponding to the source material.  I feel this would make a remarkable enhancement to an already good product.

Thanks again

/G


Future CurveEQ version will have option to capture average spectrum with the specified fallback rate.  Beside that, matching can be performed using sum of 1 or 2 source spectrums.

Wow!  Can't wait :-)
This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.