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.
would it be possible to have a more "precise" mode for Spectrumatching? Currently its missing some high frequency peaks. Would it be possible that spectrumatching uses more points (more resolution -> more precision) to be more accurate?
And another question, would it be possible to add a peakmeter for in and output? or at least for output, i use curveeq for matching my soundsystem to my room and so also in winamp and there i cant see if any clipping occurs.
No, I don't plan to add a more 'precise' mode for SpectruMatch since it was not designed for the 'deep surgery' from the very start.
I will think about peak meter, but honestly I don't feel like it is necessary. Peak meter will need some CPU. Since in pro audio environments we already have peak meters, additional peak meter is an excessive addition. Maybe you should use a dedicated peak meter plug-in with winamp?
well... just an idea, but please think about the spectrumatch precision, i think i am not the only one who needs this and your EQ is the only one that can do this (and very well already) in Windows.
As it seems aleksey wont implement a better resolution i wonder if someone knows a better implementation of such a feature in another equalizer plugin?
Actually, I have found some workaround for this limitation. If I manage to introduce it, CurveEQ will allow 60 band SpectruMatch (that's twice the current number of bands). Since this is a great number of bands, they can be edited further in the 'freehand' mode.
But now I need some comments. Do I need to implement a switch between 30 and 60 bands or not? BTW, making number of SM bands arbitrary is not technically possible.
I also feel like an undo/redo function pair will be of use too.
IMHO this 'EQ Matching', is more or less just hype ( at least for mastering ) !
There are too many things, that influence the frequency curve of a mastered song :
Does the bass play the same notes and note lenghts ...
what instruments/sounds, were used ...
Were the vocals deessed ...
Were the individual tracks compressed and/or limited ...
What kind of compression/limiting was used at mastering stage ...
If any of these things differ on two songs ( which is very likely),
the song that should be matched, to an other,
is likely to get a worse sound, than before !
A higher resolution of SpetruMatch, would even increase the EQing mistakes.
Instead, I would prefer a 'infinite peak hold', for the analyzer.
If Curve EQ could memorize a few of these Peak hold curves, it would be quite easy,
to make frequency curves similar. Which is way better than to make them match exactely !
don't know how the spectrum matching works internally, but I would suggest to use some kind of least square fitting. In this case the number of eq points would be a free parameter. The resulting eq curve does not nescessarily match the target curve precisely at any frequency but rather give an overall good match. I would find it useful to work with a few points only in some cases (in accordance with what Jan has said). Another option could be to use something like gaussian curves as a 'fit model' and afterwards calculate the spline representation to feed it to the curveeq engine. If we are talking about 30 or 60 eq points I wonder if a 'simplfy' command can be useful?
I totally agree that for mastering purposes the SpectruMatch resolution needs, if anything, to be lower -- otherwise actually we can actually end up trying to match the key of one song to another (if you see what I mean), rather than making the broader, smoother shifts that that are more commonly appplied in mastering. In defence of the original poster -- I think he wanted to use a higher res mode to take account of frquency imbalances caused by speaker/room interaction, which is a completely different (though valid) use.
When I have used the SpectruMatch feature, I have always smoothed out the matched slope (and lessened its "amplitude"), I presume that to smooth out the curve automatically, CEQ would have to do another recalculation, because just thinning out control points would not preserve the overall shape. The idea of simulating gaussian curves with control points, might provide a solution to this -- as well as allowing for a parametric style EQ to be simulated. Having said this CEQ is a great tool, now, as it stands; and these are, by no means requirements.