Great Job guys.. Har-Bal 2.0 has done wonders... BUT
Great Job guys.. Har-Bal 2.0 has done wonders... BUT
Har-Bal has performed well beyound my expectations.. great tool.. but i have noticed a slight difference in th low end on some files that i have processed with Intuiteq.. some files it has lowered my low end below 250 to a great level and response and others it has not done quit as well.. stiil requires some tweaking to match responses.. any insight?
Terence
Terence
Re: Great Job guys.. Har-Bal 2.0 has done wonders... BUT
Terencekirrp63 wrote:Har-Bal has performed well beyound my expectations.. great tool.. but i have noticed a slight difference in th low end on some files that i have processed with Intuiteq.. some files it has lowered my low end below 250 to a great level and response and others it has not done quit as well.. stiil requires some tweaking to match responses.. any insight?
Terence
The broad rationale of IntuitQ is to smooth out the peaks and valleys of the dominant parts of the spectrum. By that I mean it only does processing between the points at which the spectrum rolls off at either end of the spectrum. You'll readily note this by looking at the frequency response of the filter it comes up with on a number of different tracks by the fact that the frequency limits of each track filter is different. Basically, it is attempting to minimise masking.
A word of advice. Don't look upon intuitQ as something that is "optimum" in an absolute sense. If you can tweak it further to obtain a better sound do so. It isn't intended to take you out of the process and disagreeing with what it did is fine. Like anything, it isn't perfect but it does pick up on some things that I wouldn't.
Paavo
Hi Terrance,
In a followup to that, The way Har-Bal attempts to figure out the frequency limits over which an adjustment should be made is not fool-proof. It works well about 95% of the time but occasionally gets it wrong. It all depends on the overall shape of the roll off at either end. When I coded it I envisaged a particular shape type to look for to find the roll off points and if it doesn't actually conform to that shape then it gets it wrong. In your case I would guess that it is choosing a low frequency limit that is too low which results in the bottom end being suppressed somewhat. People who do audio restoration work on acoustically/mechanically cut recordings will have noticed this behaviour. I plan to re-think the algorithm in future to better cater for these other spectrum shapes that it has difficulty with.
Regards,
Paavo.
In a followup to that, The way Har-Bal attempts to figure out the frequency limits over which an adjustment should be made is not fool-proof. It works well about 95% of the time but occasionally gets it wrong. It all depends on the overall shape of the roll off at either end. When I coded it I envisaged a particular shape type to look for to find the roll off points and if it doesn't actually conform to that shape then it gets it wrong. In your case I would guess that it is choosing a low frequency limit that is too low which results in the bottom end being suppressed somewhat. People who do audio restoration work on acoustically/mechanically cut recordings will have noticed this behaviour. I plan to re-think the algorithm in future to better cater for these other spectrum shapes that it has difficulty with.
Regards,
Paavo.
Not that i have Har-bal but i have seen the demo and i use it for mainly spectrum analysis and when it comes time to mastering i shall be purchasing.
Paavo, just a thought, would it not help maybe to implement a genre list that should be selected before applying the intuit Q button. This might narrow things down a touch.
eg = metal = then intuit Q.
or Jazz = intuit Q.
Paavo, just a thought, would it not help maybe to implement a genre list that should be selected before applying the intuit Q button. This might narrow things down a touch.
eg = metal = then intuit Q.
or Jazz = intuit Q.
It shouldn't be necessary. If you can fix the limit points accurately enough then it will work fine everytime. I just need to make the limit point interpolation a bit more sophisticated and robust. It is pretty good for a first try though. I really don't want to go down the route of trying to categorise a genre by spectrum. Personally, I don't think it is a valid approach as genre, like anything in art, is fluid and continually changing and it is highly dependent on instrumentation in any case. The current approach (when the limit points are found correctly) stays true to production intent and that is the way I'd like to keep it.
Cheers,
Paavo.
Cheers,
Paavo.
-
- Posts: 46
- Joined: Fri Nov 18, 2005 8:32 am
- Location: Scotland
- Contact:
hey Paavo i actually think intuit Q hit's the nail on the head almost every time, yeh sure it can mess up on a rare occasion e.g on hi mids occasionaly over balances them.
But all in all it's a great tool if you make some manual adjustments then hit the intuitQ 9 times out of ten it sounds great.
But all in all it's a great tool if you make some manual adjustments then hit the intuitQ 9 times out of ten it sounds great.
http://bitheadmastering.co.uk/
24 bit / 96 Khz Digital Audio Mastering
http://www.forumage.com/index.php?mforum=controlroom333
24 bit / 96 Khz Digital Audio Mastering
http://www.forumage.com/index.php?mforum=controlroom333
thanks Paavo
Over all the release does a great job.. well done!!! no more hours of fatiging attemps to eq.. As we all know to much time listening to the same song/recording makes it that much tougher...
Terence
Terence
-
- Posts: 40
- Joined: Sat Dec 04, 2004 7:18 pm
- Location: Arkansas
- Contact:
Paavo,
Maybe in a future release let the user specify a high and low freq limit for the application of IntuitQ.
Gordon
http://www.gidluckmastering.com
Maybe in a future release let the user specify a high and low freq limit for the application of IntuitQ.
Gordon
http://www.gidluckmastering.com
I did think about that and I think it should be available to the user. However, I didn't pursue it as I wanted to get something out pretty quick and a user adjustable option would require a GUI interface that would have slowed he release. Actually, I think I'd invisage the Gui for a "semi-manual" intuitQ along the lines of a click and drag frequency selection followed by the intuitQ smoothing when you release the mouse button.
Thanks for the ideal!
Paavo.
Thanks for the ideal!
Paavo.
I've had a good chance to try IntuitQ for a while now, and my general observation is that it seems to like to cut 1k-5k a little too much for my taste, and once in a while it will go crazy around 100-200 hz and boost too much - mostly on tracks that have no heavy percussion or drums. Sometimes though, it does make a good starting point from which to fine-tune things. I also seem to like the Air feature when it's around 10%-20%.
Paavo, is the EQ and Air processed in series (if so, then which is first) or parallel ?
Mike
Paavo, is the EQ and Air processed in series (if so, then which is first) or parallel ?
Mike
Hi Mike,
I'd agree with you. I'm aware of that tendancy to cut the mids too much and have been looking at possible improvements to the algorithm. Keep in mind that this is a first draft of the intuitQ feature and with more experience with the problem at hand and the behaviour of the current algorithm I can envisage significant improvements in successive revisions.
In any case, it is easy enough to add back any lost mid range caused by intuitQ. Here is my suggest approach for doing so. Enable the original spectrum trace and after hiting intuitQ use one parametric eq edit (with low Q) to make the new spectrum overlay the orignal as closely as possible. By that I mean you'll have parts in the original that are above the new and parts below. Try and make the parts above and the parts below equal in area (ie. area inscribed by the original and the modified spectrum). I've found that this gives back any loss of mid-range whilst preserving the smoothness of the intuitQ spectrum. Another way of looking at it is if you look at the frequency response view it should be evenly spread between boost and cut. Otherwise it will change the timbre.
Cheers,
Paavo.
I'd agree with you. I'm aware of that tendancy to cut the mids too much and have been looking at possible improvements to the algorithm. Keep in mind that this is a first draft of the intuitQ feature and with more experience with the problem at hand and the behaviour of the current algorithm I can envisage significant improvements in successive revisions.
In any case, it is easy enough to add back any lost mid range caused by intuitQ. Here is my suggest approach for doing so. Enable the original spectrum trace and after hiting intuitQ use one parametric eq edit (with low Q) to make the new spectrum overlay the orignal as closely as possible. By that I mean you'll have parts in the original that are above the new and parts below. Try and make the parts above and the parts below equal in area (ie. area inscribed by the original and the modified spectrum). I've found that this gives back any loss of mid-range whilst preserving the smoothness of the intuitQ spectrum. Another way of looking at it is if you look at the frequency response view it should be evenly spread between boost and cut. Otherwise it will change the timbre.
Cheers,
Paavo.
Paavo,HarBal wrote: Here is my suggest approach for doing so. Enable the original spectrum trace and after hiting intuitQ use one parametric eq edit (with low Q) to make the new spectrum overlay the orignal as closely as possible. By that I mean you'll have parts in the original that are above the new and parts below. Try and make the parts above and the parts below equal in area (ie. area inscribed by the original and the modified spectrum). I've found that this gives back any loss of mid-range whilst preserving the smoothness of the intuitQ spectrum. Another way of looking at it is if you look at the frequency response view it should be evenly spread between boost and cut. Otherwise it will change the timbre.
Cheers,
Paavo.
Could you please clarify this procedure a little more -- especially the statement -- "use one parametric eq edit (with low Q) to make the new spectrum overlay the orignal as closely as possible."
Thanks,
Tom
Take a look at these images. The first is a track after intuitQ but before the single edit. Look at the area shaded red (where the original is above the new spectrum) and the area shaded blue (where the original is below the new spectrum). In the first there is a lot more red than there is blue. In the second after the edit there is about the same amount of each. This is what you should be aiming for.
Regards,
Paavo.
http://www.taquis.com/harbal/intuitQ_before.png
http://www.taquis.com/harbal/intuitQ_after.png
Regards,
Paavo.
http://www.taquis.com/harbal/intuitQ_before.png
http://www.taquis.com/harbal/intuitQ_after.png