what do you think, guys. do the two plugins encode the same way for m/s? if yes, it would be easy to build your own m/s-mastering eq...
xenon codec:
http://www.planetz.com/forums/viewtopic ... 1&forum=16
Brainworx bx_digital and Neutrons Xenon Codec
ok, bx_digital may could be a decent and mastering optimized 5-band-eq, we don't know it yet, but is this worth the 600 EUR gap? to build a project and routing bx-like couldn't be that difficult, what do you think?On 2006-02-24 02:13, djmicron wrote:
The quality of brainworx eq's could be the real difference.
we may could do a side-by-side test when the bx_digital is delivered... I hope it'll be soon!!
greez
katano
If Brainworx has not writed any special routines, I belive there's not difficulties to do something similar for a lot of us here on the Z. Anyway, thinking on how many boards Cw have selled until now, I belive that here on PZ we are a little percentual of the people who can buy it.
But think about how much time doing the presets takes and how confortable is to have everything under a single panel. These are really strong points that could make the difference.
Maybe there's people (and maybe some professional men) who don't have enought skill and/or free time to build a custom project. I belive that these people are very happy to spend 600 euro to resolve all their problem at the same time.
Just my thoughts
_________________
Welcome to the dawning of a new empire
<font size=-1>[ This Message was edited by: Lima on 2006-02-24 10:19 ]</font>
But think about how much time doing the presets takes and how confortable is to have everything under a single panel. These are really strong points that could make the difference.
Maybe there's people (and maybe some professional men) who don't have enought skill and/or free time to build a custom project. I belive that these people are very happy to spend 600 euro to resolve all their problem at the same time.
Just my thoughts
_________________
Welcome to the dawning of a new empire
<font size=-1>[ This Message was edited by: Lima on 2006-02-24 10:19 ]</font>
If the best of the bx is the m/s encoding function, it's not my goal to pay 600 euros for it, because i already do the job by myself and i'm able to route the cables inside the sfp environment.ok, bx_digital may could be a decent and mastering optimized 5-band-eq, we don't know it yet, but is this worth the 600 EUR gap? to build a project and routing bx-like couldn't be that difficult, what do you think?
we may could do a side-by-side test when the bx_digital is delivered... I hope it'll be soon!!
greez
katano
I must listen to the sonic quality of the eq's before give an opinion about it....
Sure, trivial for anyone to do it!
You can simply get a 12000$ SPL PQ, or a 13000$ GML 9500, or a 12000$ Avalon AD2077. If you shop around you can probably find them for a bit cheaper!
Then you model it! Of course, the actual instruments to precisely model it might cost you twice that much. Or maybe you are lucky and you have a friend with $50000 of electronics equipment in his basement.
Then you can implement it in the Scope SDK. Let's see you do it! I'm pretty curious to see it.
You can simply get a 12000$ SPL PQ, or a 13000$ GML 9500, or a 12000$ Avalon AD2077. If you shop around you can probably find them for a bit cheaper!
Then you model it! Of course, the actual instruments to precisely model it might cost you twice that much. Or maybe you are lucky and you have a friend with $50000 of electronics equipment in his basement.
Then you can implement it in the Scope SDK. Let's see you do it! I'm pretty curious to see it.
I dunno. With the bx, you can load 2-3 of them and do surround sound production. With hardware units, you need 3 of them, so roughly $35000. So 600 Euros is actually pretty cheap.
You might not get 100% exactly the same thing as $36k worth of hardware, (i.e. hardware has crazy bandwidth of like 0-400khz) but I think the previous analog emulations on this platform speak for themselves as far as what is possible in terms of quality.
Just start saving your pennies and sell a couple of those rack boxes you don't use, and you can easily afford the bx unit.
BTW, about M/S encoding.. so if I encode a L/R signal to M/S.. and then EQ each M and S differently.. and then mix them back together.. what happens? Anyone wanna show me a simple phase-correction circuit for this? =P =P =P
You might not get 100% exactly the same thing as $36k worth of hardware, (i.e. hardware has crazy bandwidth of like 0-400khz) but I think the previous analog emulations on this platform speak for themselves as far as what is possible in terms of quality.
Just start saving your pennies and sell a couple of those rack boxes you don't use, and you can easily afford the bx unit.
BTW, about M/S encoding.. so if I encode a L/R signal to M/S.. and then EQ each M and S differently.. and then mix them back together.. what happens? Anyone wanna show me a simple phase-correction circuit for this? =P =P =P
You can also use this:On 2006-02-24 15:07, symbiote wrote:
BTW, about M/S encoding.. so if I encode a L/R signal to M/S.. and then EQ each M and S differently.. and then mix them back together.. what happens? Anyone wanna show me a simple phase-correction circuit for this? =P =P =P
http://www.planetz.com/forums/viewtopic ... forum=16&3
Thanks alfonso !! I'm gonna try this out today.
Symbiote, dunno if this might help <a href="http://www.musicdsp.org/showArchiveComm ... dsp.org</a>
Symbiote, dunno if this might help <a href="http://www.musicdsp.org/showArchiveComm ... dsp.org</a>
an amazing algorithm - almost unbeatable when it comes to demonstrate the problems of math precisionOn 2006-02-25 04:34, Shroomz wrote:
Symbiote, dunno if this might help <a href="http://www.musicdsp.org/showArchiveComm ... dsp.org</a>
(you may have your snooze now, Shroomz
)the phase correction factor U is calculated with 2 trig functions and divided by a 10 digit constant pi,
then 3 times multplied by itself, the intermediate subtracted from 1 and that set to a power of 24(!) as a final result
no need to question that the algorithm itself is correct, but you'll hardly find 2 different compilers which will yield identical results even on the same source - let alone that the math rules could be implemented in a dozen different ways.
it's not that simple...
cheers, Tom
there is no explanation in my post above
I just used the algo as an example how easy it is to bring a system with a finite number of digits into trouble.
the 'not that easy' doesn't refer to the math function itself, but that it's implementation can have side effects due to (inevitable) limits of precision.
(the writer of the algorithm itself cannot explain a certain behaviour)
let's just assume that this method of phase compensation is absolutely correct - then the respective implementation(!) of the processing can produce large deviations in the final result
I assume that the folks at brainworx did their homework - and I also assume that Analog Devices provided an excellent support for exactly this type of problem.
cheers, Tom
I just used the algo as an example how easy it is to bring a system with a finite number of digits into trouble.
the 'not that easy' doesn't refer to the math function itself, but that it's implementation can have side effects due to (inevitable) limits of precision.
(the writer of the algorithm itself cannot explain a certain behaviour)
let's just assume that this method of phase compensation is absolutely correct - then the respective implementation(!) of the processing can produce large deviations in the final result
I assume that the folks at brainworx did their homework - and I also assume that Analog Devices provided an excellent support for exactly this type of problem.
cheers, Tom
You explained part of the stereo separation algo I linked to. You explained it briefly in mathematical layman's terms
Anyway, as you say the variation potential in implementing anything like this (or any dsp code for that matter) is huge (it has to be).
It's like the moog filter variations on musicdsp.org. Exactly the same applies & although they are easily implemented in a variety of manners, most will need significant optimization & tweeking.
Anyway, as you say the variation potential in implementing anything like this (or any dsp code for that matter) is huge (it has to be).
It's like the moog filter variations on musicdsp.org. Exactly the same applies & although they are easily implemented in a variety of manners, most will need significant optimization & tweeking.
honestly - I have no the slightest idea what that thing does...
yet I could pick it up (say from such a link or a book) and implement it - so just take me as the average VSTI coder... anyone may guess what the result would be without careful range checking and optimizing accordingly - and that's just a small part of the processing of a device like the one in question here.
As a non-mathwizzard you can easily spend a whole month with the linked algo - if it shows that (standard)lib functions are insufficient, then a satisfying result may even be completely out of range.
But fortunately the engineers from AD have taken (at least a significat part of) that burden from our shoulders... hopefully
it's in fact like the Moog cascade, where you have to measure and match/select each single transistor
cheers, tom
yet I could pick it up (say from such a link or a book) and implement it - so just take me as the average VSTI coder... anyone may guess what the result would be without careful range checking and optimizing accordingly - and that's just a small part of the processing of a device like the one in question here.
As a non-mathwizzard you can easily spend a whole month with the linked algo - if it shows that (standard)lib functions are insufficient, then a satisfying result may even be completely out of range.
But fortunately the engineers from AD have taken (at least a significat part of) that burden from our shoulders... hopefully
it's in fact like the Moog cascade, where you have to measure and match/select each single transistor
cheers, tom
I don't know about the stereo separation code right now as I'm a total novice myself, but I've witnessed first hand the Moog Filter variations 1 & 2 from that code archive at musicdsp.org & they sound pretty damn good when implemented properly. So good in fact, that you'll find implementations/variatios of them in assembler under the bonnet of some of the most favoured DSP harware synths on the market (clavia, access etc)
Like you've said many a time tho, they DO sound different to SFP's moog style filters.
Like you've said many a time tho, they DO sound different to SFP's moog style filters.
I don't know why you linked to that article Shroomz, my question was 100% rhetorical and not really meant to be answered.
Phase correction is always function of the other devices in the signal chain that mess with the phase. There's no 100% universally-correct phase correction algorithm. You have to look at the full signal chain, pick out the devices that affect/change phase response, and design something that will correct that phase response back to it's original curve (or as close as possible.)
Analog and digital IIR filters will affect the phase response just like they affect amplitude/frequency response. This modification of the phase is generally proportionnal to the frequency attenuation/boost, and will depend on the number of poles and zeroes in the mathematical model, just like the frequency response.
The general way to correct this phase modification is to stick an all-pass filter in the signal chain, which has linear frequency response, but not linear phase response. In the case of an EQ with variable Q/freq bands/centers, this all-pass filter needs to be variable also, and provide phase-correction over a fairly wide range.
The fact that you want a linear frequency response will impose limits on the all-pass filter design, which is why it's generally known as a "minimum phase" configuration, and not full "linear phase."
Digital FIR filters have linear phase reponse. Although there will be a delay applied to the signal, the delay is equal at all frequencies. FIR tend to have other issues, like some ripple at/near the band transitions, which you can attenuate with windowing and large-enough FIR length and other dirty things. More or less relevant to the current situation though, just mentioning it because I like to ramble about weird arcane mathematechnical stuff (it's fun!)
In the case of M/S encoding, a sample delay like on Alfonso's device will correct the delay inherent to the plugin's algorithm, but won't correct anything else. If you load a pair of normal equalizers as inserts in that device, and EQ the M and S signals differently, you will definitely get some phase issues when you reconstruct the L and R signals from M and S. It might be very minimal/unnoticeable if the EQing is very slight/delicate, but it's definitely not something you want in a mastering situation.
Not to mention having to set the sample delay by hand every time might be a bit of a hassle, and might change with the EQ's parameter, e.g. when adding/removing peak/etc EQ in a parametric EQ, the total sample delay imposed by the plugin might change, forcing you to re-adjust the delays. It's a bit distracting in a mastering situation.
The easy way to fix this is to just use a minimum/linear phase EQ, but I don't know if there's such a device for the Scope platform right now (besides bx =P.) I have some FIR designs I have to test, but they're only really simple stuff for the moment, near-ideal low pass and high pass and crossovers etc., not full EQ.
I'm just saying, given the requirements of profesionnal mastering, you will want to pay special care to maintain phase coherence throughout the whole device/signal chain.
So yes, M/S encoding and decoding is the done the same way in Xenon and bx, but there's a LOT more that goes into a mastering EQ than just that.
Also notice how all the $12k hardware mastering EQs I mentionned have the M/S stuff. It's not what you will be paying 600 euros for.
Phase correction is always function of the other devices in the signal chain that mess with the phase. There's no 100% universally-correct phase correction algorithm. You have to look at the full signal chain, pick out the devices that affect/change phase response, and design something that will correct that phase response back to it's original curve (or as close as possible.)
Analog and digital IIR filters will affect the phase response just like they affect amplitude/frequency response. This modification of the phase is generally proportionnal to the frequency attenuation/boost, and will depend on the number of poles and zeroes in the mathematical model, just like the frequency response.
The general way to correct this phase modification is to stick an all-pass filter in the signal chain, which has linear frequency response, but not linear phase response. In the case of an EQ with variable Q/freq bands/centers, this all-pass filter needs to be variable also, and provide phase-correction over a fairly wide range.
The fact that you want a linear frequency response will impose limits on the all-pass filter design, which is why it's generally known as a "minimum phase" configuration, and not full "linear phase."
Digital FIR filters have linear phase reponse. Although there will be a delay applied to the signal, the delay is equal at all frequencies. FIR tend to have other issues, like some ripple at/near the band transitions, which you can attenuate with windowing and large-enough FIR length and other dirty things. More or less relevant to the current situation though, just mentioning it because I like to ramble about weird arcane mathematechnical stuff (it's fun!)
In the case of M/S encoding, a sample delay like on Alfonso's device will correct the delay inherent to the plugin's algorithm, but won't correct anything else. If you load a pair of normal equalizers as inserts in that device, and EQ the M and S signals differently, you will definitely get some phase issues when you reconstruct the L and R signals from M and S. It might be very minimal/unnoticeable if the EQing is very slight/delicate, but it's definitely not something you want in a mastering situation.
Not to mention having to set the sample delay by hand every time might be a bit of a hassle, and might change with the EQ's parameter, e.g. when adding/removing peak/etc EQ in a parametric EQ, the total sample delay imposed by the plugin might change, forcing you to re-adjust the delays. It's a bit distracting in a mastering situation.
The easy way to fix this is to just use a minimum/linear phase EQ, but I don't know if there's such a device for the Scope platform right now (besides bx =P.) I have some FIR designs I have to test, but they're only really simple stuff for the moment, near-ideal low pass and high pass and crossovers etc., not full EQ.
I'm just saying, given the requirements of profesionnal mastering, you will want to pay special care to maintain phase coherence throughout the whole device/signal chain.
So yes, M/S encoding and decoding is the done the same way in Xenon and bx, but there's a LOT more that goes into a mastering EQ than just that.
Also notice how all the $12k hardware mastering EQs I mentionned have the M/S stuff. It's not what you will be paying 600 euros for.
great read symbiote @+
I only posted the link because I remembered having spotted the center separation article & it included a closely related phase correction algo. Not much use directly, but I hadn't realiseed your question was purely retorical
I just thought it may show a 'possible' method (concept) of phase correction for mid separationn from a stereo signal, as I thought that's what the guy had realised. As astro pointed out, the method of implementing that is where the work is & where the men are separated from the boys so to speak. These basic algos could be implemented in dozens of different ways, some poorly, some masterfully. For any 600 Euro plug these days, you're expecting masterfull (pardon the pun) implementation
btw . Alfonso's Example M/S modular patch has potential as a processing tool (route the mid signal from before the insert also to output 3 etc )
<font size=-1>[ This Message was edited by: Shroomz on 2006-02-26 09:41 ]</font>
I only posted the link because I remembered having spotted the center separation article & it included a closely related phase correction algo. Not much use directly, but I hadn't realiseed your question was purely retorical
I just thought it may show a 'possible' method (concept) of phase correction for mid separationn from a stereo signal, as I thought that's what the guy had realised. As astro pointed out, the method of implementing that is where the work is & where the men are separated from the boys so to speak. These basic algos could be implemented in dozens of different ways, some poorly, some masterfully. For any 600 Euro plug these days, you're expecting masterfull (pardon the pun) implementation
btw . Alfonso's Example M/S modular patch has potential as a processing tool (route the mid signal from before the insert also to output 3 etc
) <font size=-1>[ This Message was edited by: Shroomz on 2006-02-26 09:41 ]</font>