kcatthedog
Temp
Super Helpful Dude
Posts: 15,214
Member is Online
|
Post by kcatthedog on Jul 1, 2019 14:08:36 GMT -6
Thx: all good! Am still settling into new retirement budget so actually trying not to spend unless absolutely necessary: there will be other sales !
Hope people enjoy their new tools!
ps happy Canada day eh!
|
|
|
Post by Guitar on Jul 1, 2019 19:45:48 GMT -6
I was in Georgia but I still managed to use the extra coupon for BX Console N, which I haven't installed or demo'd yet. Pretty excited. Some folks on other forums say this is up there with the UAD Neve 88RS, which is "insane." I will never buy that plugin. I swear on my own grave. The fact that you have to buy a thousand dollar dongle to even run a few channels of it in a mix. Anyway /UAD RANT OFF. I did a listening test someone posted of SSL E channels of Waves, UAD, BX, and Softube. Thought that the BX and Softube sounded almost identical. UAD sounds a little more mid punchy or something. Waves was far behind, but this is the old code, I don't know if the CLA MiX Hub might test better in a shootout. I guess I will have to set something up. As a final note, I think the Slate stuff sounds hard, pinched, and occasionally too bright. It never won any shootouts for me, so I never bought into it. I am talking about the compressors, EQ, and tape emulations. I haven't tried the new expanded stuff. Also looking at Valhalla Delay. I've got enough amp simulators already. oh and kcatthedog I wouldn't be too paranoid about "code." I don't even think an analog mix rack is even necessary at all, I don't think it makes any difference. I would want one mostly to impress "clients" and to be "legitimate" in a photo shoot or interview. There are so many other elements that are critical, this one is way down on the list. With the exception of "options" that don't exist inside the computer. A small theory that I can say is that modeling electrical components (resistors, capacitors, semiconductors and etc) is relatively simple from a mathematical perspective. This is how most of the "good" plugins are made these days. Digital is the future of audio whether we like it or not. More to the point thought it is the present. I know some analog heads are going to bristle at these statements, but I think it might help someone out there who is trying to prioritize where to spend their limited moneys to build up their studio kit. Instruments, speakers, microphones, preamps, interface, monitor controller all day long.
|
|
kcatthedog
Temp
Super Helpful Dude
Posts: 15,214
Member is Online
|
Post by kcatthedog on Jul 1, 2019 20:06:42 GMT -6
No probs, I run a hybrid system now, sold all my ua stuff, use symphony mkii 16x16,modded delta board, ob pres and 14 channels of different comps and other plugs while mixing.
Personally, I prefer the more visceral real gear feel and I still do a first mix without the delta and comps but have Audioscape on 2 bus and wa273eq on 2 bus as inserts, but when I then add in the delta and other ob comps, for me there is more weight and presence to the mix.
I do think real materials add something code doesn’t but that’s just me: to each their own!
|
|
|
Post by Guitar on Jul 1, 2019 20:59:16 GMT -6
No probs, I run a hybrid system now, sold all my ua stuff, use symphony mkii 16x16,modded delta board, ob pres and 14 channels of different comps and other plugs while mixing. Personally, I prefer the more visceral real gear feel and I still do a first mix without the delta and comps but have Audioscape on 2 bus and wa273eq on 2 bus as inserts, but when I then add in the delta and other ob comps, for me there is more weight and presence to the mix. I do think real materials add something code doesn’t but that’s just me: to each their own! I appreciate that sentiment. Something about the difference in the working methods produces different results. One of my favorite albums is Queens Of The Stone Age self titled recorded by Joe Barresi. Somehow I just can't even imagine the same album being made with a computer. So I get what you are saying 100%. I just listend to the new Raconteurs album mixed by Vance Powell (who has posted here before). Complete opposite idea. It sounds "Modern" in every way to me, even though he used his rack of doom or whatever he calls it. I felt like it might have well just have been a software mix. The point is you can use the tools in opposite directions.
|
|
|
Post by gouge on Jul 1, 2019 21:17:05 GMT -6
the modern sound I think comes from the high freq emphasis that a lot of plugs seem to introduce. some plugs just make stuff louder for no apparent reason. waves plugs in particular can make things a little harsh and not analogue
over time though I've managed to build a very small collection of plugs that don't do this to my ears. whilst I like mixing external that setup doesn't fit in my lounge room or in the car so I've come to a place i'm happy with which is good analogue front end followed by a very limited pallet of plugs that I trust. whilst they are not quite there with the good outboard mix chains they are close enough for me these days.
|
|
|
Post by indiehouse on Jul 22, 2019 8:46:32 GMT -6
I demoed the bx_Townhouse compressor yesterday, mainly because I've been regretting not buying it during the big sale, especially now that it those sales might not happen much anymore due to the new subscription model. I A/B'd it against the UAD G-Buss MKII. I gotta say that I prefer the UAD by a large margin. It's much, much wider sounding. The Townhouse feels a little choked in comparison and definitely brings everything more toward the center. The Townhouse is a bit darker (though has a tad more weight in the low mids), where the UAD brings a little sparkle/shine. I also feel the UAD brings a little more depth where things feel a little more flat with the Townhouse.
Just a few early thoughts. I bought the UAD MKII as part of the 3+1 sale last month, so I obviously want to like it better. But people are offloading their PA plugs pretty cheap second hand, and I could score the Townhouse for $29. But I think I'll still pass, even though that's pretty cheap.
|
|
|
Post by ragan on Jul 22, 2019 10:18:02 GMT -6
I demoed the bx_Townhouse compressor yesterday, mainly because I've been regretting not buying it during the big sale, especially now that it those sales might not happen much anymore due to the new subscription model. I A/B'd it against the UAD G-Buss MKII. I gotta say that I prefer the UAD by a large margin. It's much, much wider sounding. The Townhouse feels a little choked in comparison and definitely brings everything more toward the center. The Townhouse is a bit darker (though has a tad more weight in the low mids), where the UAD brings a little sparkle/shine. I also feel the UAD brings a little more depth where things feel a little more flat with the Townhouse. Just a few early thoughts. I bought the UAD MKII as part of the 3+1 sale last month, so I obviously want to like it better. But people are offloading their PA plugs pretty cheap second hand, and I could score the Townhouse for $29. But I think I'll still pass, even though that's pretty cheap. I demo’d the Townhouse during the sale and was also unimpressed by it.
|
|
|
Post by jcoutu1 on Jul 22, 2019 10:21:06 GMT -6
...Ampeg. Man, I'm digging it.
|
|
|
Post by matt@IAA on Jul 22, 2019 10:26:07 GMT -6
GuitarThe problem with the discrete approach (modeling components) is that even then it doesn't represent what goes on in the analog domain. Let's say you model the crap out of a resistor and you get the noise, distortion (however infinitely miniscule) etc perrrrrfectly. Guess what, vintage gear was 1% or 5% or worse tolerance. Same with transistors, lets say you have a BC182 perfect. Draw up a circuit in a sim that uses 5 and you get what would happen if you had 5 completely, utterly, 100% identical transistors. Ain't no way no how. Make matters worse, when you sim a wire, you get an ideal wire. When you sim ground as 0V you get 0V. If you don't put in parasitic losses, they aren't there. Then, to add more to the funsies, you gotta consider the nonlinearity that happens in the analog domain. Especially true for tubes, but solid state as well. I'm actually pretty high on the future of digital plugins, but not the way we're doing them now. Trying to measure and recreate by algorithm is gong to underrepresent reality every time. It's an ineffective way to do it, no matter how good it gets. On the other hand, I expect neural networks could make short work of this. Give it samples from real analog gear to work with (or better yet give it feedback with an actual piece to recreate) and let the computer imitate it by itself, machine learning style. Will we know what its doing and how its doing it? No. But I am sure computers will be better at learning how to be gear than we would be at programming them how to be gear. Just like they'r better at learning how to play chess or go than we are at teaching them (or at playing the games for that matter).
|
|
|
Post by Guitar on Jul 22, 2019 10:30:19 GMT -6
Townhouse is pretty specific I guess. It's not really an alternative to the normal SSL compressor, but almost a completely different thing. I have used it successfully on a few mixes now.
I'm really impressed with Console N this week. It's really beautiful on electric guitars, make them sound "expensive" for lack of a better word. The compressor is great.
Did a shootout with UAD Neve 88RS. After a few days of tweaking and listening, I vastly prefer the Console N over the Neve 88RS. The compressor is much better, and it just sounds bigger and more alive in general. I guess I will cross the 88RS off of my future wishlist.
Console N on drum bus also works really well. Really is a great alternative to the usual SSL channel strip.
bx_bassdude is still getting a lot of use as well.
|
|
|
Post by Guitar on Jul 22, 2019 10:32:57 GMT -6
Guitar The problem with the discrete approach (modeling components) is that even then it doesn't represent what goes on in the analog domain. Let's say you model the crap out of a resistor and you get the noise, distortion (however infinitely miniscule) etc perrrrrfectly. Guess what, vintage gear was 1% or 5% or worse tolerance. Same with transistors, lets say you have a BC182 perfect. Draw up a circuit in a sim that uses 5 and you get what would happen if you had 5 completely, utterly, 100% identical transistors. Ain't no way no how. Make matters worse, when you sim a wire, you get an ideal wire. When you sim ground as 0V you get 0V. If you don't put in parasitic losses, they aren't there. Then, to add more to the funsies, you gotta consider the nonlinearity that happens in the analog domain. Especially true for tubes, but solid state as well. I'm actually pretty high on the future of digital plugins, but not the way we're doing them now. Trying to measure and recreate by algorithm is gong to underrepresent reality every time. It's an ineffective way to do it, no matter how good it gets. On the other hand, I expect neural networks could make short work of this. Give it samples from real analog gear to work with (or better yet give it feedback with an actual piece to recreate) and let the computer imitate it by itself, machine learning style. Will we know what its doing and how its doing it? No. But I am sure computers will be better at learning how to be gear than we would be at programming them how to be gear. Just like they'r better at learning how to play chess or go than we are at teaching them (or at playing the games for that matter). Very interesting. I wonder if any of the coders, UAD for example, are modeling individual components to account for the tolerance differences like you said. Brainworx is solving this riddle with their "TMT" programs. It's pretty interesting to use. You push the button and it's kind of like rolling the dice.
|
|
|
Post by jcoutu1 on Jul 22, 2019 10:43:59 GMT -6
From Brainworx...
|
|
|
Post by matt@IAA on Jul 22, 2019 12:02:45 GMT -6
That's very cool - yes, an important thing to realize. I think that the variability between components is what makes the sound stage so much better with analog. In stereo, differences are actually good..in my mind its the same as a vocal double or hard-panning stereo guitars, although obviously not as drastic.
Just consider the long tail pair input for any op amp. By definition you're assuming the two components are matched - and that they stay matched with temperature drift and whatnot. That's the whole basis of the idea, to give common mode rejection and all. But no two transistors are identical. Multiply that by XX? op amps in a signal path in an old discrete console, and you get a kind of very difficult thing to model.
To do this right, and this would be trivial mathematically but kind of a pain I think - you'd also have to include any trimpots as well. Say you have a gain string that has a final trimmer to bring it to a specific gain or distortion.. if you have two matched channels with 5% resistors, you need two different trimpot settings each time you generate the resistor values based on the possible ranges. Again, not a difficult problem, but who knows if they're actually going through that.
I'm not sure anyone is taking the time or effort to model transistors or other semiconductors to get real behavioral responses. Like a diode in a compressor sidechain to give a knee effect. That's obviously current sensitive. Unless you're doing model upon model, or you have a really good model for the diode itself (including manufacturing tolerances) your overall model is only going to sound "right" around one operating point. Same for any semiconductor component...Which is why plugins usually sound good in some circumstances and not so good in others. And transformer saturation is a whole other kettle of fish as well, extremely level dependent.
That's why I think the future is just to bypass all that and say screw it - computer, learn how to mimic this. It doesn't even need to know what "this" is - just take input signal and output signal, then let the computer listen to what happens and copy it for a few billion cycles. After a while, it'll "learn" what "Neve" sounds like for any possible setting. No programming, no algorithm. Just outcome based feedback. I mean this seems like it would be trivial compared to strategy games, or image processing, what they're doing with video deepfakes and all, because there's actually a right and wrong answer. With an image, the computer has no clue what it's doing or what right and wrong are. You have to give it connections - like, say, this is a tree, and this is a tree, and this is a tree, and over millions and millions of samples the computer learns what a tree is and can draw it. You don't even need that much here. You just say, here's the real Neve analog sample, just freaking make what you output match what it does and calibrate yourself by %error between the two signals. Done.
|
|
|
Post by ragan on Jul 22, 2019 12:13:08 GMT -6
That's very cool - yes, an important thing to realize. I think that the variability between components is what makes the sound stage so much better with analog. In stereo, differences are actually good..in my mind its the same as a vocal double or hard-panning stereo guitars, although obviously not as drastic. Just consider the long tail pair input for any op amp. By definition you're assuming the two components are matched - and that they stay matched with temperature drift and whatnot. That's the whole basis of the idea, to give common mode rejection and all. But no two transistors are identical. Multiply that by XX? op amps in a signal path in an old discrete console, and you get a kind of very difficult thing to model. To do this right, and this would be trivial mathematically but kind of a pain I think - you'd also have to include any trimpots as well. Say you have a gain string that has a final trimmer to bring it to a specific gain or distortion.. if you have two matched channels with 5% resistors, you need two different trimpot settings each time you generate the resistor values based on the possible ranges. Again, not a difficult problem, but who knows if they're actually going through that. I'm not sure anyone is taking the time or effort to model transistors or other semiconductors to get real behavioral responses. Like a diode in a compressor sidechain to give a knee effect. That's obviously current sensitive. Unless you're doing model upon model, or you have a really good model for the diode itself (including manufacturing tolerances) your overall model is only going to sound "right" around one operating point. Same for any semiconductor component...Which is why plugins usually sound good in some circumstances and not so good in others. And transformer saturation is a whole other kettle of fish as well, extremely level dependent. That's why I think the future is just to bypass all that and say screw it - computer, learn how to mimic this. It doesn't even need to know what "this" is - just take input signal and output signal, then let the computer listen to what happens and copy it for a few billion cycles. After a while, it'll "learn" what "Neve" sounds like for any possible setting. No programming, no algorithm. Just outcome based feedback. I mean this seems like it would be trivial compared to strategy games, or image processing, what they're doing with video deepfakes and all, because there's actually a right and wrong answer. With an image, the computer has no clue what it's doing or what right and wrong are. You have to give it connections - like, say, this is a tree, and this is a tree, and this is a tree, and over millions and millions of samples the computer learns what a tree is and can draw it. You don't even need that much here. You just say, here's the real Neve analog sample, just freaking make what you output match what it does and calibrate yourself by %error between the two signals. Done. You just like quadrupled my interest in the machine learning class I’ve got coming up.
|
|
kcatthedog
Temp
Super Helpful Dude
Posts: 15,214
Member is Online
|
Post by kcatthedog on Jul 22, 2019 12:25:09 GMT -6
Or just mod a delta with better op amps, etc., and run all your channels through it while mixing: its so organic !
Wiz warned me I’d get disinterested in new plug ins and he was right. I use key good ones ( fab filter, etc.) and I try drmo’s, often initially they seem better, but I hardly ever buy now.
|
|