|
Post by gouge on Sept 23, 2021 15:51:47 GMT -6
Anybody want to name a few eq plugs that don’t have this issue? So all over samples plugs don’t have cramping? i beleive acustica plugs dont. maybe someone more knowledgeable can confirm.
|
|
|
Post by earlevel on Sept 23, 2021 18:11:46 GMT -6
"Cramping" is a non-issue for music. I'm putting this bluntly, not to disagree with people who feel otherwise (happy to let people do what they want), but just to encourage people to not spend time chasing things that are meaningless to what they're trying to achieve. But yes, I do agree that if a plugin claims to model a particular piece of hardware, especially at the component level, it should be expected to give the correct response. If it's got a zero at Nyquist (which causes the cramping), it's ludicrous to say it's modeling all the electronic components. (I doubt that kind of claim is ever completely truthful.) So...maybe a little about so-called cramping, why it exists. You can try this yourself, and I think it will help it make more sense than just me showing charts. Pull up this page: www.earlevel.com/main/2021/09/02/biquad-calculator-v3/Leave the defaults, and adjust only the Fc (Hz) field. Type in 10, enter. The lowpass response has moved to lower frequency, but looks pretty much the same as the default 100 Hz. Now try 1000. Instead of a straight 12 dB/oct lowpass drop-off, the response dip accelerates near 20 kHz. This is due to the filter design. A second-order lowpass derived from an analog lowpass prototype via the bilinear z transform assumes that the goal is, well, to effectively be a lowpass filter. Such a filter would naturally have its greatest attenuation at a frequency of infinity. But a sampled signal can only go to half the sample rate, so the transform squishes an infinite response in an accelerated manner (using the Tan function). (There are other ways to design filters, it's just the choice of this generic filter design.) In doing so, it places a "zero" at half the sample rate, which is basically a digital black hole at that frequency. So, what horrible thing happened to our musical input by rolling off faster than expected near half the sample rate? Considering there isn't much change till the signal is already -60 dB, for frequencies around 20 kHz—in music, where there is very little energy near 20 kHz and certainly far louder lower frequencies going on at the same time—it's doubtful anyone could pick out that the lowpass is "just a little too good" near half the sample rate. But the phenomenon is probably more associated with peaking filters. Change the filter type to "peak". (You can click the chart to hide the phase display and view only frequency.) Increase the Gain (dB) field to 20 to exaggerate the effect. Set Fc to 100. Now to 10000. The right size is squished, because half the sample rate is force to be at 0 dB. Again, what is the musical impact? Well, that's an extreme setting—20 dB boost, with a wide Q—probably quite a bit more than you're use for "air", but still, the discrepancy is up where there is little audio content, it's competing with lower musical content that's must louder, and we doesn't hear well around 20k. I don't think many will listen and say, "hey, I can tell it's about 0 dB boost at 20k, I reckon it should be close to 10 dB!" IF SO...remember that this is digital audio, so when it's played out a DAC, it going to be lowpassed so that everything is GONE by half the sample rate (or close enough for rock and roll). In other words, it's going to get "cramped" whether you want it or not (though probably confined closer to half the sample rate). Also, I should point out that there is no cramping in high shelf filters, so if you're nervous get your "air" with a shelf
|
|
|
Post by ragan on Sept 23, 2021 18:43:57 GMT -6
"Cramping" is a non-issue for music. I'm putting this bluntly, not to disagree with people who feel otherwise (happy to let people do what they want), but just to encourage people to not spend time chasing things that are meaningless to what they're trying to achieve. But yes, I do agree that if a plugin claims to model a particular piece of hardware, especially at the component level, it should be expected to give the correct response. If it's got a zero at Nyquist (which causes the cramping), it's ludicrous to say it's modeling all the electronic components. (I doubt that kind of claim is ever completely truthful.) So...maybe a little about so-called cramping, why it exists. You can try this yourself, and I think it will help it make more sense than just me showing charts. Pull up this page: www.earlevel.com/main/2021/09/02/biquad-calculator-v3/Leave the defaults, and adjust only the Fc (Hz) field. Type in 10, enter. The lowpass response has moved to lower frequency, but look pretty much the same as the default 100 Hz. Now try 1000. Instead of a straight 12 dB/oct lowpass drop-off, the response dip accelerates near 20 kHz. This is due to the filter design. A second-order lowpass derived from an analog lowpass prototype via the bilinear z transform assumes that the goal is, well, to effectively be a lowpass filter. Such a filter would naturally have its greatest attenuation at a frequency of infinity. But a sampled signal can only go to half the sample rate, so the transform squishes an infinite response in an accelerated manner (using the Tan function). (There are other ways to design filters, it's just the choice of this generic filter design.) In doing so, it places a "zero" at half the sample rate, which is basically a digital black hole at the frequency. So, what horrible thing happened to our musical input by rolling off faster than expected near half the sample rate? Considering there isn't much change till the signal is already -60 dB, for frequencies around 20 kHz, in music, where there is very little energy near 20 kHz and certainly far louder lower frequencies going on at the same time, it's doubtful anyone could pick out that the lowpass is "just a little too good" near half the same rate. But the phenomenon is probably more associated with peaking filters. Change the filter type to "peak". (You can click the chart to hide the phase display and view only frequency.) Increase the Gain (dB) field to 20 to exaggerate the effect. Set Fc to 100. Now to 10000. The right size is squished, because half the sample rate is force to be at 0 dB. Again, what is the musical impact? Well, that's an extreme setting—20 dB boost, with a wide Q—probably quite a bit more than you're use for "air", but still, the discreapancy is up where there is little audio content, it's competing with lower musical content that's must louder, and we doesn't hear well around 20k. I don't think many will listen and say, "hey, I can tell it's about 0 dB boost at 20k, I reckon it should be close to 10 dB!" IF SO...remember that this is digital audio, so when it's played out a DAC, it going to be lowpassed so that everything is GONE by half the sample rate (or close enough for rock and roll). In other words, it's going to get "cramped" whether you want it or not (though probably confined closer to half the sample rate). Also, I should point out that there is no cramping in high shelf filters, so if you're nervous get your "air" with a shelf Always refreshing to get an actual DSP dev's input, rather than the usual armchair stuff. Thanks for chiming in, Nigel!
|
|
|
Post by gouge on Sept 24, 2021 4:14:12 GMT -6
to be honest this has been a facinating conversation.
i cant help but interperate this all to be more evidence of why digital sounds. well digital and analogue sounds, well analogue.
and to that point why hi res sounds better than low res..
|
|
|
Post by mrholmes on Sept 24, 2021 5:11:27 GMT -6
"Cramping" is a non-issue for music. I'm putting this bluntly, not to disagree with people who feel otherwise (happy to let people do what they want), but just to encourage people to not spend time chasing things that are meaningless to what they're trying to achieve. But yes, I do agree that if a plugin claims to model a particular piece of hardware, especially at the component level, it should be expected to give the correct response. If it's got a zero at Nyquist (which causes the cramping), it's ludicrous to say it's modeling all the electronic components. (I doubt that kind of claim is ever completely truthful.) So...maybe a little about so-called cramping, why it exists. You can try this yourself, and I think it will help it make more sense than just me showing charts. Pull up this page: www.earlevel.com/main/2021/09/02/biquad-calculator-v3/Leave the defaults, and adjust only the Fc (Hz) field. Type in 10, enter. The lowpass response has moved to lower frequency, but looks pretty much the same as the default 100 Hz. Now try 1000. Instead of a straight 12 dB/oct lowpass drop-off, the response dip accelerates near 20 kHz. This is due to the filter design. A second-order lowpass derived from an analog lowpass prototype via the bilinear z transform assumes that the goal is, well, to effectively be a lowpass filter. Such a filter would naturally have its greatest attenuation at a frequency of infinity. But a sampled signal can only go to half the sample rate, so the transform squishes an infinite response in an accelerated manner (using the Tan function). (There are other ways to design filters, it's just the choice of this generic filter design.) In doing so, it places a "zero" at half the sample rate, which is basically a digital black hole at that frequency. So, what horrible thing happened to our musical input by rolling off faster than expected near half the sample rate? Considering there isn't much change till the signal is already -60 dB, for frequencies around 20 kHz—in music, where there is very little energy near 20 kHz and certainly far louder lower frequencies going on at the same time—it's doubtful anyone could pick out that the lowpass is "just a little too good" near half the sample rate. But the phenomenon is probably more associated with peaking filters. Change the filter type to "peak". (You can click the chart to hide the phase display and view only frequency.) Increase the Gain (dB) field to 20 to exaggerate the effect. Set Fc to 100. Now to 10000. The right size is squished, because half the sample rate is force to be at 0 dB. Again, what is the musical impact? Well, that's an extreme setting—20 dB boost, with a wide Q—probably quite a bit more than you're use for "air", but still, the discrepancy is up where there is little audio content, it's competing with lower musical content that's must louder, and we doesn't hear well around 20k. I don't think many will listen and say, "hey, I can tell it's about 0 dB boost at 20k, I reckon it should be close to 10 dB!" IF SO...remember that this is digital audio, so when it's played out a DAC, it going to be lowpassed so that everything is GONE by half the sample rate (or close enough for rock and roll). In other words, it's going to get "cramped" whether you want it or not (though probably confined closer to half the sample rate). Also, I should point out that there is no cramping in high shelf filters, so if you're nervous get your "air" with a shelf
THX that's why I do the interview with one of my favorite dev. There is so much I still don't fully understand about digital audio, or I just miss thinking logically...
|
|
ericn
Temp
Balance Engineer
Posts: 14,809
Member is Online
|
Post by ericn on Sept 24, 2021 10:35:12 GMT -6
"Cramping" is a non-issue for music. I'm putting this bluntly, not to disagree with people who feel otherwise (happy to let people do what they want), but just to encourage people to not spend time chasing things that are meaningless to what they're trying to achieve. But yes, I do agree that if a plugin claims to model a particular piece of hardware, especially at the component level, it should be expected to give the correct response. If it's got a zero at Nyquist (which causes the cramping), it's ludicrous to say it's modeling all the electronic components. (I doubt that kind of claim is ever completely truthful.) So...maybe a little about so-called cramping, why it exists. You can try this yourself, and I think it will help it make more sense than just me showing charts. Pull up this page: www.earlevel.com/main/2021/09/02/biquad-calculator-v3/Leave the defaults, and adjust only the Fc (Hz) field. Type in 10, enter. The lowpass response has moved to lower frequency, but looks pretty much the same as the default 100 Hz. Now try 1000. Instead of a straight 12 dB/oct lowpass drop-off, the response dip accelerates near 20 kHz. This is due to the filter design. A second-order lowpass derived from an analog lowpass prototype via the bilinear z transform assumes that the goal is, well, to effectively be a lowpass filter. Such a filter would naturally have its greatest attenuation at a frequency of infinity. But a sampled signal can only go to half the sample rate, so the transform squishes an infinite response in an accelerated manner (using the Tan function). (There are other ways to design filters, it's just the choice of this generic filter design.) In doing so, it places a "zero" at half the sample rate, which is basically a digital black hole at that frequency. So, what horrible thing happened to our musical input by rolling off faster than expected near half the sample rate? Considering there isn't much change till the signal is already -60 dB, for frequencies around 20 kHz—in music, where there is very little energy near 20 kHz and certainly far louder lower frequencies going on at the same time—it's doubtful anyone could pick out that the lowpass is "just a little too good" near half the sample rate. But the phenomenon is probably more associated with peaking filters. Change the filter type to "peak". (You can click the chart to hide the phase display and view only frequency.) Increase the Gain (dB) field to 20 to exaggerate the effect. Set Fc to 100. Now to 10000. The right size is squished, because half the sample rate is force to be at 0 dB. Again, what is the musical impact? Well, that's an extreme setting—20 dB boost, with a wide Q—probably quite a bit more than you're use for "air", but still, the discrepancy is up where there is little audio content, it's competing with lower musical content that's must louder, and we doesn't hear well around 20k. I don't think many will listen and say, "hey, I can tell it's about 0 dB boost at 20k, I reckon it should be close to 10 dB!" IF SO...remember that this is digital audio, so when it's played out a DAC, it going to be lowpassed so that everything is GONE by half the sample rate (or close enough for rock and roll). In other words, it's going to get "cramped" whether you want it or not (though probably confined closer to half the sample rate). Also, I should point out that there is no cramping in high shelf filters, so if you're nervous get your "air" with a shelf Sure just pop in and bring a sane explanation! Dang it don’t you know this is the internet!
|
|
|
Post by Guitar on Sept 27, 2021 11:08:11 GMT -6
In the Dan Worall test about cramping, one of his EQ videos, I preferred the cramped EQ (ReaEQ) in the blind A/B test. Haha! It sounded better to me.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 27, 2021 11:57:29 GMT -6
In the Dan Worall test about cramping, one of his EQ videos, I preferred the cramped EQ (ReaEQ) in the blind A/B test. Haha! It sounded better to me. ReaEQ is cleaner than Fabfilter Q. He should have chosen similar technical eqs and matched curves close as possible. ReaEQ vs Crave or Oxford would’ve been better.
|
|
|
Post by gouge on Sept 27, 2021 16:16:38 GMT -6
when i first started using reaper i liked the stock plugs...
but after a while i started to not like the sound. there is a smear or something i associated with digital i didn't like.
now im using acustica with no complaints.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 27, 2021 19:13:28 GMT -6
when i first started using reaper i liked the stock plugs... but after a while i started to not like the sound. there is a smear or something i associated with digital i didn't like. now im using acustica with no complaints. The non-linear stock plugs alias. The compressor is also a total linear grotbox. Explosive lofi pump. Not great for anything else.
|
|