- Thread Starter
- #21
I do use that word to include rendering of IIR not just FIR filters
and yes I will use a video player with adjustable delay to resolve lip-sync issues
Last edited:
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Analog crossovers, only using DSP to treat issues the analog system cannot fix
- Thread starter john61ct
- Start date
and yes I will use a video player with adjustable delay to resolve lip-sync issues
- Thread Starter
- #22
As I stated, the testing / learning process is the primary goal, not any one resulting implementation.
My budget is $200 per month, with credit getting abused if it helps accelerate that process, or if I spot crazy good bargains I think unlikely to be available in future.
Incremental investments, nothing over say $700 at once
My budget is $200 per month, with credit getting abused if it helps accelerate that process, or if I spot crazy good bargains I think unlikely to be available in future.
Incremental investments, nothing over say $700 at once
If you really want to learn I would recommend you download VituixCAD and the documentation. The documents will walk you through the measurement process, which needs to be done outdoors both on and off axis. Then you enter the data into the program and start simulating and designing crossovers (analog and DSP are supported). The learning curve is steep and it is very time consuming but you will learn a lot and end up with a much better solution than trying to use fatally flawed "in room" measurements and trial and error and guessing. You can even simulate what the response differences are for analog vs DSP crossovers and or design a completely passive crossover with component values. Good luck and enjoy.
If you do it for long enough, the cost of experiment might exceed the cost of solution. Under your conditions, I really have no advice to provide.
Well I wish you luck in whatever you are trying to achieve.
With a limited budget, focus on DSP only. There is plenty to discover on that alone.
Use CamilaDSP and a multichannel DAC, some cheap amps, measurement mic, and start experimenting!
Use CamilaDSP and a multichannel DAC, some cheap amps, measurement mic, and start experimenting!
- Thread Starter
- #26
Thanks right up my alley.
I'd heard that is a great tool for targeting per-speaker compensation EQ
> based on as anechoic measurements as possible
but did not realise also well suited for modeling crossovers between enclosures including 2- 3-way speakers.
Do you reco it over say RePhase +REW for filter creation?
- Thread Starter
- #27
Which is fine, walking around the world is more expensive than a cruise holiday.
In this case the journey is the point not the destination
It’s not either/or. You can use them in tandem.
RePhase and VituixCAD are not FOSS, neither is the OS they run on
RePhase runs in wine, though. Did not have much luck with VituixCAD.- Thread Starter
- #29
Thank you for that advice. Being poor is just a momentum-limiting factor, just saving money is not a design goal
I very strongly desire to start with analog, even if that costs more long-term.
But yes I will use your approach for the DSP learning part, such a big learning curve, to be implemented in a modular fashion as needed, or as it is actually tested as superior.
- Thread Starter
- #30
Yes I'm not going to limit myself, it's just a preference.It’s not either/or. You can use them in tandem.
RePhase and VituixCAD are not FOSS, neither is the OS they run on
However I do hate WINE, with the burning intensity of a thousand suns
If you really want to go down that rabbit hole, look at something like this:
dbx 223xs Crossover Review
This is a review and detailed measurements of the dbx 223xs active analog balanced crossover. It is on kind loan from a member and costs US $245. I must say this is a nice and sturdy package for such a bargain price. Controls feel nice as well. Functionality though is confusing given the...
www.audiosciencereview.com
Not crazy expensive and transparent enough.
If you really want to have some fun for little money, check out this project:
Introducing DSPi | A powerful, user friendly and open source DSP for less than a cup of coffee
For the longest time, I have been in search of the “perfect” inexpensive audio DSP device. Something with user friendly control software that could accept a USB input, apply PEQ, delays and other essential processing and then produce a digital output. Over the holidays, I was (un)fortunate...
www.audiosciencereview.com
Keith_W
Master Contributor
Re: applying DSP loudspeaker correction to a multi-driver loudspeaker that has a passive crossover in place. The question is whether you can obtain a measurement of the minimum-phase response of your loudspeaker alone. If you can do that, you can DSP all your like. But obtaining that measurement is more difficult than it may seem.
Firstly, a passive XO makes it difficult to measure each driver individually. A driver is minimum-phase, a crossover is minimum-phase. Put them in series, and the result is a convolution (multiplication) of driver and XO. In a multiplication operation, a zero multiplied by a nonzero value is zero, so the result is guaranteed to be minimum-phase. OTOH, two minimum-phase drivers playing together is a summation operation - and a zero summed with a nonzero value is nonzero, so the result may or may not be minimum-phase. So the question as to whether this measured response can be successfully inverted very much depends on your loudspeaker.
Secondly, a multi-driver speaker is difficult to measure, if you need to measure the whole thing at once. This is particularly so if you have a speaker with woofers close to the floor or panel speakers - since the floor reflection can not be removed unless the speaker is elevated. There is usually a low frequency limit below which it is impossible to obtain an anechoic measurement of your loudspeaker. EVERYBODY encounters this problem, a 20Hz wavelength is 17.2m long, and you need to elevate the speaker to the 5-6th floor of an apartment to get that high! So alternative measurement methods are available. If you look at my eBook (link in my sig) there is a way to determine the lower freq limit of your measurement.
Fortunately, as hobbyists, we correct the bass together with the room, so it is NOT NECESSARY to obtain an anechoic measurement down to 20Hz. I would suggest you determine your transition frequency (4x Schroder) via the calculation given in the book. Let's say it is 400Hz. Above 400Hz, you can correct your loudspeaker if you can obtain an anechoic measurement down to 400Hz. Below that, correct the speaker together with the room (i.e. mic at listening position).
Is it worth it? It depends on your speaker. If it has a horribly designed crossover, for sure you would benefit from correcting it. But if the XO was fine in the first place, you won't be improving it by much. So it is very much a YMMV situation.
Firstly, a passive XO makes it difficult to measure each driver individually. A driver is minimum-phase, a crossover is minimum-phase. Put them in series, and the result is a convolution (multiplication) of driver and XO. In a multiplication operation, a zero multiplied by a nonzero value is zero, so the result is guaranteed to be minimum-phase. OTOH, two minimum-phase drivers playing together is a summation operation - and a zero summed with a nonzero value is nonzero, so the result may or may not be minimum-phase. So the question as to whether this measured response can be successfully inverted very much depends on your loudspeaker.
Secondly, a multi-driver speaker is difficult to measure, if you need to measure the whole thing at once. This is particularly so if you have a speaker with woofers close to the floor or panel speakers - since the floor reflection can not be removed unless the speaker is elevated. There is usually a low frequency limit below which it is impossible to obtain an anechoic measurement of your loudspeaker. EVERYBODY encounters this problem, a 20Hz wavelength is 17.2m long, and you need to elevate the speaker to the 5-6th floor of an apartment to get that high! So alternative measurement methods are available. If you look at my eBook (link in my sig) there is a way to determine the lower freq limit of your measurement.
Fortunately, as hobbyists, we correct the bass together with the room, so it is NOT NECESSARY to obtain an anechoic measurement down to 20Hz. I would suggest you determine your transition frequency (4x Schroder) via the calculation given in the book. Let's say it is 400Hz. Above 400Hz, you can correct your loudspeaker if you can obtain an anechoic measurement down to 400Hz. Below that, correct the speaker together with the room (i.e. mic at listening position).
Is it worth it? It depends on your speaker. If it has a horribly designed crossover, for sure you would benefit from correcting it. But if the XO was fine in the first place, you won't be improving it by much. So it is very much a YMMV situation.
Last edited:
- Thread Starter
- #33
It's going to take me some study time to try to understand much of that, but thanks.
The fact that much of the system will be mobile, broken down and re-set up in very different environments
makes things harder.
I am hoping the per-speaker compensation EQ "as anechoic as possible" and the crossovers part could be saved, locked in as profiles / .WAV separately
from the room compensation filtering
maybe even implemented by separate devices in the signal chain, the former further down in the multi-channels, just before the power amps, the latter up closer to the pre-amp where the signal is still stereo.
The fact that much of the system will be mobile, broken down and re-set up in very different environments
makes things harder.
I am hoping the per-speaker compensation EQ "as anechoic as possible" and the crossovers part could be saved, locked in as profiles / .WAV separately
from the room compensation filtering
maybe even implemented by separate devices in the signal chain, the former further down in the multi-channels, just before the power amps, the latter up closer to the pre-amp where the signal is still stereo.
Digital_Thor
Addicted to Fun and Learning
No matter what you have now...
1. Learn to measure
2. Learn what measuments show
3. Understand what any type of EQ can and can't change - no matter whether it's passive or active.
Fiddle all you want... Good times well spent - in your own mind - can never return badly.
Further... We absolutely have to make mistakes, to learn.
1. Learn to measure
2. Learn what measuments show
3. Understand what any type of EQ can and can't change - no matter whether it's passive or active.
Fiddle all you want... Good times well spent - in your own mind - can never return badly.
Further... We absolutely have to make mistakes, to learn.
My PC-DSP-based multichannel multi-SP-driver multiamplifier full-active audio setup (the latest system detail can be found #931 and #1,009 on my project thread), where I utilize optimal combination of relative gain controls plus EQs in upstream digital domain and relative gain control in analog domain (using multiple "integrated" amplifiers), would be of your reference and interest, I assume.
If you would like to try DSP controls in upstream digital domain with keeping your passive crossover network in your 3-way (or 4-way) SP system, yes,
as I wrote here, you may start trying quasi/sham-multichannel setup using PC-DSP (or Mac-DSP) and single stereo USB DAC feeding into single stereo amplifier to drive 3-way or 4-way SP system in passive LC(R) network configuration.
I wrote there;
If you can use PC (Windows or Mac) and music player software as your local and/or streaming "music source", you may tentatively "test and try" the benefits of DSP software (like EKIO for Windows which I have been using all the way through) even with single stereo USB-DAC, single stereo amplifier and ordinary 2-way or three-way passive speaker having passive LCR crossover network in it. By "benefits of DSP software", I mean various crossover configurations for multiple Fq zones, group delay, relative gain tuning/simulation between the SP drivers, phase/polarity inversion(s), various EQs, and so on.
Even though many people criticized me that such quasi/sham-multichannel setup would be quite unusual, useless and out-of-the-scope and very-much-outlaw from true active multichannel multi-amplifier approach, I actually and intensively tested and evaluated this preliminary approach prior to going into my "royal road" towards multichannel multi-amplifier fully active exploration (ref. my project thread here); I have learned a lot from such preliminary primitive studies and experiences.
Even though within the upstream DSP, of course, you can apply any kind of XOs, Gains, EQs, Group-Delays, Phases, Polarity-Inversions, etc., the critical issue would be how you would objectively measure and subjectively optimize the DSP tunings in almost-anechoic space as well as in your actual listening room acoustic environments, as many other people have already pointed.
If you would like to try DSP controls in upstream digital domain with keeping your passive crossover network in your 3-way (or 4-way) SP system, yes,
as I wrote here, you may start trying quasi/sham-multichannel setup using PC-DSP (or Mac-DSP) and single stereo USB DAC feeding into single stereo amplifier to drive 3-way or 4-way SP system in passive LC(R) network configuration.
I wrote there;
If you can use PC (Windows or Mac) and music player software as your local and/or streaming "music source", you may tentatively "test and try" the benefits of DSP software (like EKIO for Windows which I have been using all the way through) even with single stereo USB-DAC, single stereo amplifier and ordinary 2-way or three-way passive speaker having passive LCR crossover network in it. By "benefits of DSP software", I mean various crossover configurations for multiple Fq zones, group delay, relative gain tuning/simulation between the SP drivers, phase/polarity inversion(s), various EQs, and so on.
Even though many people criticized me that such quasi/sham-multichannel setup would be quite unusual, useless and out-of-the-scope and very-much-outlaw from true active multichannel multi-amplifier approach, I actually and intensively tested and evaluated this preliminary approach prior to going into my "royal road" towards multichannel multi-amplifier fully active exploration (ref. my project thread here); I have learned a lot from such preliminary primitive studies and experiences.
Even though within the upstream DSP, of course, you can apply any kind of XOs, Gains, EQs, Group-Delays, Phases, Polarity-Inversions, etc., the critical issue would be how you would objectively measure and subjectively optimize the DSP tunings in almost-anechoic space as well as in your actual listening room acoustic environments, as many other people have already pointed.
Last edited:
- Thread Starter
- #36
Yes I am building a powerful "Audiophile PC" exactly for the purpose of centrally playing with DSP most conveniently for learning purposes.
with 36 super-clean balanced / unbalanced 1/4" TRS/TR / XLR combo I/O ports and near zero latency studio grade AD/DA conversion
Also Raspberry Pi based at some point.
However, for a day to day system once "in production", I am not interested in having "the player node" (renderer) be that PC, except maybe occasionally
The signal stream from all sources, analog or digital, must have the same processing available, through standalone pass-through filtering.
Furthermore, I am trying to also work toward a layered, modular approach, so
the ad-hoc user EQ (maybe analog)
the room compensation
the per-speaker "as anechoic as possible" compensation EQ
both needing DSP? but maybe each kept separate, if not convolvers at each layer, at least the filter creation / testing measurement iterations are kept separate, learning to use the SOTA tools for each function
even say, the phase / timing delay tweaking done separately from the frequency side of bandpass / crossover filtering.
Maybe the latter kept analog, for some bits even passive, while the former requires DSP
If this turns out to all be unrealistic or just not worth doing, then I "throw up my hands" and fall back to keeping that expensive monolithic convolver running all the time
but that really rubs me the wrong way from where I stand currently.
with 36 super-clean balanced / unbalanced 1/4" TRS/TR / XLR combo I/O ports and near zero latency studio grade AD/DA conversion
Also Raspberry Pi based at some point.
However, for a day to day system once "in production", I am not interested in having "the player node" (renderer) be that PC, except maybe occasionally
The signal stream from all sources, analog or digital, must have the same processing available, through standalone pass-through filtering.
Furthermore, I am trying to also work toward a layered, modular approach, so
the ad-hoc user EQ (maybe analog)
the room compensation
the per-speaker "as anechoic as possible" compensation EQ
both needing DSP? but maybe each kept separate, if not convolvers at each layer, at least the filter creation / testing measurement iterations are kept separate, learning to use the SOTA tools for each function
even say, the phase / timing delay tweaking done separately from the frequency side of bandpass / crossover filtering.
Maybe the latter kept analog, for some bits even passive, while the former requires DSP
If this turns out to all be unrealistic or just not worth doing, then I "throw up my hands" and fall back to keeping that expensive monolithic convolver running all the time
but that really rubs me the wrong way from where I stand currently.
- Thread Starter
- #37
A good thread, I agree it's all about balance
www.audiosciencereview.com
Psychological &/or philosophical rejection of DSP
Help me understand. I've recently been browsing a bunch of "audiophile" forums in hope of learning something new about DSP, room treatment, noise reduction, etc. To my surprise most of what I found in the 2ch audio world was a widespread aversion to DSP, whether for EQ, room treatment or much...
www.audiosciencereview.com
olds1959special
Major Contributor
I have found that double-bass setups with EQ sound better than using an analog crossover.
- Thread Starter
- #39
Do you mean two subs? Stereo?
I plan for that, plus MBM couplers berween the deep trueSubs and my LS50s as mid+trouble.
So do you mean using separate filters on each box independently, as having no crossovers?
olds1959special
Major Contributor
I mean running the subs (with LPF only) and speakers directly from the output of the DAC instead of going to the subs and then to the speakers (through the crossover of the sub.) You have to apply a lot of subtractive EQ to the bass but I think it sounds better.