Hi,
I am looking to buil/purchase a dsp for room equalisation and digital crossover. Ideally it would also have digital amp connected to it. I am looking at available options.
Those that I am/have considered>
1) MiniDSP from minidsp.com the combo including digital amp for US$200. ( Most likely candidate)
2) Behringer Ultracurve and behringer digital crossover. ( a bit too expensive for my budget, about US$650)
3) PC-based software (for equalisation and crossover) followed by external USB DAC. (I like the idea, but clueless how to implemement this, is there a software one can purchase/download that can do this?)
And any other options that you may know of, please do let me know, Old Home thearte system etc...
Oon
I am looking to buil/purchase a dsp for room equalisation and digital crossover. Ideally it would also have digital amp connected to it. I am looking at available options.
Those that I am/have considered>
1) MiniDSP from minidsp.com the combo including digital amp for US$200. ( Most likely candidate)
2) Behringer Ultracurve and behringer digital crossover. ( a bit too expensive for my budget, about US$650)
3) PC-based software (for equalisation and crossover) followed by external USB DAC. (I like the idea, but clueless how to implemement this, is there a software one can purchase/download that can do this?)
And any other options that you may know of, please do let me know, Old Home thearte system etc...
Oon
Few days ago, I stumbled on thing you might be interested in:
DSP Crossover for PC. Frequency Allocator.
It's PC based software called "Loudspeaker Frequency Allocator".
I don't need it because I use a full range speakers so I don't know much about it.
I accidentally stumbled upon it when I was in a thought process on how I would do it for a friend of mine. I think that this software needs multi channel soundcard with ASIO support, you just create equalization and crossover curves in it and then you map it to your soundcard outputs and it works.
You could probably also use your digital exit with it.
I think it costs about 150$.
If you buy MiniDSP, you can buy their "STEREO CROSSOVER PLUG-IN" or something similar for 10$
I was thinking of make something similar with Foobar convolver. I would use "Convert stereo to 4 channels" DSP and then I would put 4 channel impulse file in convolver DSP. It would have lowpass filter on front speaker channels and highpass filter on surround speaker channels... then I would just plug big drivers to front plug and tweeters to surround plug.
Unfortunately, it can't work with foobar convolver because convolver DSP says "unsupported channel count" or something like that when you feed it 4 channel impulse file, it only works with stereo... maybe if we can contact the guy that made convolver and sweet-talk him to do 4 or 6 channel convolver, it would be sick and could possibly replace multi 1000 dollar systems if you know how to properly make your impulse files for DRC and crossovers.
If you already have or if you can make room equalization impulse file, you can convolve it with frequency sweep, then you save it twice and use highpass on one and lowpass on the other. Then you deconvolve each one of them to get two impulse files (you could make 3 if you need lowpass, bandpass and highpass).
The cheapest but most time consuming way of using those would be duplicating all your music, using one impulse file on one version and other impulse file on other version, then you can connect those to multi channel audio so you simply play them in foobar and you connect your speakers just like you would connect them if you used frequency allocator. More expensive way would be feeding those impulse files to a dedicated DSP hardware (or software) that can use convolution.
You can make your own impulse files by making frequency sweep, equalizing and modifying it however you want and then you deconvolve it with something like Voxengo Deconvolver (you must also have non modified sweep to do it).
I'm using 96kHz, 32bit float wav file with 100s sweep from 10Hz to 21000Hz for that purpose, I made quite complex equalizer with that technique, it even kills some of the resonance of my room and speakers and I'm tweaking it every few days, it sounds fantastic right now and that impulse file is only 350ms long.
DSP Crossover for PC. Frequency Allocator.
It's PC based software called "Loudspeaker Frequency Allocator".
I don't need it because I use a full range speakers so I don't know much about it.
I accidentally stumbled upon it when I was in a thought process on how I would do it for a friend of mine. I think that this software needs multi channel soundcard with ASIO support, you just create equalization and crossover curves in it and then you map it to your soundcard outputs and it works.
You could probably also use your digital exit with it.
I think it costs about 150$.
If you buy MiniDSP, you can buy their "STEREO CROSSOVER PLUG-IN" or something similar for 10$
I was thinking of make something similar with Foobar convolver. I would use "Convert stereo to 4 channels" DSP and then I would put 4 channel impulse file in convolver DSP. It would have lowpass filter on front speaker channels and highpass filter on surround speaker channels... then I would just plug big drivers to front plug and tweeters to surround plug.
Unfortunately, it can't work with foobar convolver because convolver DSP says "unsupported channel count" or something like that when you feed it 4 channel impulse file, it only works with stereo... maybe if we can contact the guy that made convolver and sweet-talk him to do 4 or 6 channel convolver, it would be sick and could possibly replace multi 1000 dollar systems if you know how to properly make your impulse files for DRC and crossovers.
If you already have or if you can make room equalization impulse file, you can convolve it with frequency sweep, then you save it twice and use highpass on one and lowpass on the other. Then you deconvolve each one of them to get two impulse files (you could make 3 if you need lowpass, bandpass and highpass).
The cheapest but most time consuming way of using those would be duplicating all your music, using one impulse file on one version and other impulse file on other version, then you can connect those to multi channel audio so you simply play them in foobar and you connect your speakers just like you would connect them if you used frequency allocator. More expensive way would be feeding those impulse files to a dedicated DSP hardware (or software) that can use convolution.
You can make your own impulse files by making frequency sweep, equalizing and modifying it however you want and then you deconvolve it with something like Voxengo Deconvolver (you must also have non modified sweep to do it).
I'm using 96kHz, 32bit float wav file with 100s sweep from 10Hz to 21000Hz for that purpose, I made quite complex equalizer with that technique, it even kills some of the resonance of my room and speakers and I'm tweaking it every few days, it sounds fantastic right now and that impulse file is only 350ms long.
Hello Oon,
Your option #2 need not cost that much. It looks like you are thinking two pieces of equipment; equalizer and crossover. The DCX 2496 is a crossover plus it has built in equalizers too. Perhaps you can download the manual and check out the functions to confirm that they meet your needs or not.
For your option #3 there is Ultimate Equalizer that uses your PC and sound card for digital XO, equalization and DAC. There is an online manual for this as well.
DT
All just for fun!
Hi oon,
I pointed to :
1. Behringer, - Not good in analog input output, excellent in digital area.
for equalizer DEQ, i prefer to use digital in and out.
Excellent room /speaker equalizer, you can measure freq response via microphone, adjust as your tastes.
Output behringer DCX you can choose 2 way or 3 way, graph x over, drive 2 way or 3 way, to amplifier
2.Another option, if you like to adjust via software, choose evaluation board from Analog devices, search google Evaluation board ADAU 1761.
I pointed to :
1. Behringer, - Not good in analog input output, excellent in digital area.
for equalizer DEQ, i prefer to use digital in and out.
Excellent room /speaker equalizer, you can measure freq response via microphone, adjust as your tastes.
Output behringer DCX you can choose 2 way or 3 way, graph x over, drive 2 way or 3 way, to amplifier
2.Another option, if you like to adjust via software, choose evaluation board from Analog devices, search google Evaluation board ADAU 1761.
Hi,
Thanks, I was considering this option, then I realised that a multichannel DAC would costs just as much as a minidsp. so there is not much benefits there.
Oon
Thanks, I was considering this option, then I realised that a multichannel DAC would costs just as much as a minidsp. so there is not much benefits there.
Oon
Hi DT,
Thanks, Let me have a closer look at it. I was looking at it didn't realise it has a built in equaliser.
Oon
Thanks, Let me have a closer look at it. I was looking at it didn't realise it has a built in equaliser.
Oon
Hi Adhiyasa38,
Thanks for the tip, the minidsp uses ADAU 1701. As for the evaluation borad, sound like a cool idea, but unfortunately I know nuts about programming such stuff... So will pass this idea for now..
Oon
Thanks for the tip, the minidsp uses ADAU 1701. As for the evaluation borad, sound like a cool idea, but unfortunately I know nuts about programming such stuff... So will pass this idea for now..
Oon
Build a state-of-the-art embedded standalone drc & xo
I've built and use a Shuttle Computer with Intel i7 and Lynx AES-16 card along with Acourate and Acourate Convolver to do xo between my subs and mains, loudspeaker correction, bass management, source selection, and volume control!
It sounds fantastic, transparent, better than any other system, either analog or digital that I have built or tried. So now I am completely dependent on a Windowz computer to keep my playback system up. Scary, huh?
That's why I'd like to see someone come up with a dependable embedded system that can have the same purity of tone and resolution as Acourate. Among the requirements: Superior coding, software dependability, long FIR wordlengths, a superior convolver, dithered volume control, and other features.
Now we have to conquer the remaining issues: Can we combine the above quality of sound and resolution with a more-dependable OS, one that can be an embedded OS, stripped down, and that can boot from ROM in a reasonable time. I imagine that would have to be a compact computer chassis like a Shuttle, housing a stripped-down Linux in a flash drive, an audio interface, and some kind of convolver software that has the quality to run separately-generated .bin files with full resolution. I've heard of BruteFIR as being Linux-compatible but don't know if it's the right kind of dependable software. And Acourate Convolver requires Windows to run, so we have to find another convolver for the Linux box. I envision a web-based controller that could even be controlled from an iphone for remote control. Has anyone built an embedded system like this? I am tired of rebooting and updating Windows and having the ability to run general-purpose software that this computer will never need to use. This form could even be commercialized.
There are professional DRC products which I have auditoned, and I won't name names, but I've found that none of them comes up to the sound quality and precision of correction of Acourate and Acourate Convolver. As an audiophile and an audio professional I need sonic transparency, and could not endorse any of the commercial DRC and loudspeaker-correction products prior to finding Acourate.
I've built and use a Shuttle Computer with Intel i7 and Lynx AES-16 card along with Acourate and Acourate Convolver to do xo between my subs and mains, loudspeaker correction, bass management, source selection, and volume control!
It sounds fantastic, transparent, better than any other system, either analog or digital that I have built or tried. So now I am completely dependent on a Windowz computer to keep my playback system up. Scary, huh?
That's why I'd like to see someone come up with a dependable embedded system that can have the same purity of tone and resolution as Acourate. Among the requirements: Superior coding, software dependability, long FIR wordlengths, a superior convolver, dithered volume control, and other features.
Now we have to conquer the remaining issues: Can we combine the above quality of sound and resolution with a more-dependable OS, one that can be an embedded OS, stripped down, and that can boot from ROM in a reasonable time. I imagine that would have to be a compact computer chassis like a Shuttle, housing a stripped-down Linux in a flash drive, an audio interface, and some kind of convolver software that has the quality to run separately-generated .bin files with full resolution. I've heard of BruteFIR as being Linux-compatible but don't know if it's the right kind of dependable software. And Acourate Convolver requires Windows to run, so we have to find another convolver for the Linux box. I envision a web-based controller that could even be controlled from an iphone for remote control. Has anyone built an embedded system like this? I am tired of rebooting and updating Windows and having the ability to run general-purpose software that this computer will never need to use. This form could even be commercialized.
There are professional DRC products which I have auditoned, and I won't name names, but I've found that none of them comes up to the sound quality and precision of correction of Acourate and Acourate Convolver. As an audiophile and an audio professional I need sonic transparency, and could not endorse any of the commercial DRC and loudspeaker-correction products prior to finding Acourate.
Hi,
You really use that stuff there when you are doing the mastering work? Must take time to get used to the idea, though I've also planned some similar but lighter system. You know the low-power integrated mini-ITX computer motherboards (guess the Shuttle has got one similar) now come with a lot more powerful CPU's. One such is the MSI J1800I which is also inexpensive one. It's not at i7 level yet but sure could do something useful. Problem is that the new boards don't have Linux support for all of the components like the graphics controller. The CPU takes only 10W power and is passive cooled.
So the hardware should be no problem, but the software might be. For DRC and bass management you need some measuring system and software which then creates the FIR filters or IIR coeffecients. Then you should port them into BruteFIR or other similar system. Anyway, there is some software for Linux, but maybe not so easy to setup and use as Acourate. For an example for DRC there is the DRC: DRC: Digital Room Correction which calculates the filters from pre-recorded room IR (I haven't studied it much). BruteFIR was mentioned, and I guess everything else can be found. Also some Windows software runs in Linux under Wine. It just takes time to integrate all things, but it's duable.
You promised not to reveal which DRC hardware you have tried, but if you haven't yet done it then I think it's worth checking out this little device:
DSPeaker-Home
Btw. the other diyaudio subforum "PC based" does have really many and long threads on this subject, see PC Based - diyAudio
If it's just Windows you're concerned about, you can achieve much the same results on your existing hardware with Linux, MPD, ALSA, Ecasound, Sox... All free software and nicely stable. I've been using this combination for months, and agree, best sound I've ever gotten from my speakers/amps.
I will have to investigate that other thread. I did look through the headings and though this would be the best thread.
For mastering, I know it sounds crazy to use a DRC or loudspeaker linearizer. And it did take a little time to get used to, but I'm very adaptable as long as something sounds superior.
I did wait ONE year testing and using Acourate before I entirely retired my analog filters and crossover to the garage and soon the garbage. I am a pioneer, and also not a Luddite. Not many mastering engineers like me. Most would prefer to stick with their PMCs and run with them. I love PMC speakers, but I have also heard PMCs corrected by Acourate and they are even more outstanding.
I have a good friend, a mastering engineer who just converted to the PMC digitally-corrected two-two-8 loudspeaker and all the comments from him and people I trust are that this digitally-corrected loudspeaker sounds terrific. So, even us mastering engineers are coming around to the idea. However, the two-two has nearly zero latency and so will have some computational compromises in impulse response compared to my brute-force Acourate-corrected Revels plus JL subwoofers.
Back to the other topics. I never tested the DSPeaker Anti-Mode. I just looked at it. Just doing low frequencies does not interest me because I like to have a high frequency rolloff target that I have selected through blood, sweat and tears. 🙂.
Back to the question of PC. It's good to know that it's "doable". Anyway, Acourate can still measure and generate the filters for me, as long as the convolver that I use can accept a standard data file. And to repeat the goal: We must turn this conceptual "PC" into an "appliance", one that will boot within a minute or less from something solid state that we might even call "firmware"--- directly to the convolver application and pass sound. Otherwise I might as well continue with my Windows Shuttle Computer, eh?
Has anyone accomplished that who you know of? That's the person I want to hear from.
Best wishes,
Bob
For mastering, I know it sounds crazy to use a DRC or loudspeaker linearizer. And it did take a little time to get used to, but I'm very adaptable as long as something sounds superior.
I did wait ONE year testing and using Acourate before I entirely retired my analog filters and crossover to the garage and soon the garbage. I am a pioneer, and also not a Luddite. Not many mastering engineers like me. Most would prefer to stick with their PMCs and run with them. I love PMC speakers, but I have also heard PMCs corrected by Acourate and they are even more outstanding.
I have a good friend, a mastering engineer who just converted to the PMC digitally-corrected two-two-8 loudspeaker and all the comments from him and people I trust are that this digitally-corrected loudspeaker sounds terrific. So, even us mastering engineers are coming around to the idea. However, the two-two has nearly zero latency and so will have some computational compromises in impulse response compared to my brute-force Acourate-corrected Revels plus JL subwoofers.
Back to the other topics. I never tested the DSPeaker Anti-Mode. I just looked at it. Just doing low frequencies does not interest me because I like to have a high frequency rolloff target that I have selected through blood, sweat and tears. 🙂.
Back to the question of PC. It's good to know that it's "doable". Anyway, Acourate can still measure and generate the filters for me, as long as the convolver that I use can accept a standard data file. And to repeat the goal: We must turn this conceptual "PC" into an "appliance", one that will boot within a minute or less from something solid state that we might even call "firmware"--- directly to the convolver application and pass sound. Otherwise I might as well continue with my Windows Shuttle Computer, eh?
Has anyone accomplished that who you know of? That's the person I want to hear from.
Best wishes,
Bob
Hi,
There's a thread on going from Acourate setup to brutefir on Linux, or actually using the the same way you are planning to do:
http://www.diyaudio.com/forums/pc-based/193146-linux-acourate-perfect-pair.html
Acourate has some instructions related to BruteFIR setup:
http://www.acourate.com/ConvolutionEngine.doc
How about using a pair of Sony DRE-S777 reverbs? Don't know can you put your own IR's onto them or are you restricted to the one provided by Sony.
Some DSP evaluation boards could be used but needs some knowledge. There is also the Nadja DSP crossover, another thread on this Digital Line Level subforum:
http://www.diyaudio.com/forums/digital-line-level/215379-dsp-xover-project-part-2-a.html
Don't know what kind of plans for FIR they have, but for me it looks not promising.
There's another convolver for Linux: jconvolver
On this thread there are some instruction and link to download:
[ubuntu_studio] How do you make jconv/jconvolver work
Lynx AES16 doesn't have ALSA drivers though, only OSS but that shouldn't be a problem. Finding proper Linux distro which is light enough might be.
There's a thread on going from Acourate setup to brutefir on Linux, or actually using the the same way you are planning to do:
http://www.diyaudio.com/forums/pc-based/193146-linux-acourate-perfect-pair.html
Acourate has some instructions related to BruteFIR setup:
http://www.acourate.com/ConvolutionEngine.doc
How about using a pair of Sony DRE-S777 reverbs? Don't know can you put your own IR's onto them or are you restricted to the one provided by Sony.
Some DSP evaluation boards could be used but needs some knowledge. There is also the Nadja DSP crossover, another thread on this Digital Line Level subforum:
http://www.diyaudio.com/forums/digital-line-level/215379-dsp-xover-project-part-2-a.html
Don't know what kind of plans for FIR they have, but for me it looks not promising.
There's another convolver for Linux: jconvolver
On this thread there are some instruction and link to download:
[ubuntu_studio] How do you make jconv/jconvolver work
Lynx AES16 doesn't have ALSA drivers though, only OSS but that shouldn't be a problem. Finding proper Linux distro which is light enough might be.
How about MiniDSP's AcourateDRC:
AcourateDRC | MiniDSP
OpenDRC series | MiniDSP
One box can do two FIR's, so there are only two outputs, so using multiple is a must.
AcourateDRC | MiniDSP
OpenDRC series | MiniDSP
One box can do two FIR's, so there are only two outputs, so using multiple is a must.
Dear mhelin:
You're clever! You wrote:
That's thinking outside the box! Sony did have a kit for that, but I bet they used some proprietary IR format. I need to find out about that. David Chesky offered me two of them at $1000 each, and I tested them and felt I had better reverbs here. Convolution reverb has its limitations compared to algorithmic reverb. However, it should be able to store a humongous sized IR! I wonder if he still has them for sale.
That would be fun to work on. Do you know anything about the convolution engine that Sony used?
You're clever! You wrote:
That's thinking outside the box! Sony did have a kit for that, but I bet they used some proprietary IR format. I need to find out about that. David Chesky offered me two of them at $1000 each, and I tested them and felt I had better reverbs here. Convolution reverb has its limitations compared to algorithmic reverb. However, it should be able to store a humongous sized IR! I wonder if he still has them for sale.
That would be fun to work on. Do you know anything about the convolution engine that Sony used?
That's what I'm getting as a backup to my Acourate-run computer. I'm putting together two Open DRC boxes, one for highpass stereo and one for lowpass. That will give me 6144 taps per channel for a total of 4 channels. We'll see how she sounds. Uli tells me I'll have to do minimum phase calculations as it downsamples to 48k and I didn't fully understand the implications, but I'll get it sorted out. This will give me the backup to the computer that I need and have a little fun with another FIR filter.
Apparently the Sony reverb got its power through a massive array of 32-bit float Sony DSP chips. If David still has the verbs and if he has the sampling CD rom that was an optional extra I'll buy two of them for $1000 each and have some more fun!
Haven't studied what DSP's the Sony uses, I also think it's their own DSP (at least manufactured by them, the IP might be 3rd party). I thought there must have been other similar devices but google couldn't find any. There days few use those, first they got Protools HD and now as computers are not limit I guess almost everything is native DSP plugins. Now because of that those convolution engines and external or internal DSP cards can be get for much less that they used to cost.
So another way to solve your problem could be to use some of those DSP cards, but they also need a PC or something with PCI(e) bus to stick the hardware. Most of these though use fixed point arithmetics (Motorola 56k DSP's), new Avid HDX though uses TI floating point DSP's which is closer to what is used on PC native plugins (and Acourate I think). MiniShark (MiniDSP OpenDRC) uses single Analog Devices Shark DSP which is a floating-point DSP. Those DSP cards (excluding MiniShark with I2S interfaces) don't usually come with converters or even SPDIF or other digital interfaces so in the end they are not that useful.
Found a thread on outboard convolution gear on Gearsluz :
Outboard convolution reverb? - Gearslutz.com (not much more information, someone claimed that it's FIR is implemented in time domain, so it must have really fast multipliers).
$1000 for two is really good price, go for it. OpenDRC doesn't either have enough RAM or does the FIR really brute force way in time domain (Linux BF uses FFT) beacuse it's IR length seems to be kind of limited.
So another way to solve your problem could be to use some of those DSP cards, but they also need a PC or something with PCI(e) bus to stick the hardware. Most of these though use fixed point arithmetics (Motorola 56k DSP's), new Avid HDX though uses TI floating point DSP's which is closer to what is used on PC native plugins (and Acourate I think). MiniShark (MiniDSP OpenDRC) uses single Analog Devices Shark DSP which is a floating-point DSP. Those DSP cards (excluding MiniShark with I2S interfaces) don't usually come with converters or even SPDIF or other digital interfaces so in the end they are not that useful.
Found a thread on outboard convolution gear on Gearsluz :
Outboard convolution reverb? - Gearslutz.com (not much more information, someone claimed that it's FIR is implemented in time domain, so it must have really fast multipliers).
$1000 for two is really good price, go for it. OpenDRC doesn't either have enough RAM or does the FIR really brute force way in time domain (Linux BF uses FFT) beacuse it's IR length seems to be kind of limited.
Having tested several DRC products, +1 to Bob's comment.
Wrote a step by step article here: Computer Audiophile - Advanced Acourate Digital XO Time Alignment Driver Linearization Walkthrough
Several years have now passed. So to me it is a long long time 🙄 ago when I have stopped my developments with Linux and Brutefir. You should know that I have published a dedicated version 'Brutefir on a USB stick'. It was built as a minimum realtime Linux kernel together with Brutefir.
The old package is still available by download, see
www.acourate.com/freedownload/BruteFIR_Delta2496_2.6.26.8-rt12.pdf, goto page4 for English text
www.acourate.com/freedownload/BruteFIR_Delta2496_2.6.26.8-rt12.zip
The basic idea: have a small USB stick (the development has taken place when the stick sizes have been in the range of 128 MB 😀), boot from it and the PC does the job. No unnecessary services, process and so on.
Indeed I still have a playback system with a RME HDSP9652 soundcard running this way [including filter switch / samplerate]
At the end I have stopped the Linux way for several reasons:
- my development computer (Gentoo) died
- bad support of new soundcards by ALSA
- bad support of Linux by soundcard manufacturers
- difficulties to switch between samplerates. It is not supported by Brutefir and ALSA does not deliver the required information for most soundcards
- more programming flexibility with Windows (I' not a C programmer)
That's why I have decided to develop AcourateConvolver as a Windows program. And yes, I do not want to go back to Linux.
I hope my background information is useful.
Cheers
Uli
The old package is still available by download, see
www.acourate.com/freedownload/BruteFIR_Delta2496_2.6.26.8-rt12.pdf, goto page4 for English text
www.acourate.com/freedownload/BruteFIR_Delta2496_2.6.26.8-rt12.zip
The basic idea: have a small USB stick (the development has taken place when the stick sizes have been in the range of 128 MB 😀), boot from it and the PC does the job. No unnecessary services, process and so on.
Indeed I still have a playback system with a RME HDSP9652 soundcard running this way [including filter switch / samplerate]
At the end I have stopped the Linux way for several reasons:
- my development computer (Gentoo) died
- bad support of new soundcards by ALSA
- bad support of Linux by soundcard manufacturers
- difficulties to switch between samplerates. It is not supported by Brutefir and ALSA does not deliver the required information for most soundcards
- more programming flexibility with Windows (I' not a C programmer)
That's why I have decided to develop AcourateConvolver as a Windows program. And yes, I do not want to go back to Linux.
I hope my background information is useful.
Cheers
Uli
Last edited:
Dear Uli and team: This is a pity. OS on a stick is the closest thing to an embedded system we could get right now. You get the computer power of Intel, the power of Acourate, your own filters. It is a genuine shame that soundcard manufacturers have not supported Linux.
Indeed, as Windows appears increasingly to be a dead end, superseded by Linux, Android and OSX.
- Status
- Not open for further replies.