I think with miniDSP you go:
44.1->DAC->ADC->SRC(48)->processing->DAC
Or you can send it the SPDIF (if you have a miniDIGI) and cut out the first DAC->ADC pair, and it still does SRC. I don't klnow if the onboard SRC will be better or worse than a software sRC at source.
The GroundSound units will process at 96kHz, but I don't think they have an SPDIF feed option.
Expect the next rev of the iMacs to have Light Peak/Thunderbolt (like the new MacBook Pros), which does support PCIe,
http://www.diyaudio.com/forums/digi...bolt-end-usb-firewire-hdmi-more-new-post.html
dave
Member
Joined 2009
Thanks for clearing that up, I wasn't sure about the digital input.
OK so miniDSP offers a better precision than kx-project because it processes in 24bits but they both run at 48kHz. Does miniDSP upsample the signal on the output?
Increasing 16 to 24 bits isn't an upsampling operation. SPDIF data streams have 24 bits of precision, but consumer devices only fill in the least significant 16 bits. So its just a multiplication opportunity in the crossover, AFAIK anyway.
Note that you need a miniDIGI to get SPDIF input to a miniDSP, and that means you can't use the handy off-the-shelf enclosure. The miniDSP/miniDIGI combo is $160. I'm wondering if its worth the hassle of building a custom enclosure and power supply and using these, or whether it makes more sense to just put a fanless Atom PC behind each speaker and use my choice of PC sound card.
Seemed like overkill when it first popped into mind, but once you factor in pain of building enclosures etc its tempting to consider GroundSound, and then its tempting to consider that you can pretty much build a PC and MAudio solution for the same money, give or take. Madness!
What do you guys think is the best (most powerful yet still cool) CPU for a mini-ITX system to do crossover duties? You have a few options like a dual core Atom, some of the lower power i5 chips or Phenom x4, Athlon II chips from AMD. A powerful 'mobile' processor seems like it would be good but I don't see they are easy to get hold of and don't fit most ITX motherboards.
Low power chips obviously mean the PSU would do less work and is easier to cool without a fan.
Low power chips obviously mean the PSU would do less work and is easier to cool without a fan.
It depends on your chosen software's requirements.
I have Frequency Allocator running on an old 3GHz Pentium D with 1 GB of ram (three way XO, although I don't think it affects CPU usage much). Even when running multiple other applications simultaneously, I haven't had a playback glitch. Building a mini-ITX box for Frequency Allocator, I'd go with the lowest powered processor I could find.
If you want to use another program, take a look at the user support forum. There are likely to be a number of "how much CPU power do I need for a standalone" threads.
I have Frequency Allocator running on an old 3GHz Pentium D with 1 GB of ram (three way XO, although I don't think it affects CPU usage much). Even when running multiple other applications simultaneously, I haven't had a playback glitch. Building a mini-ITX box for Frequency Allocator, I'd go with the lowest powered processor I could find.
If you want to use another program, take a look at the user support forum. There are likely to be a number of "how much CPU power do I need for a standalone" threads.
No idea - I'm fairly confident in what I need, because I'm planning to use IIR and not try to push the envelope with a lot of FFT code.
Seems you can get fanless dual core Pine Trail at 1.8GHz for 60 quid here: mini-itx.com - store - Intel Atom Mini-ITX boards but you'll need choose a case and PSU solution carefully to run a PCI card - the boards with integrated power probably code out cheaper (eg with a M300, albeit useless CF - expect to boot off a USB key). Absolutely no idea how well BruteFIR would run on that, although we've had 10 years of CPU evolution since Torsten started on this - and gcc got a good deal better at optimising too.
But don't listen to me - my current ITX is a server and definitely not silent! I'm attracted to the M350 case for convection cooling but it does imply a USB or firewire sound card. If you want silent/passive cooling, you might want to consider a PSU-less microtower case so there is air over the motherboard rather than something small with a PCI card laying right over the CPU heatsink.
You haven't said how much software you aim to do - sure the path of least resistence isn't to keep that LynxAES and just run Windows off an SSD with Frequency Allocator Light? Won't do all the magic phase stuff, but Thuneau can probably advise what sort of CPU you'd need for teh full-fat version.
Member
Joined 2009
Not to scare anyone but..
DIYaudio, meet LULL.
Lull is my entertainment center PC. It's totally silent, no optical drive. CDs are ripped to .WAV or .FLAC on my desktop in the other room. The heatsink gets hot but never to a temperature that causes a problem and my computer even sits next to house's heat vent. The only problem that I ran into with my setup was the unvailability of a PCI slot so I have to use a PCIe to PCI converter board.
Despite it being a year old I'm still in the process of building a proper case. My primary goals were total silence, HD video playback and small size. With slightly different requirements cost and other factors can be greatly optimized. There might be better components available on the market now, if I was to build another one I would search into low power laptop versions of the iCore chips.
It's a miniITX 1156 Zotac motherboard. I use an i3 CPU with the integrated HD video. Fanless heatsink, SSD and fanless laptop brick style power supply.
motherboard - Zotac H55ITX-A-E LGA1156
CPU and graphics - Intel i3 530
CPU cooler- Noctua NH-C12P
hard drive - Kingston SSD 64GB
Power Supply- picoPSU-150-XT/12.5A 150W AC-DC
sound - Creative Audigy 2 ZS with kx-project (with a PCI Express to PCI Adapter Card)
DIYaudio, meet LULL.
Lull is my entertainment center PC. It's totally silent, no optical drive. CDs are ripped to .WAV or .FLAC on my desktop in the other room. The heatsink gets hot but never to a temperature that causes a problem and my computer even sits next to house's heat vent. The only problem that I ran into with my setup was the unvailability of a PCI slot so I have to use a PCIe to PCI converter board.
Despite it being a year old I'm still in the process of building a proper case. My primary goals were total silence, HD video playback and small size. With slightly different requirements cost and other factors can be greatly optimized. There might be better components available on the market now, if I was to build another one I would search into low power laptop versions of the iCore chips.
It's a miniITX 1156 Zotac motherboard. I use an i3 CPU with the integrated HD video. Fanless heatsink, SSD and fanless laptop brick style power supply.
motherboard - Zotac H55ITX-A-E LGA1156
CPU and graphics - Intel i3 530
CPU cooler- Noctua NH-C12P
hard drive - Kingston SSD 64GB
Power Supply- picoPSU-150-XT/12.5A 150W AC-DC
sound - Creative Audigy 2 ZS with kx-project (with a PCI Express to PCI Adapter Card)
Attachments
Last edited:
Check Flowstone. This is a graphical audio compiler dedicated to digital audio.
FlowStone Graphical Programming Software
Implementing a digital audio crossover only takes a few minutes with Flowstone.
XOVER Linkwitz-Riley
XOVER Lipshitz-Vanderkooy
XOVER State Variable Filter
Actually, you can program your own digital audio analyzer. This way you can plot the magnitude and phase of any signal.
FFT-based Audio Analyzer
Now that you are in total control, you can refine your digital audio crossover at will, adding more IIR filters, adding FIR filters, adding a Linkwitz Transform (aka bass boost), etc.
Experimenting with IIR filters
Experimenting with FIRs
You can build a whole digital audio environment, automatically and interactively computing the FIR filter coefficients that you need, for implementing in one single pass, both the speaker linearization function, and the highpass and/or lowpass function you need within a crossover.
Speaker Lab with FIR filter
Speaker Lab is indeed a quite complicated audio DSP arrangement operating in real time, taking the targeted acoustic Bode plot (gain + phase) as input data, measuring the bare speaker driver acoustic Bode Plot, comparing the targeted Bode plot with the bare speaker driver Bode plot (divide magnitudes, subtract the phases) for computing the required electric correction Bode plot, and finally computing the required electric impulse response correction (iFFT) which actually is the FIR filter that you need to apply.
FlowStone Graphical Programming Software
Implementing a digital audio crossover only takes a few minutes with Flowstone.
XOVER Linkwitz-Riley
XOVER Lipshitz-Vanderkooy
XOVER State Variable Filter
Actually, you can program your own digital audio analyzer. This way you can plot the magnitude and phase of any signal.
FFT-based Audio Analyzer
Now that you are in total control, you can refine your digital audio crossover at will, adding more IIR filters, adding FIR filters, adding a Linkwitz Transform (aka bass boost), etc.
Experimenting with IIR filters
Experimenting with FIRs
You can build a whole digital audio environment, automatically and interactively computing the FIR filter coefficients that you need, for implementing in one single pass, both the speaker linearization function, and the highpass and/or lowpass function you need within a crossover.
Speaker Lab with FIR filter
Speaker Lab is indeed a quite complicated audio DSP arrangement operating in real time, taking the targeted acoustic Bode plot (gain + phase) as input data, measuring the bare speaker driver acoustic Bode Plot, comparing the targeted Bode plot with the bare speaker driver Bode plot (divide magnitudes, subtract the phases) for computing the required electric correction Bode plot, and finally computing the required electric impulse response correction (iFFT) which actually is the FIR filter that you need to apply.
Another option, from the linix camp.
MPD with alsa and sox. All software is free, and it's all running very nicely on my old single core AMD processor. See: Digital Crossover/EQ with Open-Source Software: HOWTO | Richard's Stuff
Many sound cards supported, both for playing digital source and for analog input (or digital input)... Pictured at the above site is a silent server.
MPD with alsa and sox. All software is free, and it's all running very nicely on my old single core AMD processor. See: Digital Crossover/EQ with Open-Source Software: HOWTO | Richard's Stuff
Many sound cards supported, both for playing digital source and for analog input (or digital input)... Pictured at the above site is a silent server.
Wow, that's really fantastic. I'll try this. Any chance using an ARM-powered Android Tablet instead of a X86-powered Ubunbtu Server PC?
Not sure what you mean, Jay. Richard Taylor approach is CPU-agnostic Linux Ubuntu server edition. Richard Taylor specifies the Shuttle XS35v2 and the Zotac zbox hardware as first choices, both Intel X86 (Atom) powered. Richard Taylor also specifies Raspberry Pi (ARM powered), however reading further there are three difficulties associated with Raspberry Pi.
1st difficulty is to use Richard Taylor latest rt-plugins, dealing with the fact that ARM CPUs are not equipped with x86 SSE extensions.
2nd difficulty is to edit the Makefile and remove the “-march=native” from CFLAGS, which for some reason breaks things on the Raspberry Pi.
3rd difficulty is some Raspberry Pi USB bottleneck, preventing grabbing stereo audio from USB (SPDIF stereo-in), and at the same time sending multichannel audio to USB (6 analog channels-out for a stereo 3-way crossover). Thus, with Raspberry Pi, the stereo audio must reside on the SSD, or must come from WiFi/Ethernet.
Knowing this, basing on x86 or ARM, one must take into account how moody ALSA (Advanced Linux Sound Architecture) can be when dealing with multichannel audio. You faced this, Jay, with ALSA complaining about an "out of range parameter" upon sending 4 channels of audio into a 5.1 USB audio attachment.
One shall encourage the diyAudio communauty in following the Richard Taylor way.
The diyAudio communauty will grab valuable info about x86 and ARM hardware behavior, Linux kernel behavior, ALSA behavior and multichannel audio attachments driver behavior.
As soon as ALSA stability and reliability get decent when dealing with multichannel audio, people will get motivated in writing CPU-agnostic Linux software acting as graphical front-end, generating the needed system calls.
Richard Taylor approach was to base on a Linux distribution like Linux Ubuntu server edition. There should be a diyAudio discussion, entitled "Richard Taylor Ubuntu DSP XO". I think you are best placed, Jay, for opening that diyAudio discussion.
I guess people will try basing on Android instead of Ubuntu. This way, provided Android 4.4 KitKat enables multichannel USB audio attachments in full duplex, people will get motivated for publishing Android software, converting Android Tablets into DSP XOs taking SPDIF as input. Soon or later, there will exist a diyAudio discussion entitled "Richard Taylor DSP XO - migrating to Android". Is there material for opening such discussion right now?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.