Off the shelf versus PC for high-end DAC + Streamer +DSP

I don't know much about HiFi, but a normal PC probably if you fine tune the spread spectrum will reduce electrical noise, and maybe turn off the boost of the cpu at the cost of performance, because usually when components are clocked higher or with high usage tend to introduce more electrical noise...


About sound, I think you can't go wrong with latest Creative Soundcards, they improved by miles and are now good, even the lower end AE-5 when tested had a better DAC than receivers...


This is their high end AE-9

hero.png



ESS SABRE-class 9038
DAC: 129 dB (I don't think it does 129, but it should probably still be high)
PCM: 32-bit / 384 kHz
DSD: DSD64
THD + N 0.0001% (DAC)
Swappable OP AMPS
And a good Bi-amp headphone which is enough to drive any headphones up to 600 ohms


What they fall really bad, at least the low end AE5 is recording, don't know about their high end AE9 but will probably follow the same trend off not being good for recording...




ps: They have the advantage against any DAC specially for gaming on a PC because their sound processor removes a lot of CPU usage leaving more juice for gaming...
 
Last edited:
Member
Joined 2008
Paid Member
The Motu M4 is a fabulous product. I just set it up without fuss into my stereo OB 2-way DSP crossover system (Pulseaudio crossover rack running on Linux Mint). Effortless clear sound. Highly recommended. Thanks again, Charlie, for recommending it.
 
Hi von Ah,

I have looked at Motu M4 for a completely different purpose, but the specifications mention only Windows and OSX (or whatever it is called nowadays) as supported operating systems. However, according to your post it also supports Linux. Could you please clarify?

Kindest regards,

M
 
Hi von Ah,

I have looked at Motu M4 for a completely different purpose, but the specifications mention only Windows and OSX (or whatever it is called nowadays) as supported operating systems. However, according to your post it also supports Linux. Could you please clarify?

Kindest regards,

M

You will need a recent version of the linux kernel to use the M4. The operating system should run kernel 5.4 for output only, or kernel 5.8 or higher in order to use the M4 in duplex mode (simultaneous input and output). Kernel 5.8 is widely available now as a mainline kernel and if your OS comes with an earlier kernel you can probably upgrade it to 5.8 or newer, or just wait until you get that with the next release of the OS. For example, the current LTS release of Ubuntu comes with kernel 5.4 but the next LTS release coming in April will have 5.10 or 5.11.
 
Hi von Ash, CharlieLaub, phofman,

thank you for the replies. It is a good news since, as I mentioned, no Linux support is mentioned anywhere in the specifications.

It will be a learning curve with Linux as my machines run various flavors of UNIX (OmniOS, BSD). Is there a particular distribution that you would recommend?

Kindest regards,

M
 
Member
Joined 2008
Paid Member
Mint is easy to use, has lots of support, and can be used on older/lower power machines. It’s based off of Ubuntu, I think.

I’ve really liked Solus for general computer use (surfing, gaming, etc.) but it’s relatively off on its own, not closely related to the more common distros, so I wouldn’t use it for more common audio processes. It can be very fast and reliable, and its desktop environment is attractive and smooth.

I started using Pop!OS on my laptop because it has dedicated drivers for hybrid NVIDIA graphics. It works very well on my new laptop.

My experience and expertise is limited. Others in this thread know loads more about Linux than I do.
 
Last edited:
A couple commercial multi-channel dac options:
ESI - GIGAPORT eX
Okto Research

This could be incorporated with a Raspberry Pi:
HiFiBerry DAC+ DSP | HiFiBerry

Thin client PCs can be silent and cheap. I got a HP t610 used for $40 and added a SATA SSD plus more RAM. Windows 10 or many different Linux distros can be installed.

Did not know there is a new version of the ESI Gigaport. I own the old one, and have only good things to say. Beats the heck out of the earlier version of Focusrite Scarlett 18i20, which I returned immidiately.

ESI Gigaport is not to be considered high-end, but it can for sure produce great sound. And you have all opportunities in this world to experiement with filters.
Easy hack to improve sound (not sure if that is the same with usb C though), but most importantly to iron out disturbances from USB power, is to simply disconnect the power rails via USB power splitter, like this one: USB-B adapter cable for external USB-A power supply - Audiophonics

Then use a good linear power supply, or in fact an original iphone or ipad charger (look up measurement).

Most devices, even those with inbuilt powersupply, use the USB-power in some ways ie. for correct sleep and wake mode.

Above hack is a hidden gem, that IMO all should do.
 
Many PCs today have huge internal latencies, that might result in crackle in sound. It does not matter how fast the PC is. The single most important factor to reduce issues with crackle and stutter is in my experience to disable wifi and use LAN cable.

For USB 2.0 devices, any intel Nuc is IMO way worse than using an old Thinkpad T530, that have low internal latency and can be run fanless for audio without any issues. New is not better in regards to USB 2.0 devices. That is my opinion.
 
I think I'm spoilt for choice for off the shelf DACs, there's lots which measure very well, Topping, chord, Okto, Oppo, mini DSP SHD (interesting option with DIRAC too) etc etc.

Almost all of these are exclusively stereo. You cannot just hookup multiple stereo USB DAC's and end up with a multichannel setup. The individual devices will drift apart over time. You'll either need a world clock feature to sync them or connect them through SPDIF on something that can output multichannel SPDIF (or one of the PRO level equivalents)

1 other thing that I forgot to mention, how does using Jriver/Roon with convolution filter DSP functions compare to a dedicated DSP?

A PC will just be more powerful it can do hundreds to thousands of IIR eq's and, and a dazzling number of FIR taps. A dedicated DSP is always limited. The question is: does it really matter for your usecase?

I think my existing main computer running Roon Core, HQplayer (server) with DSD encoding and a DSP program (most likely Acourate for conv filters) will work perfectly, you need a bit of grunt to do the only the fly PCM to DSD encoding, and my main computer has plenty of that. I also really don't want to have to put a silent version of this next to my hifi if I can avoid it.

Why on earth bother with DSD? Just leave everything in the PCM domain. You might also be able to skip the HQPlayer part. Roon can do convolution all by itself as well as regular IIR filers.

I'll then connect this PC over ethernet to a small headless NUC (client PC) running the HQplayer client which is outputting over USB to a multichannel DAC.

It's quite a bit of software your planning to buy.. Roon, HQplayer, Acourate

I just need to figure out if I can build things this way. I'm not 100% sure that the low processing power NUC outputting over USB works, I think it does, but need a bit more research.

I don't see how that would be an issue. We've done multichannel audio on PC's since the late 90's.. A low powered NUC is much more powerful than anything we had back then. There will even be some room for processing.


Yes or Hypex.
Why even bother with FIR-filters in the first place? This one requires to fully understand the difference between FIR and IRR, and to think about that almost none of the best active speakers in this world uses FIR, but only IRR filters.

I think at least the D&D 8C uses FIR filters on high latency mode. The gain to be had with FIR filters is room correction. The biggest issue I think is that most people don't actually know how to create the correct filters, and also not how to do proper measurements and know how to interpret them. THAT is actually what you should focus on. THAT is what makes all the difference in the end.

I would suggest buying a Minidsp SHD and measurement microphone. It's excellent quality and comes with build-in room correction that does not need that much expertise to get right. It will also do away with about two-thirds of your software costs, which covers about half of the asking price already.
 
Many PCs today have huge internal latencies, that might result in crackle in sound. It does not matter how fast the PC is. The single most important factor to reduce issues with crackle and stutter is in my experience to disable wifi and use LAN cable.

For USB 2.0 devices, any intel Nuc is IMO way worse than using an old Thinkpad T530, that have low internal latency and can be run fanless for audio without any issues. New is not better in regards to USB 2.0 devices. That is my opinion.

don't disagree with you if you're using the NUC only, but I'm not planning on doing that, I'll be handing off all the processing to a high powered machine which will be far better at processing than a thinkpad.

the connection between the PC and NUC is all gigabit ethernet so no issue there either

Almost all of these are exclusively stereo. You cannot just hookup multiple stereo USB DAC's and end up with a multichannel setup. The individual devices will drift apart over time. You'll either need a world clock feature to sync them or connect them through SPDIF on something that can output multichannel SPDIF (or one of the PRO level equivalents)

Think you have to pick a multichannel DAC rather than connecting multiple stereo DACs, there are plenty well regarded ones on the market to pick from.

A PC will just be more powerful it can do hundreds to thousands of IIR eq's and, and a dazzling number of FIR taps. A dedicated DSP is always limited. The question is: does it really matter for your usecase?

It's a fair point that it might not matter for my use case, however it will give the best quality and if I'm going to use a PC anyway I think it's a sensible idea to combine the functionality, it also comes out as much more cost effective, so I have more money to spend on better drivers and amps....

Why on earth bother with DSD? Just leave everything in the PCM domain. You might also be able to skip the HQPlayer part. Roon can do convolution all by itself as well as regular IIR filers.

From what I've read DSD is pretty much standard practice for high end audio, by converting & upsampling PCM to the maximum native rate that the DAC can handle before the audio hits the DAC you reduce the processing required and therefore any noise production on the DAC.

without this DSD conversation most DACs will be doing the upsampling internally anyway with an audible impact.

https://i.imgur.com/YXoACeG.png

Generally HQPlayer is seen as having the best implementation here, Roon can do it yes but HQPlayer is better, Roon even has the ability to use HQPlayer because of this.

I don't see how that would be an issue. We've done multichannel audio on PC's since the late 90's.. A low powered NUC is much more powerful than anything we had back then. There will even be some room for processing.

I think at least the D&D 8C uses FIR filters on high latency mode. The gain to be had with FIR filters is room correction. The biggest issue I think is that most people don't actually know how to create the correct filters, and also not how to do proper measurements and know how to interpret them. THAT is actually what you should focus on. THAT is what makes all the difference in the end.

I would suggest buying a Minidsp SHD and measurement microphone. It's excellent quality and comes with build-in room correction that does not need that much expertise to get right. It will also do away with about two-thirds of your software costs, which covers about half of the asking price already.

The goal for my build is to maximise quality at every possible stage, Mini SHD will undoubtedly deliver great results, however I think the perceived wisdom is that other solutions can deliver higher quality.

As always it's a trade off:- are the differences significant enough to warrant the additional complexity? what features are important? ease of use versus learning curve etc.

ultimately it boils down to a cost versus value decision, I'm not yet sure what the answer is but the more I research the more I lean to a PC based solution.
 
Think you have to pick a multichannel DAC rather than connecting multiple stereo DACs, there are plenty well regarded ones on the market to pick from.

Yes, but not in your list ;) Except for the two in your list, you'll end up with some of the pro audio interfaces (of which already several good ones have been mentioned).

From what I've read DSD is pretty much standard practice for high end audio, by converting & upsampling PCM to the maximum native rate that the DAC can handle before the audio hits the DAC you reduce the processing required and therefore any noise production on the DAC.

Whoever said that the DAC performs best at peak rates? I think it's pure nonsense. Look at any ASR reviews of a decent DAC. It gives you 120 SINAD with just a puny 44.1Khz source across the spectrum up until 20 kHz. It really doesn't get any better than that.

Let me ask another question: where are you going to do volume control? If digitally via the DAC, the whole DSD story is even more stupid. Volume control means that DSD gets converted to PCM, the volume control is applied, and then send for DA conversion (as in your picture).


I highly doubt that the DSD direct feature will give better performance. The multi-bit modulator inside the DAC is specifically designed to work best with the silicon. Bypassing that means that you might actually suboptimally use the DAC. Also, the analogue filters on the DAC's are not really designed to handle DSD content, since there is little filtering done in the DAC when using the DSD direct route. There will be more high-frequency noise left.

I can also explain why people that try this out think the DSD version sounds better (at least with AKM DAC's): the DSD audio is a bit louder than the PCM version. If not doing a proper level matched test, you are bound to favour the louder version.

The goal for my build is to maximise quality at every possible stage, Mini SHD will undoubtedly deliver great results, however I think the perceived wisdom is that other solutions can deliver higher quality.

"Perceived" is the right word indeed. I would say, first try a controlled, level matched double blind test with some of those upsampling tools. If you can actually hear the difference you can worry about those things (assuming that none of these things changes frequency response or phase).
 
Yes, but not in your list ;) Except for the two in your list, you'll end up with some of the pro audio interfaces (of which already several good ones have been mentioned).

Whoever said that the DAC performs best at peak rates? I think it's pure nonsense. Look at any ASR reviews of a decent DAC. It gives you 120 SINAD with just a puny 44.1Khz source across the spectrum up until 20 kHz. It really doesn't get any better than that.

Let me ask another question: where are you going to do volume control? If digitally via the DAC, the whole DSD story is even more stupid. Volume control means that DSD gets converted to PCM, the volume control is applied, and then send for DA conversion (as in your picture).

I highly doubt that the DSD direct feature will give better performance. The multi-bit modulator inside the DAC is specifically designed to work best with the silicon. Bypassing that means that you might actually suboptimally use the DAC. Also, the analogue filters on the DAC's are not really designed to handle DSD content, since there is little filtering done in the DAC when using the DSD direct route. There will be more high-frequency noise left.

I can also explain why people that try this out think the DSD version sounds better (at least with AKM DAC's): the DSD audio is a bit louder than the PCM version. If not doing a proper level matched test, you are bound to favour the louder version.

"Perceived" is the right word indeed. I would say, first try a controlled, level matched double blind test with some of those upsampling tools. If you can actually hear the difference you can worry about those things (assuming that none of these things changes frequency response or phase).

Clearly I need to do more research about DSD versus PCM, and about DAC implementation. I know a bit but I'm far from knowledgeable enough to make the right/best decision yet.

I've not yet decided on volume control, I think I may have mentioned in another thread that I've tried a DAC straight into a power amp with volume control on the DAC before and didn't like it. That doesn't mean it's not a good answer with a different model, but something I need to think through and I suspect try versus a pre-amp.

I guess having the option to try both PCM and DSD might be sensible, but I need to read up more to see if/how it alters my choice in DAC and/or the use of HQPlayer/DSP etc.
 
I've not yet decided on volume control, I think I may have mentioned in another thread that I've tried a DAC straight into a power amp with volume control on the DAC before and didn't like it. That doesn't mean it's not a good answer with a different model, but something I need to think through and I suspect try versus a pre-amp.

It depends a bit on your speakers. With very sensitive speakers you might be better off with an analogue volume control. A digital one will raise the noise floor a bit when you're listening at a lower volume. Just know that a multichannel analogue volume control will add some additional cost, that can be significant depending on the implementation.
 
Many PCs today have huge internal latencies, that might result in crackle in sound. It does not matter how fast the PC is.

What exactly do you mean by "huge"?

I don't find latency to be a problem AT ALL. It's under 50msec for sure. Unless you are doing something wrong? It's more about how large your buffering is (number of samples and sample rate).
 
Yes or Hypex.

Why even bother with FIR-filters in the first place? This one requires to fully understand the difference between FIR and IRR, and to think about that almost none of the best active speakers in this world uses FIR, but only IRR filters.

If you can't answer that question, it's clear that YOU don't understand the difference between IIR and FIR filters, what they can do and can't do, etc.

Using your logic, one could post the question that "none of the best speakers in the world use DSP, so why use it in the first place?".
 
It depends a bit on your speakers. With very sensitive speakers you might be better off with an analogue volume control. A digital one will raise the noise floor a bit when you're listening at a lower volume. Just know that a multichannel analogue volume control will add some additional cost, that can be significant depending on the implementation.

Not yet sure what route I'm going down for the mains, in fact I keep thinking that I'll end up build a few different types longer term.

I tend to listen at a decent volume so doubt the noise floor will be an issue (though as a rule I'd rather not increase the noise floor unless it's for a good trade off in quality).

Thanks for the pointer on DACs, I've been doing a lot more reading, seems PCM might be a better answer for me. I've also been looking into SD versus R2R and I think I've concluded that it doesn't matter provided the DAC measures well and sounds good - think I'm pretty much on pick based on features and preference (not sure if I have a preference though! :) ).