A novel way to massively improve the SQ of computer audio streaming

[guiltyboxswapper]

2 hours ago, Exocer said:

 

I've tried enabling and streaming to HQPlayer Embedded after all of this HQPlayer talk... and so far I feel MPD or BubbleUPNP sounds more transparent, dynamic, and desirable in my setup. I am in no way equipped to run DSD modulators and my DAC would convert DSD back to PCM (Yggdrasil). I have not adjusted any of the HQPlayer configurations though, as all I did within GentooPlayer was enable the HQPlayer service. 

 

Research will commence now...For those not upsampling or converting to DSD with modulators have you tried/compared HQPlayer to a simple uPNP renderer such as BubbleUPNP or MPD?

 

 

I have ran the MPD equivalent with upnp support, but the problem is my DAC is not PCM first so couldn't tell.  

 

Having said that, this isn't the first time I've seen PCM lovers make the switch to HQPlayer + NAA again with or without upsampling for SQ reasons, so you're not alone!

 

Don't feel obliged to switch to DSD upsampling especially if your DAC is natively PCM only, there's been a lot of effort by Jussi on the PCM upsampling side these days (to the point a few people have sold their Chord MScaler units - granted with a good host PC source).   

 

Worth looking in the HQPlayer threads on this forum for ideal settings for your DAC (Yggdrasil).

[guiltyboxswapper]

1 hour ago, romaz said:

Your post really isn't about the Extreme versus your DIY build but rather your preference for DSD upsampling versus NOS. 

 

Maybe @hols should simply do a comparison without HQPlayer and let us know how he gets on?

[romaz]

4 hours ago, Johnseye said:

Its interface for song/album/artist sorting is worse than Stylus (and Roon) unfortunately. 

 

I am finding HQPDcontrol on a tablet to be my preferred way of interfacing with HQP but I agree, this is where Roon really shines.

 

4 hours ago, Johnseye said:

If I want to browse not so much, and it has no Qobuz or other streaming integration so there's another key limitation. 

 

Yes, these are some of the downsides of HQP standalone which is why I'm eagerly awaiting Emile's new software player, TAS which will allow both Qobuz and Tidal playback and will have an HQP interface.  

 

4 hours ago, Johnseye said:

I've also found that the music stored locally on an Optane drive sounds better than over the network with HQP.

 

Yes, agreed.  Optane versus SSD is well discussed on this thread already and as @lmitche explained to me some time ago, Optane drives are based on memristor storage which means they need no refresh activity to hold data. SSDs need constant activity and that's why Optanes are supposedly quieter. They're quiet unless you access them and then they're so fast that the window of activity is really short.  Optane drives as internal storage provide a wonderful zero point of reference and I'll explain.

 

Today, I view the network to be a source of noise and network devices to be merely noise shapers more than noise removers.  Yes, I'm sure some switches are providing noise reduction or even isolation but these switches then add their own signature.  Properly implemented, they can provide a desirable signature (more bass, bigger sound stage, etc) but not necessarily a truly neutral or "noise free" one.  If one wants to hear what true silence sounds like, then play back a file from your internal storage and unplug your device from the network because how can any network configuration sound better than "no network."  With an unoptimized network, the difference will potentially be quite stark.  As you add 1 network switch followed by a 2nd switch, a 5th switch, even a 10th switch, I'm willing to bet the sound signature will continue to change with each switch since I believe each device is merely adding further to the sound signature by modulating noise in some fashion.  As proof of concept, if you connect 2 eRs to your network in serial via optical and you swap out the SFPs for a different brand, even through the eR's moat, why is there a sound difference?  If you then switch modems or routers upstream of whatever switch you own, why is there a sound difference if your switch is supposedly isolating so well and is regenerating a pristine signal with it's fancy OCXO?  In my view, as long as you are hearing a difference when compared to local playback with the network unplugged, then you are still hearing noise.  Some will use their network as a means to tune and I think that's perfectly fine but those who suggest they are getting better playback from their network than they are from local storage, imho, are preferring the noise created by their network.  

 

With Emile's latest round of tweaks, the network has less impact on my Extreme today than it has in the past to the extent that I can now leave my Extreme plugged into my network with no significant drawback and so this is good but I would never have fully appreciated it without having the point of reference of unplugging from the network and playing back from internal storage.

 

[Exocer]

31 minutes ago, guiltyboxswapper said:

 

I have ran the MPD equivalent with upnp support, but the problem is my DAC is not PCM first so couldn't tell.  

 

Having said that, this isn't the first time I've seen PCM lovers make the switch to HQPlayer + NAA again with or without upsampling for SQ reasons, so you're not alone!

 

Don't feel obliged to switch to DSD upsampling especially if your DAC is natively PCM only, there's been a lot of effort by Jussi on the PCM upsampling side these days (to the point a few people have sold their Chord MScaler units - granted with a good host PC source).   

 

Worth looking in the HQPlayer threads on this forum for ideal settings for your DAC (Yggdrasil).

 

Very helpful! Thank you @guiltyboxswapper. Now, to figure out why one would need an NAA vs Embedded alone on a single box setup.

 

So much reading to do . 

 

On another note, I have tried Audirvana DSD512 upsampling from my laptop (it sports 32gb of ECC ram, 2TB SSD, Xeon E-2176M which turboes up to 4.2ghz for this duty) to GentooPlayer based MPD renderer and I do notice an interesting result in terms of "space". I wish I were better with describing these differences but I can say that without DSD support whatever the Yggy does with DSD is still very audible. I can easily see myself preferring this mode for specific genres. Nice to have options.

[romaz]

2 hours ago, Exocer said:

 

I've tried enabling and streaming to HQPlayer Embedded after all of this HQPlayer talk... and so far I feel MPD or BubbleUPNP sounds more transparent, dynamic, and desirable in my setup. I am in no way equipped to run DSD modulators and my DAC would convert DSD back to PCM (Yggdrasil). I have not adjusted any of the HQPlayer configurations though, as all I did within GentooPlayer was enable the HQPlayer service. 

 

Research will commence now...For those not upsampling or converting to DSD with modulators have you tried/compared HQPlayer to a simple uPNP renderer such as BubbleUPNP or MPD?

 

 

There is definitely a case for NOS and sometimes this comes down to personal preference but also, there are some DACs that just sound better when fed a bit-perfect signal.

[The Computer Audiophile]

I believe the Chord DACs run at or near 100% of their FPGA capacity and are pushed to their limits. I wonder if pushing the Extreme to its limits would cause the Chord DACs to back of this limit internally and change the sound. 

[romaz]

3 hours ago, ASRMichael said:

I have been looking at the HQPlayer Pro option myself. Why? 
 

Up until I went through Convolution filter Process with Mitch I was using Stylus. IMHO neither, Roon or HQPlayer sounded as good as Stylus on my system. 

 

Now I have Convolution filters, I clearly need to use either Roon or HQPlayer. I thought I would give Roon a try again, in the hope it would sound good. I’ve made a lot of SQ improvements to my system since I last used Roon, so didn’t know what to expect. Rather than wasted time explaining, I moved quickly onto HQplayer. 
 

In the last two weeks, I’ve tried every possible setting in HQPlayer. I’ve now settled with 64DSD, EXT2, A7SEC. With great results. 
 

I’m loving the effects of Convolution Filters. Everything is dialled in. Am I now happy? No, there’s a trade-off...Convolution improvement, yes, versus the SQ improvement using just Stylus. I’m not saying HQplayer sounds bad, it’s just missing something that Stylus has. (I suspect light weight processing). So what’s next?....If it’s lightweight processing then HQP Pro makes sense, offline upscaling.

 

I downloaded HQPlayer Pro. Created some files offline, both PCM & DSD. Imported into Stylus. I’m thinking bingo, best of both worlds here, then I say keep listening. Then song ends. Unfortunately the Pro trial only gives 1 minute of the recorded song of listening time. I find it virtually impossible to trial properly. 
 

It’s £2800 to buy..am I buying it, if @Miska allows more than 1m payback. At least a song worth. 
 

Being an audiophile can be very draining at times!!!!

 

I, too, was frustrated by the limitations of the unregistered version of HQP and 1 minute of listening time is indeed very short and so what did I do?  I bought HQP Pro and I am both pleased but also have regrets.  I am pleased because it did everything it claims to do but I have regrets because somehow, I can't get it to run on my Extreme and so as of now, I am forced to run it on a server that is far inferior to the Extreme.  The process has been educational and what I am finding is that the quality of the files that you generate with HQP Pro is very much dependent on the quality of the server used to create these files.  Don't think you can just create files on any server and end up with excellent results.  For years, there has been debate about whether there are SQ differences between different CD ripping software with some swearing they can hear differences (I belong to this camp) and others suggesting there could possibly be no difference but my experience with creating new masters via offline upsampling with HQP Pro strongly supports that noise created by your server becomes permanently embedded in the new recording and even play back of these files through the Extreme won't fix it.  What I am hoping is that eventually, I will be able to run HQP Pro on my Extreme but for the time being, I'm using my HQP Pro as a proverbial boat anchor.

[romaz]

21 minutes ago, The Computer Audiophile said:

I believe the Chord DACs run at or near 100% of their FPGA capacity and are pushed to their limits. I wonder if pushing the Extreme to its limits would cause the Chord DACs to back of this limit internally and change the sound. 

 

This is true and is the advantage of pairing a Chord DAC with Chord's M-Scaler as it offloads upsampling to a separate device.  If a Chord DAC is relieved of it's upsampling duties, the DAC's FPGA essentially runs idle and even Rob Watts himself has suggested this is one reason why M-Scaler improves SQ.

[Blackmorec]

1 hour ago, romaz said:

 

I am finding HQPDcontrol on a tablet to be my preferred way of interfacing with HQP but I agree, this is where Roon really shines.

 

 

Yes, these are some of the downsides of HQP standalone which is why I'm eagerly awaiting Emile's new software player, TAS which will allow both Qobuz and Tidal playback and will have an HQP interface.  

 

 

Yes, agreed.  Optane versus SSD is well discussed on this thread already and as @lmitche explained to me some time ago, Optane drives are based on memristor storage which means they need no refresh activity to hold data. SSDs need constant activity and that's why Optanes are supposedly quieter. They're quiet unless you access them and then they're so fast that the window of activity is really short.  Optane drives as internal storage provide a wonderful zero point of reference and I'll explain.

 

Today, I view the network to be a source of noise and network devices to be merely noise shapers more than noise removers.  Yes, I'm sure some switches are providing noise reduction or even isolation but these switches then add their own signature.  Properly implemented, they can provide a desirable signature (more bass, bigger sound stage, etc) but not necessarily a truly neutral or "noise free" one.  If one wants to hear what true silence sounds like, then play back a file from your internal storage and unplug your device from the network because how can any network configuration sound better than "no network."  With an unoptimized network, the difference will potentially be quite stark.  As you add 1 network switch followed by a 2nd switch, a 5th switch, even a 10th switch, I'm willing to bet the sound signature will continue to change with each switch since I believe each device is merely adding further to the sound signature by modulating noise in some fashion.  As proof of concept, if you connect 2 eRs to your network in serial via optical and you swap out the SFPs for a different brand, even through the eR's moat, why is there a sound difference?  If you then switch modems or routers upstream of whatever switch you own, why is there a sound difference if your switch is supposedly isolating so well and is regenerating a pristine signal with it's fancy OCXO?  In my view, as long as you are hearing a difference when compared to local playback with the network unplugged, then you are still hearing noise.  Some will use their network as a means to tune and I think that's perfectly fine but those who suggest they are getting better playback from their network than they are from local storage, imho, are preferring the noise created by their network.  

 

With Emile's latest round of tweaks, the network has less impact on my Extreme today than it has in the past to the extent that I can now leave my Extreme plugged into my network with no significant drawback and so this is good but I would never have fully appreciated it without having the point of reference of unplugging from the network and playing back from internal storage.

 

Very interesting points. I have found that instead of adding multiple series switches to my network I get the best results by reducing vibration, optimising ethernet, AC and DC cable screening and most important, upgrading the network power supplies. My system comprises Synergistic Research Network and power cabling, Mundorf silver/gold JSSG360 DC cabling, all network components and PSs on Atacama racks and all network components powered by SJ DC3s.  The DC3s are currently undergoing upgrade to add DC4 regulation, Statement-level Transformers, internal casework damping, internal Mundorf S/G wiring and Isoacoustic footers.  

In the past I tried things like USB reclockers but they actually reduced SQ vs a simple SR USB cable

 

Back to your point, there’s no doubt that the network is delivering both a data stream and a noise stream. Adding more devices such as switches means adding more cables, more power supplies and therefore more noise. Adding then removing noise is basically noise shaping. In my experience the fewest number of high quality devices, linked by cables with the best possible screening and the purest possible power supplies seems the way to go. Also I believe there’s benefit in reducing the complexity of the noise spectrum by standardising on 1 type of network PS, thereby reducing harmonics generated when the noise spectra from multiple different PSs interact. 

[austinpop]

1 hour ago, The Computer Audiophile said:

I believe the Chord DACs run at or near 100% of their FPGA capacity and are pushed to their limits. I wonder if pushing the Extreme to its limits would cause the Chord DACs to back of this limit internally and change the sound. 

Adding to Roy’s response, which is spot on.

 

The nice thing is you don’t have to push the Extreme to its limit at all either. Using HQPlayer to do PCM upsampling to 705.6/768 is a negligible load on the Extreme, AND it allows the FPGA on the DAC to run idle. Win/win.

[ted_b]

2 hours ago, romaz said:

It's easy to think this and I used to wonder about this myself until my Extreme arrived and I heard what it could do with bit-perfect playback.

I meant as an NAA, but point taken. 

 

Rajiv and Roy, thanks for the responses about HQP and PCM upsampling.  Jussi has done a great job there, and even though he and Emile have not worked out how to run V4 on both cpu's (thanks Roy) it sounds like the slight cpu load is a great tradeoff vs the superiority of Jussi's engine and dither/filter choices.  I've always loved Rob's work, and have been a staunch supporter of Chord for many years.  There is no DAC that I've heard that surpasses PCM musicality better than a Chord.  But for me, a personal decision, I also have loads of DSD recordings...so my DAC and upsampling choices need to have a native DSD option, both 2 channel and multi. 

[ASRMichael]

1 hour ago, romaz said:

 

I, too, was frustrated by the limitations of the unregistered version of HQP and 1 minute of listening time is indeed very short and so what did I do?  I bought HQP Pro and I am both pleased but also have regrets.  I am pleased because it did everything it claims to do but I have regrets because somehow, I can't get it to run on my Extreme and so as of now, I am forced to run it on a server that is far inferior to the Extreme.  The process has been educational and what I am finding is that the quality of the files that you generate with HQP Pro is very much dependent on the quality of the server used to create these files.  Don't think you can just create files on any server and end up with excellent results.  For years, there has been debate about whether there are SQ differences between different CD ripping software with some swearing they can hear differences (I belong to this camp) and others suggesting there could possibly be no difference but my experience with creating new masters via offline upsampling with HQP Pro strongly supports that noise created by your server becomes permanently embedded in the new recording and even play back of these files through the Extreme won't fix it.  What I am hoping is that eventually, I will be able to run HQP Pro on my Extreme but for the time being, I'm using my HQP Pro as a proverbial boat anchor.

Noise in file —- Or the file is created with the errors created in the process, this then being duplicated on audio server, and more errors again. who knows!!! Maybe we’ll find out in 10 years time when we know more about digital affect on audio. 
 

I’m glad you mention this now, imagine i’d already spent countless hours creating my library offline files, & having to redo them........

[matthias]

@romaz

@ray-dude

@austinpop

 

Does someone have experiences with testing TAS (Taiko Audio Server) ?

Thanks

 

Matt

[ray-dude]

I have not yet, although much of the urgency went away after the July update (I’m at “better than pre Great Disruption” levels of SQ). Very much looking forward to testing TAS, esp as Emile and team get HQP integration working

[cat6man]

3 hours ago, romaz said:

 

Thanks for explaining.  If you are going to compare an HQP setup (intended for DSD upsampling) versus a machine like the Extreme that is not set up at all for HQP upsampling, then the results with be expectedly different.  If you really like the sound of DSD 256 ASDM7EC, then I would expect that your server that gives you this is going to sound better than a server that doesn't.  Your post really isn't about the Extreme versus your DIY build but rather your preference for DSD upsampling versus NOS. 

 

romaz,

thanks for clearly stating what i've been trying to clarify in my head

 

hols,

have you ever compared the SQ of your build vs. the extreme under identical, bit-exact conditions?

if not, would you be willing to do so please?

i think a lot of us would like to hear your conclusions and help us to understand your server build vs. extreme SQ.

 

thanks

[cat6man]

tangential to the topics but hopefully of interest

 

i've heard Rob talk about Chord products and his design philosophy a couple of times now and find him quite interesting.  i've also consistently found the Dave/scaler combo to be one of the best sounds at head-fi shows (and the scaler made a big difference).

 

at the last NYC audio meet, he had a lecture/presentation, after which he took questions from the audience.

i asked his thoughts on servers, networking, noise, galvanic isolation, etc

his response was that you should just run digital from your cellphone and the result will be as good as you can get.

[MarkusBarkus]

25 minutes ago, cat6man said:

...his response was that you should just run digital from your cellphone and the result will be as good as you can get.

 

[The Computer Audiophile]

1 minute ago, MarkusBarkus said:

 

That’s an analog phone :~)

[MarkusBarkus]

6 minutes ago, The Computer Audiophile said:

That’s an analog phone :~)

...don't be fooled. It's the new Chord "old school" DAC. 

[austinpop]

35 minutes ago, ted_b said:

I meant as an NAA, but point taken. 

 

Rajiv and Roy, thanks for the responses about HQP and PCM upsampling.  Jussi has done a great job there, and even though he and Emile have not worked out how to run V4 on both cpu's (thanks Roy) it sounds like the slight cpu load is a great tradeoff vs the superiority of Jussi's engine and dither/filter choices.  I've always loved Rob's work, and have been a staunch supporter of Chord for many years.  There is no DAC that I've heard that surpasses PCM musicality better than a Chord.  But for me, a personal decision, I also have loads of DSD recordings...so my DAC and upsampling choices need to have a native DSD option, both 2 channel and multi. 

 

Hi Ted,

 

As you've been following this thread, you've seen us go through all combinations of distributed and standalone topologies in the quest for SQ. Roy's excellent experiments a year or more ago with Roon showed that when holding the endpoint constant, and varying the quality of the upstream server, that upstream server's speed and quality (PSU etc) had an effect on SQ. Ultimately, we found the ideal (at least for some of us) to be a standalone server like the Extreme. But of course this is predicated on very low utilization, which is a consequence of bit-perfect or PCM upsampling playback.

 

I too have a lot of DSD content, but I was actually pleasantly surprised to find how well Rob's M-Scaler, TT2, and DAVE handle DSD content. So I'm not sure I would equate the possession of loads of DSD recordings to naturally pose a requirement of native DSD in a DAC. Still, we all have our preferences, and certainly some DACs are known to shine with native DSD input at high resolutions like 512. But that takes us to DSD upsampling.

 

 

The challenge with DSD upsampling is that it is so CPU-intensive. Once you introduce this behavior, then the assumption of a low-noise standalone server doesn't hold. And even if you say — use the Extreme as an NAA and shift the CPU-intensive work to an upstream server, then Roy's findings with the impact of upstream servers makes me think the quality of this server would be very important too.

 

This then pushes us to posit that maybe a high-end server built for HQPlayer — like the Pink Faun or the SGM 2015? — in combination with an Extreme running NAA would yield the desired SQ? Who knows, and that's a heck of a lot of $$$$! Maybe someone will try this one day.

 

I really think CPU-intensive DSD upsampling requires its own set of intrepid explorers, to find the optimal SQ and configuration. They do have their work cut out for them!

[ted_b]

10 minutes ago, austinpop said:

I really think CPU-intensive DSD upsampling requires its own set of intrepid explorers, to find the optimal SQ and configuration. They do have their work cut out for them!

This is true in most sense, but, remember, we can, for the paltry sum of $3k (slight tongue-in-cheek when compared to Extreme budgets) somewhat alleviate the issue, or at least move it to some other time (i.e upsample offline with HQP Pro).  Then the debate starts as to what does a good upsampled file-making server look like, and does it need the same strengths as Extreme.   Ah, open one door, find five more to explore!    Thanks to you and Roy (and Larry and others), we have much fewer doors to deal with.

[austinpop]

2 hours ago, ted_b said:

This is true in most sense, but, remember, we can, for the paltry sum of $3k (slight tongue-in-cheek when compared to Extreme budgets) somewhat alleviate the issue, or at least move it to some other time (i.e upsample offline with HQP Pro).  Then the debate starts as to what does a good upsampled file-making server look like, and does it need the same strengths as Extreme.   Ah, open one door, find five more to explore!    Thanks to you and Roy (and Larry and others), we have much fewer doors to deal with.

 

Hi Ted,

 

Actually, you may have hit on the perfect solution — use an offline HQPlayer Pro to generate the upsampled files. Yes it requires a lot of storage (which is relatively cheap), but it does mean that during playback, you can adhere to a bit-perfect regime.

 

4 hours ago, romaz said:

The process has been educational and what I am finding is that the quality of the files that you generate with HQP Pro is very much dependent on the quality of the server used to create these files.

 

Of course, this is still a problem, and a real head-scratcher. But for DSD upsamplers, this may be a small price to pay to optimize the playback path.

 

 

[romaz]

3 hours ago, matthias said:

@romaz

@ray-dude

@austinpop

 

Does someone have experiences with testing TAS (Taiko Audio Server) ?

Thanks

 

Matt

 

Yes, I agree with @ray-dude, my Extreme is currently in a state of "perfection" where I am just loving what I'm getting.  Periodically, I will step back to Roon so that I can explore Qobuz or Tidal and when I find something really good, basically, I buy it and store it locally onto my Extreme and play it back via HQP.

 

I have had the opportunity to test one particular feature that will be part of TAS and while I don't have the authority to really speak about it, it is enough to know that TAS + HQP should sound better than HQP by itself.  

[cat6man]

5 hours ago, romaz said:

 

There is definitely a case for NOS and sometimes this comes down to personal preference but also, there are some DACs that just sound better when fed a bit-perfect signal.

 

TotalDAC, in my experience, is one of those

[Johnseye]

5 hours ago, romaz said:

 

I, too, was frustrated by the limitations of the unregistered version of HQP and 1 minute of listening time is indeed very short and so what did I do?  I bought HQP Pro and I am both pleased but also have regrets.  I am pleased because it did everything it claims to do but I have regrets because somehow, I can't get it to run on my Extreme and so as of now, I am forced to run it on a server that is far inferior to the Extreme.  The process has been educational and what I am finding is that the quality of the files that you generate with HQP Pro is very much dependent on the quality of the server used to create these files.  Don't think you can just create files on any server and end up with excellent results.  For years, there has been debate about whether there are SQ differences between different CD ripping software with some swearing they can hear differences (I belong to this camp) and others suggesting there could possibly be no difference but my experience with creating new masters via offline upsampling with HQP Pro strongly supports that noise created by your server becomes permanently embedded in the new recording and even play back of these files through the Extreme won't fix it.  What I am hoping is that eventually, I will be able to run HQP Pro on my Extreme but for the time being, I'm using my HQP Pro as a proverbial boat anchor.

 

Very interesting hypothesis and/or discovery.  I've made efforts in the past to re-rip CDs.  I haven't taken the splurge with HQP Pro yet, but it is more food for thought when doing so.  Thanks for this info.

 

Another downside of HQP Pro is that you can't use it seamlessly with Roon, so if you're using Roon it isn't a replacement for HQP Desktop or Embeded.