NAA endpoint, useless?

[widemediaphotography]

Hello everyone.
More and more often you hear about the necessity of endpoints in systems that involve a NAA (Network Audio Adapter).
With the availability of a server, whose task is to make audio files available for multiple end users in different locations, I believe that this solution, when applied to stand-alone systems with a single renderer, is truly just a complication with no benefits.

I have started a series of tests where I essentially contrast a simple system:

Audio-pc (DIY) / WINDOWS 11 (ReviOS) with a local library / foobar2000 – USB DAC (smsl DO300), one classic and much used stand alone "bit-perfect" configuration

with a system:

PC/Server (any OS) – ethernet-- Raspberry/minipc (some dedicated OS/HQplayer NAA/but never Windows) as an NAA Endpoint – USB DAC (smsl DO300)

It's easy to find on the web variations of the second solution, some very sophisticated with dedicated Endpoints like sMS-200 or with DACs that include an endpoint as some models (GUSTARD R26 for Roon) where the presence of such "jerry-rigged contraptions" would be justified, even through a series of scientific contradictions.

There are different reasons given for this necessity, from the origin which stems from the need to use upsampling and advanced filters on powerful PCs that may not be galvanically isolated with the DAC.

The first contradiction would be that the Raspberry/NUC would remain connected via USB to the DAC, which wouldn't isolate the DAC at all. The second is based on the belief that Galvanic isolation would significantly improve sound quality and then I wonder why not prefer an already isolated Optical or Electric connection like AES/I2S, but we all still prefer USB, not just for bandwidth reasons.

This is not the thread in which we simplify with phrases like "if you don't hear differences, live in peace", but a discussion where technical arguments are brought together, possibly supported with parameters and measurements.

The fact that I have not only found complications and useless things at the moment does not mean that the problem is solved for me and for everyone. There are thousands of people out there who use completely dedicated hardware and software for that purpose, so before stating that it is a collective madness, it would be worth saying everything that needs to be said.

I also understood that talking about HQP here is not very welcome, but if we have a scientific approach we must accept any comparison...

Please, let me know what you think, really.

thanks
wide

 

[nerdstrike]

I googled "aes i2s" hoping to understand the interconnect and almost every hit was audiophilia... There's a problem I never realised existed!

Your two setups have huge amounts of varying software involved. Lots of scope for divergence from surprise settings/bad defaults in there. I recommend being extra careful when reporting your first "aha! I told you so!"

 

[MaxwellsEq]

I believe that this solution, when applied to stand-alone systems with a single renderer, is truly just a complication with no benefits.

I've not looked into NAA, but from what I've seen, you may be right, certainly for domestic use cases. I know a bit about very large media centres using large servers and thousands of independent workstations editing and playing audio. These ARE necessarily complex based on current, clock-dependent IEEE1588 architectures. But it's possible to build a NAS-based domestic model with multiple devices reading audio from the NAS, bit-perfectly.

 

[ShadowFiend]

The most important feature of NAA/Roon Endpoint is that you can place the music storage and computing PC out of listening room. It will remove all kind of noise from fans/disk spin. I always find it amusing when people concerning about 0.00x SINAD number but using a PC/laptop as a music source with their noisy fans nearby.

 

[MaxwellsEq]

The most important feature of NAA/Roon Endpoint is that you can place the music storage and computing PC out of listening room. It will remove all kind of noise from fans/disk spin. I always find it amusing when people concerning about 0.00x SINAD number but using a PC/laptop as a music source with their noisy fans nearby.

Or people who live in countries with air conditioning and air-based heating. They must enjoy the spring and fall when their rooms are quiet!

I use a dedicated hardware "streamer" that outputs S/PDIF into my DAC. No NAA required and zero mechanical noise.

 

[widemediaphotography]

The most important feature of NAA/Roon Endpoint is that you can place the music storage and computing PC out of listening room. It will remove all kind of noise from fans/disk spin. I always find it amusing when people concerning about 0.00x SINAD number but using a PC/laptop as a music source with their noisy fans nearby.

My DIY audio-pc is already zero noise and zero ground loop (has even integrated one topping HS02 + ultralinear low noise PSU)

 

[widemediaphotography]

"Another important issue would be to verify whether indeed a PC + specialized software for upsampling and various filters + a native DSD DAC can perform better than a PC with bit perfect software and a good D-S DAC. If someone has not already done so, it would be necessary to create in the digital domain a set of about thirty sweeps at known frequencies ranging from 20 to 20kHz, for example square waves, and overlay the acquired results on the analog output of the DAC to verify if indeed a PC + software perform better than the DAC. Perhaps someone has already done this and it would be interesting to evaluate the results."

 

[voodooless]

square waves

Please no! If an upsampled band limited square wave looks a proper square wave, you can be sure that the upsampling is flawed.

 

[Vincent Kars]

The first contradiction would be that the Raspberry/NUC would remain connected via USB to the DAC, which wouldn't isolate the DAC at all.

Never heard of an USB isolator?
Never heard of an USB receiver doing a good job?

 

[SSS]

"Another important issue would be to verify whether indeed a PC + specialized software for upsampling and various filters + a native DSD DAC can perform better than a PC with bit perfect software and a good D-S DAC. If someone has not already done so, it would be necessary to create in the digital domain a set of about thirty sweeps at known frequencies ranging from 20 to 20kHz, for example square waves, and overlay the acquired results on the analog output of the DAC to verify if indeed a PC + software perform better than the DAC. Perhaps someone has already done this and it would be interesting to evaluate the results."

If using a PC with specialized software player and feed a DAC then the audio signal to the DAC will be bitperfect. If the analog output differs then the DAC is responsible not the PC. Of course it is not easy to compare the analog signal with the bits in an audio file because of a ADC of perfect quality needed. Since no ADC is perfect in a strict sense differences will occur.

 

[widemediaphotography]

Please no! If an upsampled band limited square wave looks a proper square wave, you can be sure that the upsampling is flawed.

What methodology would you use to verify, at the various frequencies of the audible range, whether a PC+SW+DSD direct is actually better than a good D-S DAC? At least in theory this is the case, because the computing capacity and versatility of a PC is certainly better than any DAC chip and HW filter implementation.

 

[widemediaphotography]

Never heard of an USB isolator?
Never heard of an USB receiver doing a good job?

I already wrote that my PC audio has a built-in HS02 Topping and ultralinear low-noise power supply. In another thread that my smsl DO300 DAC does a great job because even without an isolator (not all of them work well) even turning up the volume of the Violectric V222 amplifier to the maximum I don't hear any noise during pauses through the HIFIMAN HE1000 headphones, same goes for floorstanding speakers.
In the introduction I indicated that there would be no discussion about whether my audio chain did a good job, but whether the NAA endpoint complication made it better and whether there are measures to prove this.

 

[widemediaphotography]

If using a PC with specialized software player and feed a DAC then the audio signal to the DAC will be bitperfect. If the analog output differs then the DAC is responsible not the PC. Of course it is not easy to compare the analog signal with the bits in an audio file because of a ADC of perfect quality needed. Since no ADC is perfect in a strict sense differences will occur.

I agree when you say that the comparison is not easy, but we also cannot accept that golden ears prevail over measures. There will be a way to clarify one thesis rather than another in a scientific sense. If I believe that using a NAA endpoint is just an “improvised contraction” there will be scientific evidence to either confirm it or establish that I am wrong.