Sound quality of Roon build 537 [retracted claim]

Windows enterprise 2019 Ltsc is nothing like the iot version. It is a full windows without the modern apps, store or Cortana. I’ve been using it to run my windows machines for years . Best win 10 version hands down imho.

Daniel, typical Microsoft branding confusion. It may be a locked down version of full Windows 10 Enterprise, but Microsoft still slaps the IoT label on to it. Same thing with Skype and Skype for Business - two different technologies given the same branding.

Anyone fancy an EUR 825 chopping board ?

https://taikoaudio.com/product/daiza/

Not to mention the block of wood with a wire coming out of it for €500…

Supreme pricing indeed - US$40k for a 3m PC… makes the SGM Extreme look like a bargain.

Haha, I love this bit:

“We do not have much of an understanding of how this sonic signature migration occurs, but it is identifiable and can work nicely in some systems.
Audiophiles in Asia, Europe and the US have found that Setchi’s can improve transparency, clarity, resolution of details and tonality.”

I feel sorry for those poor Africans who cannot hear the difference… :rofl:

3 Likes

It can but that is more for keeping disruptive software at bay, it’s a sticking plaster for duff software, which is what it sounds like JPLAY is given the depth it tries to work at.

I’m getting a feeling that there is a lot running on this super server which is trying to take control of cores or sub systems and going down that route is a never ending path of pain.

Also the CPUs are likely to be relatively low spec devices in that Intel range - limited cache which will have a significant effect if he’s running a lot of memory intensive processes. Would be better off with one CPU with high clock speed and as much cache as possible.

When you talking about transport (software and or hardware) that is interfaced to a USB-DAC. The timing needed to send the data to a USB-DAC must be such that the DAC asynchronous memory plays a major role of holding the data. This is because the asynchronous memory is clocked by a highly stable clock system (Low jitter and noise)

Ideally, the faster the source is able to fill the asynchronous memory the better it is the SQ. I did a few experiments to vary the amount of buffering from 100 to 1000ms, and the SQ begins to degrade by a lot when hit above 400ms. Modern USB-DACs has larger asynchronous memory so the source buffering no need to be that large anymore. 100ms to 200ms works fine for USB connected DACs.

1 Like

Canadians, Mexicans, Central and South Americans too, it seems.

3 Likes

That makes sense for Central America. Greater centrifugal force near the equator.

4 Likes

Completely agree. Except, what you are referring to is bandwidth, not latency. In the road and toll both analogy, bandwidth is number of cars coming off the road per second (to fill your buffers) and it really doesn’t matter how long the car was on the road, or if the road was shorter or longer - latency.

Now The bandwidth of USB is anywhere from 400 mbps (ISB 2.0) to 10gbps and up (USB 3.0). Modern computer architecture “is” fast enough to feed these rates and is mostly limited by the source (disk or network for example). A 5 minute long 24 bit DSD file is about 150 MB, or about 5 mbps.

If your PC is struggling to keep up with that rate, then you’re likely doing some heavy DSP processing (HQPlayer?) on underpowered hardware. Modern hardware can easily feed 4K video files, often with real time transcoding. Simply serving a computer audio file to an IO port should be child’s play.

4 Likes

Put differently, if a computer can process 10,000 widgets per second, and if we added a few extra processing functions, and let’s say that rate drops to 9,995. This change alone cannot affect SQ, as we only need 5 widgets per sec, and we are well above that. If we now throw in some massive DSP and upscaling into the mix, and the processing drops to below 5 per second, then - yes it would affect SQ.

So context matters - and this is what Danny alluded to when he said it’s like the bus driver is blaming the extra passenger (ignoring the weight of the bus and the 1000 other passengers)

1 Like

@Larry_Gelman I feel the same way after a bottle of my favorite Bordeaux!

2 Likes

This is absolutely true, but it’s also important that we keep in mind that the effect on SQ is stuff like skips, dropouts, etc. It is not “soundstage” or anything like that. I’m not saying soundstage isn’t a noticeable part of SQ (it is) but it’s fully and completely encoded in the digital information reaching the DAC. Affecting the digital signal layer has no effect unless the signal degradation also affects the higher, information, layer. This is unlike analog, where the information is the signal.

5 Likes

That’s called a write-only memory.

This is the quote from that link that I love:

“When a Setchi is connected, there is an immediate effect. There is also a secondary effect with takes two to three days to become quite noticeable.”

A € 500.00 grounding device connected to nothing but what is in the box and has a secondary affect that takes two to three days to become noticeable? I guess it doesn’t’t take much to separate a fool from his money…

2 Likes

Remember, centrifugal force is a fictitious force. It’s not real…so it is perfect for this discussion!

Is this SQ discussion going only about connection via USB? Guess many users are using Ethernet connection between Roon / Server and streamer.

I’m embarrassed to say I hadn’t remembered that, but I like your take on it making it even more appropriate!

USB 2.0 have more than ample of bandwidth. I’m referring to a USB output buffering at the source when you connect to a USB-DAC. The rule of thumb is to have minimum buffering at the source and a large asynchronous buffer or memory at the DAC side. This means there’s less latency involved.

Unfortunately, most USB buffering at the source is fixed at high level and it is non-adjustable. Fixing at higher buffering ensures older USB-DAC with less asynchronous memory able to cope without over-fill. (Safety net). It will great if it has an adjustable output buffer. You need to adjust to the lowest where you start hearing some ‘clicking’ or dropout in the sound. Then slowly increase by 50 to 100ms until no dropout is heard, you will have a optimum buffering settings to the particular type of DAC used.

I would recommend the use of integrated DAC streamer; this is most direct point to point connection via I2S that yield much better SQ if it is properly constructed :grinning: