Does highend music server make no difference to sound quality as per Roon’s advice or has the advice been proved as WRONG?

I have read this thread three years ago and was influenced and skeptical about buying highend music server, mostly because Roon’s general advice that music servers make no difference to sound quality but only need a good fiber isolation set up somewhere in the streaming chain

After three years, is this opinion still correct as we have seen more proof and user reports on the opposite with new innovations from Antipodes, Taiko and many other highend music server makers.

Besides streaming audio, the only other job of a music server is to do DSP. If you don’t use that, any PC will do. Even if you do use DSP and stream to different rooms simultaneously, a $1,000 machine is probably still too much.

If you get your music direct from a server (USB/HDMI DAC direct connected or sound card or whatever) then anything in that server can potentially have an impact on the sound.

The general consens (sort of, that’s hard to achieve in this hobby) is to keep the computer out of the audio chain.

So, if you use an endpoint (or more of them) then all the server has to do is to make sure that those end points are getting what they are expecting (very bit perfect :slight_smile: information). For the most of it any computer does that. Any other tasks (library indexing, DSP, transcoding and so on) that a media server must do, are directly related to the pure computing power. An i7 machine will generally do better than an i5 and so on.

If you made it up to here, just one more thing: there are also end points and end points!

1 Like

Danny did not say servers make no difference. Here is what Danny said:

Danny believes the server should be considered to have no influence on sound quality, so the server should be high power per dollar, and preferably high enough power to cope easily with anything Roon throws at it. I have no doubt that Danny’s views are honestly held.

Danny replied:
Not quite… I believe that the server either does or does not have an influence on sound quality.

If it does not, get the server that provides the best experience in all regards, ignoring any claims about sound quality.

If it does, there are things you can do (like isolation via networked endpoints) to eliminate the explainable effects that you believe might be attributing to negative sound quality. Therefore you should get the server that provides the best experience in all other regards, ignoring the claims about sound quality.

1 Like

The server also feeds the dac a signal that is far less noisy that what a computer provides.

Isn’t the DAC fed a digital signal? How is that ‘noisy’?

And this is exactly why I left the Audiogon forums…
Geoff Kait would have fun here methinks!

It seems no Roon endpoint on the market fully immune to upstream servers or does not benefit from better music server design till today, even world-class streamers from MSB, DcS, Lampazitor, etc.

There will always be room for improvement. But, as the SQ keeps gets better, it is more and more difficult to attain even further improvement, and usually more and more expensive.

I thnk it all really depends on what your doing with the server. If you just using it as such and its away from all the audio kit and accessed only via network then I doubt any high priced alleged high end server makes a difference. If your using it as a streamer then having less noisy compoenets and lots of isolation of PSUs etc will lkely help it rasie the bar. As Roon strongly recommnends separating core from playback then one would assume it does matter . Some still prefer local playback from devices such as Innous to network playback it also cuts down on devices, some rave about Melco and they are just a gloridied nas but your spending money that might be better used elsewhere or roughly amounts to the same thing as separating the core and having a decent network endpoint when compar the overall cost is taken in to account. Be interesting to see if spending 3000k+ on a Zenith is any better than a nuc and endpoint of say 1500.

I believe that we have seen more user reports to the contrary but, afaik, still no proof.

I wouldn’t worry about it. If people make the mistake of thinking some kind of high-end “audio computer” is necessary and spend extra cash on it instead of, say, better bottles for their wine cellar, or donations to the charities that bring clean water to the benighted places of the world… Well, people make mistakes continuously, 24 x 7, around the world. Not much to be done about it.