What is the Best music playback media?

I have a very simple question regarding the “noise” from the music playback.

“Noise” such as electric noise on network / USB key / HD / SSD plus “Noise” such as the mechanical induce jitters from the CD transport / Hard drive mechanism.

Suppose I have a 44.1/16 WAV file on different media:

  1. On standard Redbook CD (e.g. play via PST)
  2. On DATA Disk (such as DVD+RW, play from CD player, e.g PST)
  3. On the USB key (e.g. play via USB port on CD player or music server)
  4. On the Hard drive of a music server, (e.g. Melco N10)
  5. On the SSD drive of a music serve, (e.g. PSA Octave music server)
  6. On NAS file server (e.g Synology DS720+, play from streaming device)
  7. On PC (or MAC) (play from USB port on a DAC)
  8. On music streaming service site (e.g. Qobuz, play from streaming device)

What is the best media to play?

Does “galvanically isolated output stage” a magic silver bullet to make difference music playback irrelevant?

I benefitted from galvanic isolation in the media server. So, I would imagine it is good in more than one place.

If a component/device makes noise that no one can hear, is it still noise?


On my system its;

  1. Melco built-in audio graded SSD
  2. Extern usb-SSD into Melco (very very close)
  3. Streaming Qobuz (very close)
  4. Disc played from PWT to DS (close)

(Etherregen (Uptone lps1.2) > Melco N1ZS10 > AQ diamond usb > Matrix (full loaded Farad super3) > AQ diamond hdmi > Directstream)

Great thought Palouse…

Even though equipment sourced noise may be below the speaker’s ability to
resolve…perhaps with high efficiency speakers might loose a bit of resolution
clarity because of equipment sourced noise…Forgot where; but it seems I read
that the minutest faintest whisper of a detail would be as in parts per million as in
a chemistry wise analogy…

In chem labs the water used is multiple times distilled for control to ensuring a result
that is free from contamination of the processes…understand I am not a chemist…
but recall the term ppm parts per million…

So for those with beautifully treated quiet music rooms…this “equipment sourced
noise” could have an adverse effect overall …however slight this may be.

So my .02 thinking is that yes ultimately it is still noise… :frowning_face:

Best wishes …

I was being partially flippant and partially rhetorical while waiting for my coffee to brew—but looking at it the way you describe it, yes, it is noise. Between my tinnitus and the forced air furnace I don’t feel the need for a “better” music system, unfortunately! Though it maybe cheaper than a new furnace! Hmmm :thinking:


Hard to do but …being content with what we have…is great gain

So being flippant you were eh? Hmm so instead of using bug spray
to fight off the ants…just flip 'em out there :innocent:

Some of the options are easy and cheap enough to setup and do comparisons on. I recommend you try in your system with your music and assess with your own ears. If you can’t hear a difference then chances are you’ll be fine with any of those option.

I recently compared between my PC going over USB over ethernet device and an Intel NUC (no fancy ssd) and I heard minor but identifiable and more enjoyable differences in treble extensions and sound staging. So now next step is get the NUC a LPS.

Thanks everyone for your valuable comment.
For me, music file is just a music file (0/1 in binary form), it does not matter where I store it. I think it all come down to how the gear is been designed to handle / isolate the unwanted “noise” during the process to retrieve the “music”(I.e. PCM or DSD) from the music file.
As my own story goes, I have DSDAC, and I never thought CD transport will make any sonic performance difference, since any transport will get the same digital bits. Later, I have PST, I can hear a great sonic performance improvement on my system. My redbook CDs come to live. How come a “disk spinner” will make sound difference? Might be just a psychological effect that PST is an “expensive” disk spinner.

It is not psychological. I do not have the PST yet, but when I first got the DMP, I have the same reaction with this player. The sound was much better than my server the Aurender N10 at that time. But my DMP was going into the DS DAC with I2s which is the better input and the Aurender was using the inferior USB input. I decided to get the Matrix USB to I2s converter and try the Aurender through the I2s input. Long story short, The Aurender with the Matrix and premium cable and LPS blew away the DMP in sound performance. One day when I get the PST, we’ll see how that measure up to the server.

1 Like

Great topic! I’ve tried many options over the year to improve the “source” problem. In trying to provide “real time” music most “disc spinners” create jitter and data errors that are not all corrected. Then, the interface to the internal or external DAC is generally flawed.
In my understanding, the challenge is to provide the cleanest and least noisy bit stream to the
DAC (assuming it reclocks the data). That’s why the PST and DS DAC with the I2S interface work so well. Hence, it should not matter how the WAV file “travels about and gets close” to the interface (USB, I2S, etc) - it’s the final conversion to the bitstream, plus the interface into the DAC input that present the first challenges. Then the DAC has to do a great job…!

1 Like

The problem with the question is there is no way to test this without changing so many variables the test is no longer tracking the 1 variable you want; it is tracking other variables.

In my system I have a Rega transport and an Allo USBridge Signature (digital transport). They are both hooked up to the same DAC but different inputs on the DAC. Does that give me the ability to compare “noise” between Redbook and files? Absolutely not. Just changing the input interface on the DAC invalidates the test as we now have different “noise” characteristics influenced by things other than the source media.

Pick the media of convenience and optimize it. Each source will have ways to optimize. I’m on the path of digital separates. Keep the spinny vibrating high frequency stuff on the other side of the house and optimize the bit conversion to analog where the speakers live. I’ve got a ways to go on my journey but its the path I’ve chosen. Also, my amp produces more noise than anything else in my system right now so if I wanted to “fix” noise I’d focus on that.

Then, and I cannot stress this enough, if you’re room isn’t treated don’t bother. It’s amazing how “not quiet” a system becomes the deader you make your room. If you’ve got a live room don’t bother so much with trying to drive the “noise” down. My family room system might be noisy by audiophile standards but I dare anyone to tell me they can “hear the noise” over the refrigerator compressor in the kitchen which is behind and to the left of the sweet spot.

You’re confusing different types of jitter here. In a digital system serving files / streaming a lot of the “jitter” you hear about has nothing to do with the jitter that influences playback. As example “network jitter” cannot influence playback beyond the playback failing if the jitter becomes so irradiate you underrun a queue / buffer. Otherwise, the playout buffer has its own clock which is completely independent of whats happening before it.

But, to answer your question, the “best” is whatever you’ve optimized for in your playback system. In my network that is Roon. Why Roon? Because it levels the playing field between streaming and files. Everything comes into the Core and gets converted to RAAT. I then optimize the RAAT playback path. Doesn’t matter what the source is (files, streaming, internet radio, etc.) I get consistent characteristics of the playback system. And, upgrades, benefit a wider range of source media.


Indeed, as said

You have Roon, and I have PST, since for me, PST reduces my CD playback “noise” in my HIFI setting.

Not really confuse with “jitter”, the

I means the mechanical vibration of spinning drive (CD / HD) impacts to data access quality. Not the digital type of jitter that DSDAC has already taken care to it.

Most CD mechanisms never implemented error correction properly. Additionally, reading from a CD and feeding a DAC is almost real time. Very shallow buffers are used so most CD transports don’t have the ability to read the same bits more than twice before you have to give-up, feed no or the wring data, and move on. Reading data via a HDD is very different. There are very stages within the drive itself, the filesystem, the OS, etc. which guarantee data integrity. If that data cannot be reliably read a read error occurs and you get nothing. You don’t get the wrong data you actually get nothing. Now, depending on the filesystem, how many times it tries to read the data could be very long (minutes, again the system is guaranteeing data integrity and it has no idea the application is time sensitive to the data being requested ). You’re playback would have failed way before the system has reported a fatal error and abort.

You mention HDDs, spinning disks, which are more rare these days but lets talk about them for a bit and how robust they actually are and their failure modes. First, to level set, a HDD works by having one or more platters with a suspended head per platter. The flying height, or distance between head and platter, can be about 5 nm (nanometers). And these things operate just fine, with no data loss, on planes, trains, and automobiles. The failure mode from “vibration” is catastrophic in that the head comes in contact with the platter. That, effectively, destroys the drive. What I’m trying to point out here… vibration at the HDD has no effect on data access quality. The drive is designed to operate with vibration from its very nature of being portable. It will operate this way right up to the point it parks the head (almost all drives have the technology to detect motion and park a head before it risks crashing into the platter). If there is an emergency park operation the result of that can be the OS crashing and you needing to reboot the machine. But, again, these are “stop playback” kinds of failure not “got the wrong bit from the HDD” failures.

But, all modern drives like SSD, nvme, m.2, USB flash, etc. have no moving parts. They have different failure modes but vibration is not one of them. These drives are, effectively, immune from vibration. These drives, their file system layer, and the OS still guarantee data integrity.


The problem with Roon is that it is processor heavy. I use Roon on an Innuos Zen Mk3, optimised for low noise, which I have again optimised with a decent switch, mains conditioning and sending data via ethernet rather than usb. Innuos are further optimising by providing their own streamer software inside 6 weeks. This has involved various updates to improve file mangement. So I will probably stop using Roon on the Innuos, and use it as a standalone server/streamer, much like the planned Octave player. I will run Roon for my other endpoints off a TS-473 QNAP.

There are loads of variables, but one well-designed box and decent software should be able to manage it. It seems to be more how the data is managed than where or what it is stored on, or streamed from.

1 Like

Why does it matter if its processor heavy if its isolated away from everything else? If you mean RAAT is “heavy” compared to other streaming protocols then, yes, that may impact sound quality. But Roon Core being CPU / memory bound / heavy really doesn’t matter. Just throw hardware at it and be happy.

It really doesn’t make sense to optimize the hardware of a Roon Core for low noise when the reference Roon architecture eliminates any noise from the core landing at any of the D/A stages of playback.