Interesting about gigabit speeds in home audio. I’ve read that given how little tha bit rate required for music, some have argued that gigabit implementation works against the whole premise of Diretta, as it pushes ever larger caches of data with less frequency. I think the problems of latency and cache management have effects similar to jitter on the server side.
The UpTone product I’m waiting for is their promised replacement for the discontinued ISO REGEN usb decrappifier. Now that David’s dual pi isolator effectively filters out any network interference, my attention turns to the signal path into my active loudspeakers…
Hi Veryy interesting. May I ask for some help understanding?
Myconfiguration is just using a Rock configuration in a Nuc Intel i5 optically isolated from the router , with usb output directly to my usb DAC with a usb cleaner between the Nuc and the dac. Quality is quite good But I believe your would work Do you think your setup added to mine will improve quality?
Welcome to the thread! That is an excellent question.
Based on your description, I believe you are in the ideal position to hear an improvement from this project. You’ve hit on the exact reason many of us started down this “tinkering” path.
The philosophy behind Roon’s architecture for sound quality is to separate the “heavy lifting” (the Roon Core) from the lightweight “music player” (the Roon Output/Endpoint). This is done to protect your sensitive DAC from the electrical noise the Core generates.
In your current setup, your NUC is doing both jobs at the same time. It’s acting as:
- The noisy Roon Core (managing the library, background analysis, DSP, unpacking FLACs, talking to streaming services, handling control requests, etc.)
- The Roon Output (running the Roon Bridge software that feeds the USB signal to your DAC)
All of that server activity creates a constant “tidal wave” of electrical noise (RFI/EMI) inside the NUC, just inches away from the USB port that’s feeding your DAC.
Your “USB cleaner” is like putting a Band-Aid on a shotgun wound—it’s trying to filter an overwhelming amount of noise. (Your optical isolation from the router is great, but the main noise source is inside the NUC itself.)
This is what my project aims to fix.
It would allow you to finally separate those two jobs, just as Roon’s designers intended:
- Your NUC would only be the Roon Core (the brain, which you’d keep on your network but relocate to a different room or closet away from where you listen to music).
- The Diretta Host + Target would become your new, dedicated, ultra-quiet Roon Endpoint (the music player), which then connects to your DAC.
This project would physically isolate your DAC from all that server noise. It’s the ultimate expression of Roon’s core-and-endpoint philosophy. I am very confident you would hear a substantial improvement and have fun doing it!
I’m guessing you’re well aware of current alternatives to Uptone’s ISO REGEN from Schiit and iFi and probably others. What were you hoping for and are these for other active speakers than the ones mentioned you’re using with USB into AES/EBU?
Very interested in proceeding to experiment.
That said, I read the entire thread but, unless I missed something (I’m a proud senior citizen so it is possible), I did not see any external links to how-to instructions, or even a high-level bulleted list of steps. Or maybe an active users group?
Thnx for any feedback/suggestions.
P.S. I’m a retired software engineer. I have multiple Ropieee endpoints (a couple with 3rd-party HATs) that I setup myself. Recently trying a pure streamer (no DAC) – the FiiO SR11. I’ve tried linear power supplies, DDC reclockers, better cables and – IN MY EXPERIENCE – they don’t make an easily discernible difference. I use a Motu 828 (2024 version) for my interface/DAC to a Parasound A21+ to Magnepan 1.7i’s. I’m just a lifelong audiophile still trying to improve things by “eliminating doubt” via bang-for-the-buck experiments.
Yes, I am using a Singxer UIP-1 PRO in the meantime which connects to my Singxer SU-6 DDC for conversion to AES/EBU. Knowing Alex and John, I expect the UpTone product to far outclass what some think is the best of the bunch (for now).
I convert the DDC SLR connection to Meridian’s SpeakerLink wiring over CAT8 so that I can easily plug in to my Meridian DSP9 active speakers. I have a Network Acoustics M-Link filter at their input.
The older I get the more I would hope to simplify stuff. Never seems to happen with any of the technology. Guess life’s just about chasing rainbows…
Hi @Michael_Sparrow,
Welcome! I’m so glad you’re interested. Your background as a retired software engineer who has already set up multiple Ropieee endpoints makes you the perfect person for this project.
My apologies that the “how-to” links got buried in the discussion!
The main guide is a 66-page document I published on GitHub. Please don’t let the page count intimidate you; as a software engineer, you’ll find it’s very straightforward. I gave a high-level summary (which I called a “three coffee project”) back in post #8, but here is the direct link to the full instructions:
- Full Guide: Building a Dedicated Diretta Link with AudioLinux on Raspberry Pi
- High-Level Summary: Post #8 in this thread
A Note on Your P.S.
I really want to respond to your P.S. because I think we might be kindred spirits.
This is what makes you the ideal person to try this.
This project is not another “magic cable” or “esoteric power supply” tweak. It’s a fundamental architectural change to how Roon’s network data is processed and delivered to your DAC.
Like you, I was a skeptic. I started this thread by saying I used to believe the DAC was at least 90% of the sound. This project is the “bang-for-the-buck experiment” that finally “eliminated doubt” for me and proved that the transport was the real bottleneck.
Unlike those other tweaks, this is not a subtle change. It’s a “quantum jump in sound quality” (as I called it) that is very easily discernible.
Given your background and your “eliminating doubt” philosophy, I am genuinely excited to hear what you think if you decide to build it. Please take a look at the guide, and feel free to ask any questions you have!
Hi David, while on holiday for a few weeks, I keep a keen eye on this thread. One small suggestion: for future reference, maybe it’s helpful if you edit your very first post here, and add these links there.
Thanks for your enthusiasm!
Thanks so much Great to get so clear information. I will start soon and when done inform about results
Best Regards
Javier
Thank you @David_Snyder for your quick and comprehensive reply. (And yes, I missed the link – or, by the time I read the whole thread, forgot.)
As we head into winter here in Michigan, I typically have more “indoor time” to work on “eliminating doubt”. I already have multiple “surplus” R Pi’s so my inve$tment will be reduced. Looking forward to working on this project to keep my mind sharp – and perhaps ultimately gaining some sound quality improvement as a byproduct.
(BTW: I think I first heard the audiophile’s “eliminating doubt” theory from a YouTube video from either John Darko or The British Audiophile. I don’t want to imply I was the originator.)
Excellent suggestion. Not sure why I didn’t think of that. Done! Thanks. ![]()
Hi @Michael_Sparrow,
Just had one more quick thought, in case it saves you a headache…
When you check your “surplus” RPi’s, make sure they are Raspberry Pi 4 models (or newer).
The real-time kernel that AudioLinux uses (which is key to its performance) trades efficiency for timing accuracy and requires extra processing power. Unfortunately, RPi 3 models or older won’t work for this specific project.
Just a quick heads-up! Happy tinkering.
- Speakers first. 50%
- Then source (including quality of master and DAC and preamp/poweramp). 25%
- Then Room 25%
Hi @Rhythmatist,
Thanks for posting. That’s a perfect summary of the traditional audiophile wisdom, and it’s a set of ratios I agreed with for most of my life.
The entire premise of this “Tinkering” thread is that my experience with this <$300 DIY transport project profoundly shook that belief.
I’m not convinced the 50/25/25 model is wrong, but I am now convinced that the “transport” component—which is just one part of your 25% “source” slice—has a far larger impact on the final sound than I ever imagined.
It’s a head-scratcher, for sure! That’s why we’re tinkering.
But it’s all wrong. DAC never was 90% of sound. Nowadays all chips sound +/- the same. Amplifier is 90% of sound. And of course in pair with headphones or speakers. And transport irrelevant. It’s digital. It’s same everywhere. So many efforts for basically nothing.
Hi @Sergei_Sokolov,
You’re responding very literally to the title of my thread!
As I explained in my first post, the “90%” wasn’t a scientific ratio; it was my old belief about the ratio of influence (90% DAC vs. 10% Transport). The entire point of this thread is that a sub-$300 DIY project has altered my understanding of how much influence the transport has over the sound quality of a digital source.
As I also said in that first post, I respect the “bits are bits” philosophy, but this thread isn’t for debating it. It’s a “Tinkering” space for those of us who enjoy sharing our subjective experiences, “real or imagined.”
Your view that the transport is “irrelevant” is a common one, but it’s not the topic we’re exploring here. If you’d like to try the project, I’m happy to answer any questions!
I still think “garbage in garbage out” given a quality level as input the matter is not deteriorate it, you can not improve, yes you can create a nice type of sound according your taste But it is a different think than hifi
As well roon acoustics is very important.
The most productive and efficient investment is a high quality recording
The less efficient investment are speakers
Anothef extremely destroyer of quality unless you put good enough money and significant one are cables