Roon in 2025, 10 years on: Sound quality, thoughts, and suggestions. And an increasingly AI-generated Internet.

Notice it has been "10 Years" for Roon as per the top left.
Roon 1.0 was released May 2015.

Obviously, it's impossible to get on all the discussion forum about the myriad subtopics in this audiophile hobby. However, I do get newsletters in my inbox and I see frequently asked questions that inspire many of the topics addressed here.

If we've come across each other on the audio forums (places like Steve Hoffman's site, or Audiophile Style), you'll notice that I often will respond with links to articles on this blog. I've posted more than 600 long-form articles already, the vast majority digging pretty deep into the audio topic. One of my intentions over the years is for this to be a repository of information, articles I can use as "landmarks" when I engage in discussions with fellow audiophiles; how I understand the hobby including big-picture philosophy, psychology, the debates. Contained within these pages are the thoughts, character, understandings of "Archimago the anonymous audiophile from the early 21st Century" with contributions from guest posters. Familiarity with these articles I think will provide a good starting point for "rational audiophiles", without necessarily going too deeply into technical details.

Apart from discussing individual products, broadly, I think we've covered many of the usual issues raised in audio circles already, much of it analyzed with data to seek out the facts, rather than mere opinions. Feel free on the main page of this blog to use the "Search This Blog" widget as I suspect you'll find articles addressing many questions.

So, coming back to a frequently asked question, for this post, let's think about Roon software in 2025 (currently version 2.51 build 1534). In a recent Roon forum newsletter, I saw this question posed by Eric_Pell:

"I am wondering for years why the most expensive app in the world of hifi and high end only has a sound quality which is only ok. If you compare Roon with JPLAY i think JPLAY has only a few people to develope (sic) software and improve sound quality. But here the difference. What is Roon doing to improve sound quality and do they compare with JPLAY or not?"

If you click on the link, you'll see many thoughtful responses already which I agree with.

Indeed, Eric is right, Roon is not inexpensive. The current lifetime price is US$830, US$15/month, or US$150/year ($12.49/month). These prices have "inflated" over the years. When I started using Roon 1.6 back in 2019, the lifetime price was US$500 (or US$119/year subscription). Obviously, this kind of pricing structure requires some thought about regularity of use and value. One could easily enjoy using something like JRiver (~US$70 for single OS) as UPnP server, Lyrion Music Server (previously Logitech Music Server, based on Squeezebox origins) free, or free foobar even depending on your devices for managing the music library. Furthermore, purpose-made audio computers/streamers will also likely have their own library server software installed.

Beyond the software subscription, on the hardware side, Roon sells their Nucleus computers (US$500 Intel Celeron-based Nucleus One, or the more powerful fanless US$4000 Nucleus Titan running Linux-based "ROCK" OS said to be based on the Intel NUC12WSHi7 inside). Since a server computer does not really affect sound quality coming out from your DAC, and doesn't have to be very expensive, I'm more than happy still using my home Server computer with tons of storage, powered by an Intel i7-7700K CPU across mixed 2.5/10GbE wired ethernet, for much less money than the Nucleus Titan.

Unless you put your server computer in the soundroom, there's no need for it to be silent (ideally fanless). As discussed before, I'm happy running fanless MiniPC's these days (like this, or this) with HDMI output for playback up to multichannel 5.1/7.1 audio to my Integra AV receiver over Roon.

So, going back to Eric's question/comment, including sound quality, here's a relatively comprehensive answer in 2025:

Roon is a comprehensive piece of software that provides a very good user interface front-end incorporating deep metadata (artwork, artist information, album information, file technical details) so users can find out more about their personal music library with integration of music from some online services (Tidal, Qobuz, KKBOX, Dropbox, Internet radio). It acts in a client-server fashion (ie. distributed computer audio model) and seeks to unify all the audio endpoints in your home network (computers, phones, tablets, audiophile streamers, Chromecast, Squeezebox, AV receivers, AirPlay, etc.) under one system. Since 2022, it has the ability to stream to your mobile iOS/Android device (ARC) when away from home.

Regarding sound quality, the software is capable of lossless bit-perfect transmission. I've confirmed this multiple times over the years and compared with the likes of JRiver. Lossless playback, assuming proper software settings, to a high quality hi-res DAC will be very quiet with jitter performance excellent these days (these factors are determined much more by the DAC). This is the highest fidelity/transparency one can achieve. As discussed and tested over the years, I have found no difference between lossless playback from digital sources.

Regarding Eric's question about JPLAY, I do not believe this software is capable of improving sound quality regardless of what they say. I have previously examined JPLAY (also here) and noted that it does nothing to the sound objectively nor was there an improvement in my subjective listening. If anything, I was weary about a bug I ran into with Kernel Streaming in 2013 and still not fixed by 2015 which I sincerely hope has been rectified now! As you might recall, way back in 2012, JPLAY riding on top of JRiver with false claims of sound improvements led JRiver to issue a statement that it was a "hoax". As such, I would not touch JPLAY nor recommend their hardware JCAT products (typically ethernet cards and silliness around "audiophile" ethernet switch) unless there is objective evidence of meaningful positive effect.

Roon also incorporates a reasonably thorough set of audio DSP functions which they call "MUSE Precision Audio Control" featuring such filters:

My main use for this is for room-correction DSP using the "Convolution" function (typically with FIR filters created in Acourate or Audiolense). But there are facilities for speaker setup (phase inversion, individual speaker distance/gain adjustments), headphone listening (Audeze presets, crossfeed). Roon also allows for extra headroom (avoid intersample overload) and has its own sample rate conversion system.

For those who believe Roon doesn't sound as good as another bit-perfect software player, make sure you're A/B listening at the same volume level, using the appropriate driver (like ASIO, WASAPI), have turned off DSP (including "Headroom management" and software upsampling), and in "Device Setup", check if you're using "Device volume" for volume control. To be extra sure, see if someone can help blindly switch the audio for you and check if you can still consistently hear a difference.

You can decide after installing and trying out the Roon demo if the price justifies the performance and features. All I can say is that I use it all the time as my main streaming system, and I'm glad I have the lifetime account since 2019.

Roon is not perfect. For example, once awhile I run into a situation where I'm sure I have an album but the search function doesn't find it. Sometimes I have to manually edit the metadata if the automatic system mislabels it, which can happen with "deluxe edition" albums with bonus content. On occasion, there are multiple artist entries in the database that are the same person/band but the albums are split between these entries, or it gets confused with artists of the same name (eg. did you know there's the grunge USA Nirvana and the British Nirvana). Optimizations and bug fixes are always welcome and new builds get released regularly.

It is a complex piece of software and although generally it's not hard to figure out, the configuration might be daunting for those who don't feel adequately computer savvy. This is why the company-supported Nucleus computers might be a good choice for those who can benefit from the support if the price is right, and Roon forum discussions are helpful.

If I were a strictly 2-channel audiophile with a large music library, I would be reasonably satisfied with the Roon software as it is currently (can always still use fixes of course). As far as I can tell, there have been no significant features added since 2022 and ARC. However, as an audiophile who's a bit more perfectionistic and want to progress beyond basic 2-channel stereo, here's a little suggestion list I hope Roon can consider:

1. Improve multichannel playback support, please.

For me, 2-channel digital audio is obviously mature technology; DACs have achieved all that's needed for transparency and all that remain is to strive for increased "immersion" and 3D "realism".

My hope is that Roon can expand its support for multichannel formats, common ones like EAC3 and TrueHD decoding for example would be a good start (as per this feature request, if it makes sense to you, go vote!). Already, these codecs are supported in the open-source ffmpeg. As discussed recently, there's no need to fear lossy encoding at high enough bitrate. I already have some multichannel EAC3 or lossless TrueHD streams ripped from BluRay in my music directories as .mp4/.m4a files, and this will become more common in the years ahead. Another media data container to consider would be .mkv files as we're starting to see downloads using this, like Anette Askvik's Liberty (Immersive Edition) released last month.

Even if for now, all Roon can support is 5.1/7.1 multichannel playback over RAAT, that's a good start. In time, maybe bit-streaming to allow for Atmos height channels could be an evolutionary step for supported decoder devices, or discrete software decoding to 7.1.4 would be fantastic if supported by the playback hardware.

2. An option to prioritize multichannel albums as the "primary" version by default when importing new content.

As a multichannel fan, when I import new albums, I would rather have the multichannel version be the primary version even if the imported content is lossy (like 5.1 AAC, or EAC3) - as discussed, there is no need to worry about lossy EAC3-JOC Atmos for example.

In general, I'd rather listen to a good high bitrate 5.1 AAC mix (like 512kbps), than even 2-channel 24/192 FLAC (consistent with my "Post Hi-Res" stance).

Perhaps a switch in the Settings → Library → Import page like this can be implemented:

3. Just as there's Crossfeed DSP, why not add Crosstalk Cancellation (XTC) DSP in MUSE?

Roon does not have a system to incorporate third party plug-ins like JRiver which can use VST, for example. As such, we're dependent on Roon to implement their DSP features.

If it makes sense to add crossfeed for headphone listeners (allowing some of the opposite channel to seep in, creating the illusion of sound outside the head), why not then also implement a crosstalk cancellation (XTC) option for 2-channel loudspeaker playback? This would be a great way to help improve the immersive experience for audiophiles who want to experiment. A taste of the BACCH-like effect. I can imagine a simple interface like what Home Audio Fidelity's X-talk Shaper offers would fit nicely within the Roon esthetic.

4. Add slider to directory/file explorer for faster scrolling!

This one should be obvious, and hopefully easy to improve the user experience. Sometimes there's a need to point Roon to a file, such as a cover image, and it's a real pain with large libraries to scroll down to find an album or artist since the file explorer window always starts at "A" and there could be hundreds of names to scroll down!

There's some inconsistency in Roon's UI in that there is a scroll bar such as when going into the "Folders" option from the main screen to browse the library. Please make this universal through the software.

5. Change to "bit-depth/bit-rate" nomenclature. 🧐

Which is more important? Higher bit-depth or higher samplerate?

While I would not claim that 24-bit audio is easily audible, high samplerates with frequency response beyond 22kHz would also not be audible either! However, as discussed years ago, there is something to be said about having higher bit-depth for DACs and higher precision DSP since this minimizes the noise floor (and maintains highest dynamic range).

As such, I would argue that bit-depth is more important which is why I've consistently listed that first as in "24/96" for 24-bit/96kHz, not "96/24" (even if DTS presumably wants the bigger number first like in the old "DTS 96/24" format).

BTW, the 5.1 version of Queen's "Another One Bites the Dust" from The Game DVD-A sounds excellent when upmixed with Auro-Matic!

A small matter I suppose, and one of preference. Either way, it's good to be consistent with this especially if used when tagging the music library.

There might be other items I can think up, but these are some main wish-list items or just day-to-day annoyances. As you probably know, in November 2023, Roon was bought over by Harman (subsidiary of Samsung). I don't know if this has affected the staffing in development of the software. I do hope that they continue to provide strong support obviously! It's always good to see products be progressive with new features being added especially when I think multichannel immersive audio is still an area for audiophiles to mature into. Given the price of Roon, I believe they need to be proactive to justify value, especially in the eyes of those using the subscription model.

Here's wishing the best for Roon and users over the next 10 years.

--------------------

"My grandfather's friend took this photo 70 years ago while diving in the prehistoric seas (about 150 million years back), traveling there using a time machine he had built.
The Buddha was already there with his seahorse Kanthaka. The government confiscated the time machine decades ago. Government leaks have confirmed that it's being used for all kinds of dark purposes by the deep state."*

In anticipation of an increasingly AI-generated Internet (and media in general).

Although I have used AI to create banner images and in some articles, to illustrate AI-generated texts (like here), unless specified, this blog has been the work of human minds and I expect it to remain so. 😊

Already, there have been discussions (and worries) that much of the content on the Internet is AI-generated. This is particularly true for non-English content due to parallel machine translation as examined in this paper "A Shocking Amount of the Web is Machine Translated: Insights from Multi-Way Parallelism" that came out in 2024 - I believe the headline number of "57%" was the "shocking" value to suggest that more than half of content searched online originated from such machine translations.

Social media platforms like LinkedIn and Meta have embraced AI-generated content with a study suggesting "Over 54 percent of longer English-language posts on LinkedIn are likely AI-generated" in late 2024. More recently, in early 2025, we're now seeing numbers like "74.2% of newly created web pages in April 2025 include AI-generated content" (this would apply to many of my posts on this blog). To me, the key is that at least "only" 2.5% is "pure AI" while 26% "pure human", I think this statistic is probably still healthy. Presumably, as time goes on we could quite easily see a shift to more "pure AI" articles although how far this shift happens we shall see as the field enters the realm of agentic AIs that could be instructed to complete complex tasks including creating long articles with minimal human input.

While machine translation has been with us since the early 2000's (maybe even late '90s), generative AI as in ChatGPT and image creation has only been in the last 3-4 years (ChatGPT from OpenAI publicly available only since November 30, 2022). It is amazing how quickly these tools have become incorporated into blogs and social media posts at accelerating pace! A reflection of the power that these tools can have for content production.

As with any technology, we have to balance the positive with the negative. For much of the information content we consume, I don't see AI as automatically in any way being inferior to human-curated or created materials. Just like with human writers or creators, so long as what is presented reflects intended, balanced truths and realities, what a machine produces can be just as meaningful and useful. Note that this can be more complex than just factual presentations because the subtle tone of an article and inherent biases do need to be considered. As media consumers, we will need to be increasingly more discerning to make sure we understand the world and can make judgments against incorrect or biased content. Is the educational system catering to address this? I've also seen many older folks have trouble with obviously fake news articles, and many do not actively seek information from multiple sources for confirmation before accepting the contents.

As I discussed last year, AI generated content is expanding multimodally and will continue to improve in apparent realism. Journalistic sources will also be using the technology. Watch as these tools become used to easily create content coming from sources with low journalistic standards. This is what I fear most - that low-quality content massively outnumber high-quality material because of the ease of creation. For example, low-quality subjective-only articles or the scripts for an opinion-based YouTube review will be much easier to create compared to ones that take time to plan, to test, to measure.

Examples include much of the content in magazines like The Absolute Sound or Hi-Fi+. Fluff articles like this "AudioQuest system upgrades" (Dec 2024, Hi-Fi+), or this review of the Synergistic cable (TAS, 2023) would be sooooo easy to write! Just throw in the names of the products we want to highlight, throw in basic advertising talking points, add a few pictures (perhaps from the company), some light text editing, and it'll probably be done and proof-read within an hour. In fact, the more snake-oilish the product, the easier to claim almost whatever one desires as these articles are mainly works of creative fiction that a computer can easily "imagine".

Seriously light work, and easy for those making money off this kind of thing.

Within the audiophile world, I believe we've been inundated with way too much misinformation from all kinds of sources whether online, from questionable companies, from fellow hobbyists who have accepted claims without serious consideration, and in typical magazines. I hope rational audiophiles by now have developed a healthy level of skepticism to what we read/watch/hear, with desire for more objectivity to anchor us in reality.

[On a somewhat humorous side note, I see that Danny Richie of GR Research posted this YouTube video hocking his cryo-treated speaker cables with a rope through the middle as if special tech. As discussed before, those cables are basically Kimber knock-offs. 

He also has a strange desire to create fear by disputing the robustness of digital audio, wanting us to believe in all kinds of "noise" without evidence - this is typical of many out there who believe in voodoo and need to sell products to fix an imagined issue. I get a sense that he doesn't actually understand digital playback systems nor actually has tested himself. Pretty silly stuff in 2025!

"Not flat"?! Who is he talking about? 🤔

As usual, these guys will claim that if one doesn't hear a difference with special tweaks and cables, it's because your system/ear/brain isn't discerning enough. It's like The Emperor's New Clothes suggesting that a "true" audiophile knows, and can hear these claims, and if you don't, well, I guess you're just not sophisticated enough or have the right kind of hardware/room/ears. Seriously, let's just cut the crap and show audiophiles in what way a product improves anything, grounded in physical reality. 

I also noticed that Richie doesn't understand how the placebo mechanism works, as if it is consciously controllable (which is why blinding is essential).

Maybe Danny Richie is a gifted crossover/speaker designer, that's cool. But he's acting as usual like a "UFO Believer" with nothing to show, mere words, appealing to sympathy, while posing hypocritically as if he's against "Flat-Earthers" - those who believe strange things, ignoring the science and physical evidence - he needs to take a good look at the man in the mirror holding his fat rope-filled, cryogenically-treated cable. 

Sad to see this level of poor insight. Age and experience do not imply "wisdom", if one has issues acknowledging even basic truths.

"Don't believe the hype."]

* A paraphrased AI-generated comment, in the voice of a conspiracy theorist describing the grainy, monochromatic image.

Addendum: June 8, 2025

Want to see a funny AI-generated "catalog" of audio snake oil? Check out this "Audiophile.Rocks Catalog" with some kind of "IQ Test", released in December 2023. I have no idea how serious this is, but I bet the guy doesn't mind money coming in by PayPal!

Hilarious that there's even a subjective review of such things. As is usual with these kinds of reviews, when the reviewer knows the product seems crazy, they typically will admit to be "very skeptical about these types of audio tweaks" until they're not. Inevitably shocked by what they then later subjectively, ostensibly, experienced. A typical pattern for apologists as they try to buffer the magnitude of craziness.

Of course this is just low-IQ and stupid, maybe even delusional, or perhaps sociopathic if one is less generous. The attitude that we must "try it first before dismissing it" that some audiophiles advocate is silly. The onus is on those who make and sell such questionable items to provide evidence first.

It's important as audiophiles to encourage consumers to think critically. Since audio is very much mature engineered technology, if an item does not appear to perform within scientific domains, we must not encourage others to "give it a try". This is not close-minded thinking, it's just a reflection of wisdom. A natural response to the obvious recognition that there are liars and conmen in this world out to make a buck by any means possible.

If a friend has a diagnosis of cancer, we should not be encouraging him to ignore science and put money into questionable naturopathic treatments and psychic surgery, no matter the strong subjective opinions and anecdotes within certain echo-chambers, right? If he still thinks he wants to try these things, then fine, within legal freedoms, such things are permissible.

We don't need to try everything in order to know that many ideas are worthless; potentially causing harm - moral, financial, healthwise, or in other ways.

--------------------

Now that we're in June, I suspect many of you are looking forward to vacations ahead!

To end, here are a few more pictures from Turkey and Greece from my Sony α6700 with just the 18-135mm kit lens; an excellent travel companion with great photo quality, very capable vibration reduction for challenging light situations, and highly capable of punchy, smooth 4K/60fps HFR videos:

Balloons at sunrise over Göreme, Cappadocia, in Turkey.

Blue Mosque, Istanbul.

Hagia Irene, Istanbul.

Basilica Cistern of Istanbul.

Hagia Sophia, Istanbul - a look inside.

Prayer, at Kariye Mosque / Chora Church, Istanbul.

Lots of cats in this part of the world!
Photographed on Patmos, Greece.

View from the Acropolis of Lindos, Rhodes, Greece.

Rhodes, port area.

Oia, Santorini. (Everyone visits Santorini when in the Greek Isles, right?)

Santorini evening, in the caldera.

A culinary delight. Grilled octopus. Yum.

Acropolis (and Parthenon) by night, Athens.

Cheers, wherever you might be heading, whatever you're doing this summer (for those of us in the Northern Hemisphere)! Regardless, I hope you're enjoying the music.