New (was back in 2021) iFi Zen Stream device

Sorry for the delayed post, but I just caught this. Roon detected 768/512 for my DAC when it was in bridge mode too, but if I tried to upsample to DSD 512 in Roon the songs would just cycle through one after the other indicating incompatibility with the endpoint capabilities. It still does the same with the new firmware.

I’m pretty sure that with a bridge, Roon will always detect the capabilities of the attached DAC, but if there are other devices upstream that cannot handle that throughout Roon won’t play it. And although the device is now Roon Ready and is being detected by Roon as such, the bridge is still active in the audio settings—from what I understand from iFi the bridge is baked into the ZS hardware (either on a permanent BIOS, CMOS or the firmware architecture itself).

I think this bridge configuration is highly unique to how iFi navigated their Roon functionality with this prototype streamer, and may be without precedent. Typical Roon Ready devices don’t show a bridge in the signal path—they’re just endpoints unto themselves. Can’t say for sure, but given iFi’s history, something tells me that before they go any further trying to retrofit consumer demands into this iteration of the ZS they’ll just release a v2 or Signature version that is Roon Ready out of the gate, possibly with higher throughput, and jettison the bridge altogether.

What does it mean for this device to be Roon Ready? How does that make it different from a Pi running RoPieee? Is there additional functionality?

Bill there are a number of requirements around visibility of of Signal path and how it works that apparently caused Hegel a lot of issues and maybe Primare as well. I for one never expected iFi to complete this work, but I am glad they did as it will give a small number of people the confidence to try these, especially in Roon only mode.

In terms of how it works and function, there is not a lot of difference in functionality that I see between Roon Bridge and Roon Ready. You now only get one device to select as opposed to having separate entries for USB and SPDIF, so that might be considered a slight loss of functionality though there is a way to swap that.

1 Like

Thanks, Michael. I’m a little curious Roon went ahead with this, though it makes sense from iFi’s point of view. New marketing claim, and they can ship the device with Roon Bridge already loaded.

Yes and that’s exactly what they did for the last year.
iFI always said they were aiming for RR, but they said a lot of other stuff as well.
I think most of us user’s were shocked to see it finally released.

I can actually see the sense for Roon as it gives them a high quality low cost streamer that is available from the likes of Amazon that works really well and is not a DIY Pi based solution (I say this as someone with 5 Pi’s plugged in and running in the house)
IFI have made a lot of mis step’s with the marketing of this Stream, but it’s a great little device


Agreed, it’s terrific that they kept their promises. Marantz claimed that their SACD 30N (also a streamer, $3000 price tag) would eventually be upgraded to Roon Ready in a future firmware release, and that was almost two years ago…:flushed:

1 Like

So they are right on schedule then.


Well, the ZS was released on May 20 of 2021–at least they beat the one-year mark :wink:

1 Like

Compared to some that is quite an achievement.
It seemed like an eternity but really only 10 months :grin:

1 Like

Yes, a difficult ten months for many, but given how the Zen Stream now seamlessly integrates with Roon, iFi’s $399 is the only honest price point I’ve seen anywhere in this market. it amazes me that folks are willing to shell out $3,000 for a streamer. What NAIM, Cambridge and Cocktail Audio charge for their streamer-only options is appalling. I get it if they have internal DACs, DSP options or other factors that could affect the sound signature, but streamer-only components lie strictly in the digital domain and have zero impact on the end result, unless they have very poor implementations that could introduce jitter into the signal path. And the ZS is designed explicitly to produce an unadulterated digital stream.

So, for anyone considering the ZS who visits this thread, disregard whatever any other company promotes in their marketing portfolios because unless you need a streamer/DAC combo, all well-made streamer-only components have identical sound quality. You’re better served to invest in a well-measuring DAC and set aside the majority of your cash for your amp and headphones. What you pay extra for with expensive streamers are fancy hardware GUIs, elaborate aesthetics or those built-in DACs, many of which are proprietary and have measurements that don’t justify their price.

I get that some folks like the aesthetics and cache of “premium audiophile” brands, or may desire one of those gorgeously hi-res hardware GUIs with a plethora of menu options. But if you use Roon they’re largely superfluous. Which is why I think that the ZS now offers the ideal Roon endpoint—if iFi manages to expand its capabilities to PCM 768/DSD 512, I think it’ll qualify for best in class at any price point.


Fair points Steven, personally I’m not sure it needs to go to 768 PCM or DSD512, but maybe if it did some people would be more interested in including it int their thinking.
As long as it works well and is reliable in doing this why not. I will be interested if this is possible or if this comes in a Stream 2 option, or maybe a plus type option that doubles the price but is better than higher end streamers.
Hopefully iFi have learned enough about this market now that it would be easier for them to develop something along the lines of a Zen Stream DAC+ :grin:


I agree, I’m not too troubled by its sample rate “limitations”. I’d like to be able to upsample to 512 in Roon, but I think it’s mainly because it exists as an option that I can’t access. Whether it would offer any audible benefits I can’t say, but I doubt it. It would mainly be to reduce quantization noise to the fullest possible extent, but DSD 256 already moves it well-beyond 20,000 hz. It’s just FOMO.

1 Like

I’ve talked about it before, including here:

iFi has incorporated a desire to make the ZenStream an endpoint NAA and the goal of HQplayer is to do upsampling. Today, many dacs reach DSD512 and PCM 768. This is the case with the Gryphon (for iFi). Some even exceed these limits.
RPI3 and 4 based streamers… at least I can attest that it works on an Allo USBridge (CM3) and on an RPI4. I was even able to switch to PCM 1.5 and DSD 1024.
It’s a bit frustrating for the ZS especially if your dac goes through the dop (just DSD128).
There is no need to sell a more expensive product unless you have to change the motherboard.
For the moment iFi has not confirmed that an evolution of the kernel would allow to unlock these limitations. Does the hardware and the drivers developed allow it ?
But iFi has said that they hear the request without answering if it is theoretically possible.
For RPI, successive kernels have allowed to unlock these thresholds.
After the RR validation, the move to Volumio 3 is perhaps a step forward.
For the moment iFi does not say if the hardware choices limit the development and if the ZenStream will be able to do better. Which is already excellent considering the price. We have to wait and listen to the music.


Yes I have a Pi 4a that does DSD512, but it does not sound as good as the stream at DSD256. Sadly the old kernel used on the rock board that is used in the Stream is way behind current Linux kernel used in the Pi and it has not been updated in some time.

That’s why I think it will require a new board to hit these features as it is unlikely to be supported in the Rock board

It would be great for them to eke more performance out of the board, but I am not sure how far they can take it.

Well said, just curious how you foresee Volumio 3 offering any tangible, or more importantly audible improvements? I think the ZS already achieves the best one can attain from a sonic perspective, in that it performs as a purportedly jitter-proof transport (I don’t have any measurements to confirm this, but RPIs already achieve benchmark standards for digital transport and iFi has included some tech that allegedly buffers any theoretical errors).

I can see where Volumio 3 would benefit folks using DIY kits (and might improve the experience for those unfortunates availing themselves of the ZS’s native GUI) but for Roon users what do you think we can expect to see improved with the upgrade?

1 Like

Personally, I see the move to Volumio 3 as a modernization of the OS that can improve the SQ in a very small way and a move to a much newer NAA version.
The change would be mostly interesting for the interface, which is not my need, to provide a better user experience.
The level of hardware quality for the ZenStream is very high but having tested an RPI4 or Allo USBridge with a quality power supply and also an iFi Silencer (USB), the OS optimization can be heard (latency ?). The switch to GentooPlayer or piCorePlayer is quite surprising (especially piCorePlayer + NAA = the image is only 148mo ! ).
I’ve been using Zenstream for 2 months and found it better when I installed it, but I haven’t done a new test since.
My ZenStream is only allocated to decoding HQPlayer powered by Roon.
Almost all unnecessary services are switched off (7 tasks including ssh and htop). It runs at 600Mhz (cpufreq) powered by 7.5v (MeanWell medical grade + iFi DC filter).
I turn it on and forget about it… but frustrated that I can’t do DSD256 (dop limitation of my RME until my Pegasus comes back).

1 Like

Again, how will it affect that sound quality in any way? How can your RPI sound any different than the ZS? Do you have measurements to support this?

There is no DAC in these devices; they exist entirely in the digital realm—meaning prior to the audible, musical realm. They are pure digital transports. If they were not sending an accurate signal into the DAC, that would manifest as stutters, glitches, drop outs—not changes in the sound signature. Given that they are network streamers there are protocols built into that method of communication to resolve errors or packet losses—they get rejected and requeried.

Where I’m going with this is the science—all pure network streamers sound identical to each other, until the signal they transmit is sent to a DAC. What you are suggesting is impossible, no offense or hard feelings intended…

1 Like

I’m not offended, don’t worry otherwise it’s no longer possible to discuss. :wink:
I agree with you on the theory but you make a difference with ZenStream which limits jitter.
Otherwise, an RPI4 is sufficient and much cheaper. I don’t necessarily have a theoretical explanation and it is difficult to get measurements. They do exist, such as latency as a function of kernel, but then it’s empirical to associate a measurement with an audito result.
We can recognize that beyond the interface, most users of RPI-based streamers prefer such or such distributions. I regret that the ZenSream does not have a compute 4 with an iFi audio motherboard.
We could really have an open environment !


What jitter? Roon endpoints buffer the data.

Ping and jitter are important with live streaming.