SOLVED: Is USB audio quality resolved with the Raspberry Pi 4? i.e. No need for SPDIF Output HAT's

I have recently traded in my Sonos equipment for Raspberry Pi 4’s. 2 with Allo Digione/Signature HAT’s (1 into Schiit Modius DAC, Vidar amp.), 2 with HiFiBerry Amp2 HAT’s; all running Ropieee XL.

I understand that the audio out of the USB of an RPi 3 was particularly poor due to the shared ethernet and USB bus and thus the Allo SPDIF HATs (and others) do a fantastic job of mitigating those problems.

With the Raspberry Pi 4, did the audio quality out of the USB improve to no longer necessitate a SPDIF HAT?

Could I run into my Modius (with Unison USB) or Modi or any other USB signal ‘cleaner’ and get the same or better audio quality?

Short answer: Yes, Pi4 USB is fine

Longer answer: There are people that complain about electrical noise leakage between devices and USB will therefor always be garbage. Personally, I do not subscribe to this malarkey. Any such noise is so far down in the noise floor as to be inaudible.


Can’t you run the experiment and report the results? Ideally, with a careful listener who does not know which input is which? I’d think you could run an USB cable from the Pi 4 with the Digione into the USB socket on the Modius, while leaving the S/PDIF cable.

My systems are different, but my faithful tester (my musical wife) heard a difference when I upgraded my Yggdrasil to Unison from Gen 5 USB. The USB source was the original Allo USBridge with a Sparky SBC, powered by an Sbooster LPS. This source with Unison also beat AES from an Auralic Aries Femto streamer.

In general, USB on earlier Schiit DACs was well behind S/PDIF or AES. Unison is a solid upgrade that brings quality to a competitive region with S/PDIF coax or AES (at least for Yggdrasil, but reports suggest that’s true on other Schiit DACs too). Currently I run my Yggdrasil with a Pi 4 + Pi2AES, as AES from the Pi2AES gives a small improvement over the above USB source, but we are talking slight and possibly totally personal taste dependent changes (again, my wife was the tester).

1 Like

Thank you, appreciate the feedback. I can of course but am hoping to garner all other opinions in conjunction with my own. My modius is in the mail and I also don’t yet have another “plain” Pi to ABX test.

Archimago hasn’t been able to measure significant-audible differences. Your USB should be fine.


Yeah, I wouldn’t worry about it. As you say, the issue with the Pi 3 has been resolved with the Pi 4, as the Ethernet and the USB are on separate channels now. And the slight increase in speed of the Pi 4 also slightly improves response time. Should work just fine.

The Modius looks like a good pick.


I’m finding that the Pi4 (RopieeeXL) fed into decent dacs (Schiit multibit, RME ADI-2 DAC etc) is just as good as going from dedicated network bridges (eg Allo USBridge signature or Pro-Ject Audio Streambox S2 Ultra). On less proficient dacs, I still find that an iFi iPurifier 3 helps, but I’ve been really pleased with the Pi4 solution - simple, robust and effective. I think the unison usb interface will make anything more than a simple Pi4 solution a waste of money.


The news is slowly getting out, before the DAC, it’s just data. A Raspberry Pi bridge works great!


Data encoded into noisy electrical signals that leak into the analog side of some less well constructed DACs. Some DACs (like my Soekris dac1541, or the Unison-upgraded Yggdrasil) reject that electrical noise enough that their USB quality is roughly on a par with S/PDIF coax. Others, not so well.

Rather than asserting broad generalizations that are both technically and empirically wrong, I’d rather we stick with specific gear we have actually evaluated?


It’s not malarkey. Maybe you can’t hear the difference with your setup. But the Pi4 still has a noisy USB implementation even if it is not sharing circuitry with the Ethernet port. Noise messes with soundstage, instrument and voice separation, and a variety of other things.

You don’t have to hear the noise to hear the effects…

1 Like

Interesting. Evidence? Measurements?


You must be kidding. Consumer-grade digital circuitry has a lot more switching noise (power and data) than that.

Because the Pi4 is a noisy small general purpose computer. It was not designed to be ultra quiet. So it’s USB implementation is polluted by that noise.

I hear you saying this, but what makes you think it’s noisy? What noise? Evidence? Measurements?


Bill, do you have evidence for every comment, opinion, and fact you state? No, you don’t…

Look, I don’t have to measure every computer to know that a general purpose computer with a standard USB implementation is going to have a bunch of noise on the USB bus. The Pi4 was not designed with low noise in mind. It was designed to fit a form factor. Low noise is really only critical in audio applications. You have to design with low noise in mind to get low noise. For general computing it is a non-issue and the Raspberry Pi family are general computing devices…just very small devices.

We’re not talking about me. We’re talking about what you said, that “the Pi4 is a noisy small general purpose computer.” I don’t know why you say that. I don’t know whether to believe that or not. I doubt it, but I’d like to keep an open mind. Do you have any evidence to back that up?

Perhaps you believe that to be the case, but it doesn’t appear to actually be the case. Take a look at these measurements.


If you want to believe that the RPI family and all computers don’t make any noise that effects audio, you go right ahead…

Archimago is not measuring the noise on the USB bus in any case. He is measuring what a particular DAC puts out using various inputs. I have no way of knowing his test methods or tools. He and ASIR on the same page which causes me concern. He is in the “bits are bits” camp which I am certainly not…

I recently did an A/B test between the USB from a NUC83i running ROCK and a Rpi4 with Ropieee. Now, one caveat, the NUC had the regular power supply and the Rpi had the iFi low noise 5volt/2.5amp power supply.
I was very subjective, as I doubted there would be any difference. Plus, I would have preferred it if there was no difference as using the USB from the NUC meant for a simpler setup.

While very close, the soundstage from the Rpi4 seemed slightly fuller with more depth. I had to go back and forth to hear the difference, but it was there. So, I think using the USB from a Rpi is perfectly fine.


Yep. And demonstrating no discernable additional noise.

He’s pretty clear and detailed in his description of test methods and tools.

I’m sorry to hear that this causes you concern. Not sure just what there is to be concerned over, that two folks you don’t know and apparently don’t want to know, apparently agree with each other on certain things.

It’s the camp called “reality”.


It depends on both the source computer and the DAC. Sources vary a lot on noise level (some older laptop USB ports are especially bad), and DACs vary a lot on their USB noise rejection ability. Measurements or (preferably blind) listening tests on a particular source-DAC pair are only applicable to that pair. I have no doubt that Archimago reported accurately on the particular hardware he studied. But I could name several otherwise very good DACs with noise-sensitive USB receivers.