why expensive streamers

@mgrif104 

I apologize if my frustration came across as hostile.  

there are two separate camps on this topic lurking here on this forum.  One argues that spending a significant amount of $ on a streamer can be a waste of your precious resources, because the science, measurements and their experience tells them that beyond a basic level, there’s no difference in sound.  The other camp argues that based on their experience, there are clear differences, and therefore recommend spending upwards of $10k in some situations in order to optimize one’s system. I'm squarely in the first camp. So here’s my view:

Putting aside reclocking and oversampling, the only distinctions across streamers  have to do with levels of jitter and other sources of noise.  These are easily measurable.  Anybody who tells you that a more expensive streamer is less noisy while simultaneously claiming that measurements don’t matter is not credible.  If your DAC is good at rejecting noise (e.g., filtering, reclocking, etc.), you can assemble your own streamer for as little as $150 and it will sound great ( I’ve done this as an experiment myself).  However, if you want something less kludgy than a Raspberry Pi, and you want a streamer that measures and in some situations with some dacs may sound even better than a Pi (and measures and sounds comparable to some of the extremely high $ streamers out there), then there are a bunch of well regarded products that cost between $400 and $1k.  They include the iFi Zen, the Volumio Rivo, Holo Red, Pro-Ject Stream Box S2 Ultra, the Primare Np5, and a few others.

@jaytor 

respectfully disagree. maybe i’m not following, but not sure your objection is any different for measuring dacs.  regardless, others have done it; for example, you can measure at several different dacs with a control source (e.g., a computer) showing jitter and signal to noise ratios - at the dacs - for both the computer and the streamer, and then do same for other streamers against same control (computer).  that’s just one way, which is how paul miller (hifi news) has done it.  if you look  at his measurements, it shows that dac performance on jitter and noise dominates. that outcome is consistent with what the folks at Benchmark have said - that their dac’s performance is agnostic as to the streamer.  

noise is easy to measure, which is why I find it extraordinary when high $ streamers are advertised as lowering noise without any measurements/data to back it up.  However, there are a few places where some analysis of the differences - or lack thereof - across different streamers - is done more comprehensively.  These include hifinews (paul miller’s measurements sections of streamer reviews), archimago, goldensound, and, yes, ASR.  (For a reaction to the point from @blisshifi, see Amir’s video entitled “Is digital transmission really analog?”)

i’ve met a bunch of folks over the years who think they’re perfect….including my father-in-law, my oldest sister, the list goes on.

@nonoise 

sorry, noise in the digital domain is not mystical, it’s easily measured. now in the analog realm, I sing a very different song.  Measurements are very limiting there.  

@nonoise 

Streaming is not a new technology.  I for one have been streaming since about 2004 (starting with the original Yamaha Musiccast system).  Moreover, the fundamental job of the streamer, moving a data stream from server to dac, is rather simple.  And sorry to be a skipping record, but even most high $ streamer proponents agree it’s just about jitter and other noise.  So measure it!

Btw, I volunteer for the reassembler chamber…

@jaytor 

yes, it’s the combined system, precisely.  We don’t care what a streamer sounds like without a dac or source, because it has no sound.  So if you say it adds noise to the system - i.e., it conveys noise that manifests thru the dac that can be heard - then you have to test it in the context of the system.  There is no alternative.  And that’s what the tests I’ve seen do.  The results of those tests demonstrate that noise from a streamer can carry thru to the dac, in certain cases, but even in those, it is a question as to whether any marginal increase in noise from a streamer was audible, relative to the noise that was coming from the dac.  In most cases that I’ve seen, with any decent streamer and dac, there is virtually no additional noise that comes from the streamer.  

Candidly, I find extraordinary the reluctance on the part of so many sophisticated audiophiles to follow this basic level of logic and analysis when contemplating spending thousands of dollars on what is really just an accessory (again, putting advanced processing like dsp, reclocking, and upsampling aside).  And as a result of this reluctance, a number of firms have dived into this niche market and, imho, are fleecing their customers.  Don’t get me wrong: their stuff can be beautiful and well-engineered, but without significant sonic benefit.  The only reason anybody should spend alot of $ on these niche players is if they think the UI is worth it (again, imho).  But I’m a Roon guy, so spending another dollar for somebody else’s UI is akin to pouring that dollar down the drain.

@12many 

Not sure what your dac is, but any differences you hear may have to do with different processing.  The N150 definitely reclocks, for example.  In addition, if the n150 is upsampling, but the Node isn’t, you’ll definitely hear a difference.  May not matter to you, since you like it better, but for purposes of lessons learned, it’s important to filter out (pun intended) any differences in processing going on.  

@mgrif104 

you can’t attribute the differences to noise when objective measures of this most easy to measure quality show that’s not it.  that’s all i’ve been saying.  that’s like saying an engine puts out more horsepower when the data show us otherwise.  

@mgrif104 

I’m astonished that you can point to jitter at the same time that it’s been demonstrated that jitter is virtually nonexistent in any decent streamer.  How can it be that these manufacturers have discovered a secret in removing jitter, noise (and fairy dust?) without showing us how to measure this new, exciting branch of science?  Why don’t you ask them to prove it?  Why don’t they volunteer this information that would prove their brilliance? It’s all mystifying to me. 

@12many 

The only Bluesound product I have is their Powernode, which drives my outdoor speakers on the deck; I’m using the internal DAC cuz it’s obviously about as far as you can get from a critical listening environment.  The Aurenders I’ve heard, because they’re ubiquitous these days, are all in dealer or show conditions, and have never been in A/B situations.  My A/B experience is among my current crop of streamers, which include a Raspberry Pi, an iFi Zen, a Pro-Ject Streambox S2 Ultra, and the streamer portion of my Gold Note DS10.  My experience is consistent with what the data and theory tell me, which is that I hear no difference between a $150 streamer and a $1000 streamer, with my DACs.  

At the high end, I spent an afternoon with a friend at a dealer comparing an Innuos Pulsar ($8000) to the streaming portion of a Simaudio 280D streaming dac ($3900), with a DCS dac.  No difference.  (This experiment was for my friend’s benefit, as he wanted to get comfortable with replacing his DCS with one of the new Simaudio North products - he chose the top of the line 891 - $25k!!)

All of this is consistent with the lab measurements that paul miller at hifi news has been doing on streamers.  I’ve put together a spreadsheet (sorry, I’m a recently retired finance guy) summarizing his results for a bunch of products, including Volumio Rivo (£1k), Aurender N200 (£7k), Melco N20 (£7k), Grimm MU1 (£10k), Aurender N30SA (£24k), and an Antipodes Oladra (£25k).  He compared the SNR and jitter results from these streamers in conjunction with 3 different DACs - AQ Dragonfly, iFi Neo iDSD, and a Mytek Brooklyn (i.e., from really cheap to moderate price, but nothing expensive).  His findings?  Jitter measurements for the $200 Dragonfly ranged from 105 psec to 150 psec, with the highest number coming from the Aurender N200.  Fear not though.  Remember, a psec = 1/1000 nsec, and the general rule is that we can’t detect jitter unless it’s about 30 nsec, which is 30,000 psec!

What about the iFi and the Mytek?  Even better jitter rejection, with ranges of 9-18 psec and 5-10 psec, respectively.  And the results on SNR tests were similar.  There, the only material noise issue was with the Dragonfly.  

In summary, the science, my experience, and actual measurements tell me the same thing.  A competent streamer should not be a source of audible noise; I am convinced that any differences we hear are either due to differences in signal processing (e.g., reclocking, up sampling, etc.), our DACs, or, yes, confirmation bias.  I also believe that most of the niche companies in the streamer space like to bundle their products - server, streamer, dac - because then it’s awfully difficult to pinpoint where that great sound you’re hearing is coming from.

I hope this is helpful.  I’d be happy to give a better, more comprehensive summary of the hifi news methodology, but in the interest of brevity (lol!!)….

@mclinnguy 

hah, nice try! but let’s stay on topic.

@mgrif104 

I agree to disagree.  And I do believe that you are earnest in your beliefs.  I guess my frustration is this feeling that we live in a post-truth world.  I show you factual evidence of virtually no jitter across a wide range of streamers, but because you want to believe in the jitter fantasy, you insist there is audible jitter, offering. no evidence other than that you hear isomething.  A different approach would be: “ok, i guess you’ve shown me it’s probably not jitter.  But I hear something. Maybe I need to work out more carefully what it is.”. Regardless, peace be with you.

@cleeds 

those are called rhetorical questions. 

. 

@cleeds 

Geez, the second comment was an apology for the first.  Doesn’t anyone even accept apologies anymore? 

@mitch2 

To me, the issue is not whether there’s no jitter, it’s whether there’s a substantial difference in jitter across streamers, how much jitter we’re talking about, and the extent to which different DACs reject jitter from streamers.

As I’ve previously indicated, I have found the “lab report” section of HiFi News streamer reviews incredibly insightful.  I’ve looked at the results from 10 different streamer reviews over about 4 years; in 5 of them they use a usb connection to a computer as their control scenario vs. the streamer, and then they use up to 3 different DACs with both the computer and the streamer.  All 10 of the reviews I’ve pulled feature the Mytek Brooklyn dac, 6 include the iFi Neo iDSD, and 4 include the AQ Dragonfly.  Here are some takeaways:

1) The Mytek has low jitter levels itself, and also is very effective at rejecting any streamer-induced jitter;

2) The iFi is a low jitter DAC, but is not nearly as effective at rejecting streamer-induced jitter; and 

3) The amount of jitter produced by all 10 of the streamers, priced from £1k to £25k, is very low, and doesn’t vary materially by price.

How did I arrive  at these conclusions?  Here are several snippets of the data:

- the single highest jitter number, 550 psec, was with the computer/iFi combo, substantially more than the computer/Dragonfly number (300 psec).

- but, when paired with a decent streamer (e.g., the Volumio Riva or Aurender N200), the iFi’s jitter amount was substantially less than the Dragonfly (18 & 9 psec v. 150 & 135 psec, respectively.

- the Mytek’s jitter levels were never above 10 psec, whether using the computer or one of the 10 streamers.  

So while a decent DAC, in this case the iFi, may not be engineered to reject jitter from a streamer, you don’t need to spend much to have a very low-jitter streamer.  That’s why a Raspberry Pi can sound so good - many DACs, even moderately priced, do a good job of rejecting jitter from a streamer.  But to be safe, you can invest a little more than $150 and get a very low-jitter streamer.  

@mclinnguy 

my main point in this whole thread has been that it is very common for many of us to cite “lower noise” to explain differences in sound that we can’t otherwise explain.  but there’s this weird cognitive dissonance among many when confronted with actual measurements of noise; they cite marketing materials promising “decreased jitter and noise” and their experience of improved sound, rather than accepting actual evidence and then trying to figure out more plausible explanations for what they’re hearing.

by way of example, on another thread a while back, the op was insisting that a new streamer’s noise level had to explain the improvement in sound he was experiencing.  when pushed to reveal what his actual A/B experience was, it turned out he preferred a non-oversampling algorithm that deferred to the native sample rate of his new set-up vs. the always oversampled algorithm of his old setup.  in other words, he was definitely hearing a difference, he just hadn’t done the right analysis to figure out why.