Digital Sources computer versus Streaming

[Purite Audio]

Just following on from an idea proposed on another thread, which is technically the best solution? I suspect the devil is in the detail/implementation but I would be interested to hear a technical explanation . Keith.

 

[Purite Audio]

Just to expand, proponents of streamers are often ( all the time ) crowing about the supposed low jitter that is 'inherent' with streamers compared to the noisy computer environment, is there more to streamers than an Ethernet spdif interface? Keith.

 

[cfmsp]

 

Keith,

 

As far as I know, there's still a DAC and a clock involved, the quality of which will be the determining factor in the jitter, same as for Firewire, and Async USB.

 

 

I don't pretend to understand this category completely, but have known for awhile that there's potential for some of these implementations to rival Firewire/Async USB.

 

it seems there are implementations called 'network players', which I think you're calling streamers, and then there's another group which are quite similar - the two computer plus dac solution, which John Swenson has described on AA, which Andrew described recently here on CA, and which I believe is being implemented by Matan Arazi (sp?) in the system that was demoed at the CA Symposium.

 

these seem to be very similar in concept - they isolate the playback computer from the DAC, by inserting a device in between which has the bare minimum of processing capability required to receive and pass along the data.

 

clay

 

 

 

 

 

[firefly0071]

Stating the obvious, computers can crash particularly if it is used as a music server and general tasks like accessing the internet, etc.

 

 

 

[cfmsp]

 

It's pretty well known that music playback on Windows/PCs is susceptible (and more so than Mac OS X) to interference from other tasks running, but I've not heard much about systems crashing.

 

I've personally not experienced a 'system crash' since 10.3.9.

 

YMMV, and apparently did.

 

clay

 

 

[Audio_ELF]

As I see it, "computer audio" can be implemented in 3 distinct methods. The common factor in all is that instead of the file being read from a CD (or DVD) it is always read from a hard drive (in some location). Each have many pros and cons - I've just listed some of them below.

 

1. Pure computer based system - this is the typical Mac or PC system with a computer attached directly to a DAC; using FireWire, USB or direct SPDIF / AES connection.

Pros - Very versatile, everything can be changed and designed around the users requirements - you don't like Windows, use a Mac OSX or Linux. Choice of software suiting different people's requirements. Infinitely upgradable. Supports all file formats either natively or easy (without loss of quality with loss less formats) transcoding. Easy upgrade of DAC. Low cost entry (though can spend a lot). Can easily be expanded to multi-room using low cost streamers such as Squeezebox or Apple TV.

Cons - Jitter can be issue transferring data to external DAC. Computers can be tricky to set up. Cheep computers tend to be noisy for listening room. Control requires other devices - universal remotes limited (need monitor), devices such as iTouch as remote are expensive. Many audiophiles just don't want a computer in their listening room (yes prejudiced I know - but it's true)

2. HiFi component - basically a computer system "in a (pretty) box" using components chosen for their physical (and electronic) noise level, often combined with a high quality DAC. Typical of this system would be the Naim HDX and Macintosh MS750.

Pros - Bespoke software can be easier to use with remote. Limited options makes system easier to support. Components chosen for noise so (should) be quiet. High quality audio out of the box with Naim HDX type device.

Cons - Limited support for file formats. Expensive. Limited potential for upgrade. Adding external DAC can introduce jitter. Not all built for sound quality.

3. Network Streamer - a device which talks to a computer server (or NAS) to retrieve the music files to play, these are then decoded in the box and fed to a high quality DAC. Typical device that would fall into this category would be Linn DS devices and Logitech Transporter (also the other Squeezebox but the DAC is of lower quality).

Pros - No hard disk in listening room. High quality possible. Potential to remove jittery interface to DAC.

Cons - Expensive. Limited support for file formats. Control systems not always very user friendly compared with (e.g.) iTunes. Limited range of devices in the HiFi market.

 

I'm sure that other people can think of many more pros and cons for each type of device. There are also lots of devices that can also cross the boundaries. One thing I think varies between people is how much potential sound quality are you willing to give up, for the sake of convenience features. i.e. people argue that iTunes can easily be beaten for sound quality if you use (for example) Wave Editor as your playback software, however there is little argument that iTunes using an iTouch for remote control is (one of) the most intuitive, user friendly ways to listen to music. Yes there is still the potential to improve on audio quality from a computer - devices such as Matan's server which splits it into two seam to be a possible way to go - high power computer decoding and controlling, then passing the data to a second box which converts it to a signal for a DAC, even better if the DAC is built into that second box (IMO).

 

Reading a forum such as Computer Audiophile you get the view that every high end audiophile immediately is willing to go the computer route, but if you read a less focused forum, there are many more posts from people who are sticking (for now at least) to medium like CDs. One good example is High End Naim Users. Many have (for example) a system with Naims CDX2 with mid-high end, even though there is the option (sticking with a all Naim system) of the Naim HDX (which is as good quality as the CDX2) or soon a computer with a Naim DAC, they will stick rigidly to CDs. There is also a large crowd of audiophiles who are sticking with vinyl. As for computer vs. HiFi device vs. network streamers - I'm not sure there'll ever be one true way. Just as there is continual argument of SPDIF vs TOSLink vs AES vs FireWire vs USB vs i2s vs all-on-one; then different people will keep using new methods to try to wring the best out of their systems. Some people will be happy buying a MacBook and Wavelength DAC and listening to the music; others won't rest till their personally written software and designed DAC is perfect (an aim which is beyond the reach of most); and still more will continually fret if using a FireFace 400 rather than their TC Konnect 8 will reduce jitter by 5 or 10%, or if they should replace the MacMini with a custom built silent PC so they can install a Lynx AES16 PCI card.

 

Eloise.

 

[BobH]

My take on this is implementation. It all comes down to what the box manufacturer regarded as their primary connection preference, because that's where the effort and the technical expertise will have been most applied. I admit to being a bit of an evangelist as far as my Squeezy is concerned because it is a very good low cost solution. Not the be all and end all, to be sure, but excellent for the outlay.

 

I'm able to test between :

 

Computer direct to dac via usb

Computer via wired network to squeezy to dac

Computer via wireless network to squeezy to dac

 

And the order in which I have listed them is the order in which I prefer them. So, for me, I get a better sound sending the audio via the most convoluted route - very unintuitive!

 

There are a lot of variables in there, such as the quality of the usb implementations, the quality of the playback software, (server vs iTunes, MM, Foobar etc), and such like. However, the important thing for me is that the Dac's major interface is coax spdif - the usb being an order option - and the Squeezy expects to be connected via a wireless network. (SlimDevices actually go so far as to say, in the manual, that they expect you to get the best sound quality via a wireless network - I don't disagree with them!)

 

Technically I suspect that the arguments for and against the different interfaces could well rival the average jitter thread for length and disagreement but in the end I think, in the real world, it comes down to which one you would like to use and then finding a manufacturer whose kit has been designed primarily to work with that interface. That's my take on it, anyway! Not very technical but then I'm all idiot, with little or no savant!

 

(apologies to i_s - only joking!)

 

[Audio_ELF]

Just out of curiosity and to clarify ... you find the Squeezebox via wireless and then onto the DAC the best sounding solution?

 

Also what DAC is it you use? And have you ever tried any FireWire interface between computer and DAC?

 

Eloise

 

[BobH]

Just out of curiosity and to clarify ... you find the Squeezebox via wireless and then onto the DAC the best sounding solution?

 

Yes! Actually, much, much better. The set-up is : Samsung NC10 Netbook, (running the SlimServer software), with USB attached storage / Linksys(Cisco) WRT54G2 wireless / Squeezebox Duet - output via coax spdif / Promitheus Dac (24/96 Nos)

 

My Promitheus has the choice of either spdif or usb and the Squeezebox can be set up either wired or wireless. I've tried all combinations and the above set up works best by quite a margin. All the usual hyperbole applies but it mainly boiled down to clarity and dynamics. Whether that is a function of the software, server vs media player, usb vs network, etc etc I've no idea - I ran out of inclination once I'd got the sound I wanted!

 

No, I've not had the opportunity to try either a well implemented USB dac (the Promitheus usb input is dreadful!) or a Firewire dac. I know the usb on the Promitheus is dreadful because if I swap it out for my Tascam U122 usb soundcard the Tascam is tons better!!

 

I have to say that when and if I get round to it, the two implementations I shall be most interested in trying is Firewire and AES. I find it interesting that most of our UK manufacturers have gone down the network route. That's not to say that sound quality was the absolute reason they've done it, but it is noteworthy none the less. :~)

 

[Audio_ELF]

Bob's last comment made me think about how some of the high end audio / HiFi manufacturers have approached computer audio in the last half dozen or so years ... thought this might be a useful summary (in no particular order) of companies that I can think of at the mid-high end of HiFi. I'm not here thinking of specialist companies such as Wavelength or Weiss, more general audio companies and what "computer audio" products they produce.

 

Naim

HDX - "custom computer" (provides multi-room via NaimNet / NetStream devices) controled via front pannel touch screen or Flash capable browser.

NaimUnity - all in one including UPnP streamer.

SuperNait - integrated amplifier with inbuilt DAC

Naim DAC - first DAC device - SPDIF input, USB connection for iPod though no direct computer input (details tbc).

Arcam

MS250 - custom computer provided 4 HQ analogue outputs plus zone 1 SPDIF, control via basic remote + LCD or monitor. Product seams to have been dropped after only a few months.

No standalone DAC, CD player with SPDIF in or stereo amplifier with built in DAC

Linn

4 levels of DS streamer utilising UPnP controlled via Windows interface or generic UPnP control, now available an iPod interface.

Akurate Kontrol pre-amp has inbuilt DAC (also DD decoding) - no USB / FireWire.

Previous produced high end HDD based multi-room system

Cyrus

Cyrus 8XPd and Pre XPd integrated and pre-amp with inbuilt DAC and USB connection.

Cyrus DAC X(P) - standalone DAC (optionally with pre-amp) - no USB / FireWire.

In past produced a version of the Imerge server with improved Zone 1 output.

Musical Fidelity

Various DACs with USB input including V-DAC at budget end and DAC-X v8 higher up

A1008 integrated amp with DAC inc. USB

Bryston

BDA-1 DAC has USB input

Both integrated and pre-amps available with built in DAC but no USB / FireWire

Cambridge Audio

Azur 640H custom computer music server.

DAC Magic has USB connection.

McIntosh

High end custom computer music server.

Unsure of any DACs

Audio Research

DAC7 has USB connection

Wadia

Adding USB to new "decoding computers" (DAC)

iTransport "digital" iPod dock

dCS

Puccini U-Clock and Scarlatti Upsamper both have async USB supporting upto 24/96

 

 

That about exhausts my product knowledge of the mainstream companies and their approach to Computer Audio for the current time. I just thought it was interesting to see how different audio companies have approached the "next generation".

 

[johniboy]

Nice summary. Interesting that none of the non-pro companies use firewire, only spdif / aes, usb or network. At CA most people argue that this due to the need of specific drivers / software that non-pro companies can not develop themselves (so why don´t they outsource this task?? IT outsourcing is not that new...). Or can anyone think of other reasons why all these companies neglect firewire?

 

[idiot_savant]

Hi all, firstly, I'd like to thank Eloise for a succinct summing up of the current state of the market, and give my explanation of the lack of firewire interfaces in the market.

 

To put it as bluntly as possible, if you are a hifi manufacturer, looking to provide a good quality interface between a PC/Mac and your DAC, and you have a choice between: SPDIF, USB and Firewire.

 

Now, SPDIF is over 20 years old, so from a marketing point of view, doesn't sound modern or computery enough. Add to this that it's known as being suboptimal in jitter performance, and to do it right is really hard.

Next you might try firewire. Nice and modern, different, and very related to PCs and Macs. Additionally, plenty of bandwidth for plenty of channels at high sample rates. Unfortunately, there is no agreement on a universal implementation of how it should work, so, to develop your DAC, you're going to need to develop drivers. This, again, is hard, and probably doesn't fit the skillset of your engineering team very well, especially as you're going to need Windows, Mac and Linux drivers to keep up to date on all the millions of permutations. Now, you've done all the hard work, start to sell your DAC, and have to tell customers how to install the drivers, and indeed sometimes tell them that they're going to need a firewire card to add to their PC, which they may not be able to do ( any netbooks with firewire ports? )

So, now we come to the currently most popular: USB. Every PC and Mac made in the last 10 years has a USB port capable of sending audio to a DAC, it's computery for marketing, and it's got native OS support in every operating system!

Even better ( for some manufacturers, anyway ), it can be an extremely easy thing to add to your product - something like a PCM270x basically converts USB audio to SPDIF or I2S, which your DAC already understands - this solution is jittery as hell, but undeniably has a lot to do with the proliferation of USB audio.

 

Fortunately for audiophiles, there are some manufacturers who aren't afraid of a bit of hard work, and do come up with novel solutions, so there are good SPDIF implementations ( typically using a word clock or two to improve the jitter ), there are firewire implementations where the drivers work, and there are USB implementations ( specifically async ones ) that don't have any jitter issues,

 

Finally, note that all of the blurb about jitter is only to do with interface jitter, which is a contributing cause of jitter, but by no means the only one...

 

your friendly neighbourhood idiot

 

[Purite Audio]

IS, 'idiot' it is becoming increasingly difficult to use that moniker ! Very sensible , (as always) piece , seems to lead nicely onto jitter? Regards Keith.

 

[krzys]

Ok I confess I use Squeezebox (wired) ;-)). I don't know or can prove if this is the best solution to generate SPDIF signal for the DAC, but the interface and the commodity of installation are superior. The distance between PC and the SPDIF source become irrelevant. This is the strong point for "streamers".

And as nobody seems to pretend that the SPDIF from the SB is audibly inferior to a Transporter for example (by a noticeable margin), the SB seems to be a very good performance/price value

 

 

[idiot_savant]

Right, this is a very difficult subject to cover, and the source of much confusion here, and in the audiophile world in general. I am hoping that CG will step in if I go wrong, but we'll start with the basics....

 

As I have explained elsewhere ( mostly in the DAC forum ), making a good DAC is hard. Not only have you got to have an excellent "thing" that converts digital values into analogue ones, but you've may have to transform those values in some way, perform filtering on probably the input and the output, but you also have to make sure you do it all at the right time. Now, if you look at some of the stuff on the Multibit/Delta Sigma thread, you will see that some of this is counter-intuitive and needs thinking about, whereas everyone can understand something simple like jitter, right? This is probably why the community has become obsessed about it - it's something subtle, and different people do different things to "fix" it, and best of all, some people give out some numbers to tell you how good their DAC is.

Unfortunately, jitter isn't as simple as is made out, and yes, it is an important facet to consider with ultimate DAC output, but it's only part of the issue.

Right, so what is jitter, and why do we worry about it? Firstly, an old-fashioned view, that I hope people have got over is that "jitter makes the DAC gets the bits wrong" - this could be true in extreme cases, but will cause extreme discontinuities in the DAC output - typically, a sample may change sign ( e.g. go from -ve to +ve ), which will be heard as crackling, as opposed to the "veil" that people often refer to.

No, jitter is simply a small error in timing at the DAC output - the DACs job is make it's output even in time, and if this varies, the error in the time domain translates into an error in the amplitude domain. Typically, jitter is periodic, and this means that when the DAC reconstructs the signal output, it will add sidebands ( tones equally spaced either side of the original signal ) to the original signal that are related to characteristics of the jitter - this is what stereophile are looking for in their measurements,

 

the above is true for ALL DACs. If you like, I can go on later to talk about what causes jitter, and so on,

 

your friendly neighbourhood idiot

 

[johniboy]

IS, thank you for the comprehensive comparisons. This all makes much sense. Nevertheless, I have some more questions. Even though the most topics were already discussed in this forum, I would like to hear your (and the others) opinions - for the sake of curiosity

 

I understand that a good spdif interface is difficult to implement (just as probably every interface - there is no free lunch) and is maybe not the most modern and best solution. So far so good.

Now lets assume there are 3 other interfaces competing with each other:

1. usb

2. firewire

3. ethernet

 

I don´t want to add too much to the usb discussions but just the fact that there are so heavily different opinions about the sq (see The State of USB Audio by Alan Taffel thread) of this interface makes me skeptical. Ok, its really plug and play and every computer has it BUT as you mentioned, its jittery (if not very well implemented) and is also not compatible to all computers. I will quote Gordon from the thread mentioned above about Windows 2000 and async usb:

"First Windows 2000 is not 100% Class 1 USB Compliant. Windows 2000 did not include Asynchronous protocol (and other things as well) and one of the reason's why I was demanding of TAS about the Computer Hardware not being up to the capabilities of some of the dacs."

So if I take the usb limitations into account, I am left with a handful of good dacs, ayre, wavelength and dcs. Most of them not really a bargain...

 

Now lets look at the other 2 interfaces. I think they are pretty close, both having advantages and dis-advantages.

Ethernet is pretty widespread already in non-pro audio. Linn, T+A, logitech, and a couple of low-mid end hifi receivers with built in audio streaming use ethernet (or wifi). True, you do not need to implement drivers BUT you need to implement the data handling and player part yourself.

Advantage: you do not need a computer, a NAS also does the work.

Dis-advantage: Some (eg Linn) have a terrible user interface and the community is not large enough (yet) to make better open source software available in foreseeable time (in my opinion).

Handling is comparable for ethernet vs. firewire: Ethernet devices have to be connected to your network (if you have one!). For most non-computer audiophiles this is a real challenge! For this reason, the Linn dealer sets up your system, not you! This could also be done with firewire dacs. You could bundle the DAC with a firewire card if necessary. For laptops that do not have firewire, you can use pcmcia firewire adapters. Cost: 20$.

 

So why not more firewire? I see your argument with the driver but this could be outsourced and is probably not that more difficult than implementing a decent user interface. Pro-audio dacs mostly use firewire, why not the audiophile companies? Does ethernet have sq advantages over firewire? Can a computer connected to a dac via firewire sound as good as a NAS connected to a DAC via ethernet? Does the NAS - DAC connection have advantages because its less "noisy"?

 

These are great discussions!

 

Johniboy

 

 

 

 

[johniboy]

sorry, i got stuck on the interface question. But this relates very well to jitter, so please go on!

 

[idiot_savant]

As I alluded to earlier, writing drivers is a big deal for hifi companies - who, don't forget in have a limited number of customers

If you consider how many times, e.g. Creative Labs release a set of drivers, and consider the user base they have, you begin to understand the task they are taking on.

The other problem you have with say outsourcing the writing of your drivers is that you may get lucky and get a good set of drivers from a good team - what happens if they break, when say Windows 8 comes out or you want to add a feature? Similarly, Eloise has said that since the Macs changed their firewire chips, there have been problems reported, which probably means tweaks to the drivers - more expensive outsourcing for the manufacturer. It's just not a "natural fit" of expertise within a hifi company.

 

As for Ethernet, as you mention, that's yet another field of expertise/customer support for the manufacturer - I've steered clear of it so far, but may come back to it. There are at least two competing, incompatible standards in the market ( Linn and Naim use UPnP, whilst Logitech/Squeeze use Squeezecenter ). Not to mention Sooloos. Or Sonos.

 

I'm not saying technically any one is better than any of the others, but I hope it's clear as to why USB has such traction, even though there are many rubbish implementations, and only 3 ( at current count ) good ones?

 

And if you add in the "cool" factor for controlling iTunes with an iPhone/iPod Touch, well that can't hurt sales for the USB/firewire DACs...

 

your friendly neighbourhood idiot

 

 

 

 

 

[johniboy]

yes, maybe i do underestimate the difficulty of writing drivers... thanks for clarifying. it would be great if you could elaborate a bit more on ethernet streamers and dacs.

 

[idiot_savant]

Right, so we've explained that ALL DACs have an intrinsic amount of jitter due to the fact that oscillators oscillate in the real world, and this is a subject in it's own right...

 

I'm also going to ignore the use of ASRCs here...

 

So let us look at "interface jitter" - this is the area where how you connect the DAC to the source of audio comes into play. The basic thing with digital audio interfaces, is that the DAC has to consume as much data as the source generates. If you imagine that the DAC and the source disagree, only by a little bit on what 44100Hz is, for example, then the DAC wil either have too much stuff, and will have to throw some of it away, or not enough stuff, and have to repeat samples. The size of the buffer in your DAC will determine how much the difference in rates causes this to happen, but it will happen eventually. So, you need to match the two clocks somehow.

The classic approach for this is to use SPDIF, where the clock and data are combined into one wire, and the DAC can feed the embedded clock through a PLL to match the two clocks, and the speed at which the DAC tracks the SPDIF clock will define the amount of interface jitter it receives.

Now, some jitter will come through just because the DAC is tracking the source. How much will vary on the quality of the source, and the quality of the PLL in the DAC. Additional problems are caused by the data in the SPDIF stream modulating the clock ( data-dependent jitter ). Now, you can get very good SPDIF implementations, but these are not cheap.

Now, the idea of all of the more modern interfaces ( ethernet, async USB, firewire ) is that the DAC is the thing that generates the clock, and the source ( NAS/PC/Mac ) adjusts the rate at which it sends the data to ensure that the DAC never runs out or overflows. The precise implementation details vary, but this is the important point.

So (in theory at least) all 3 methods should be able to reduce interface jitter to zero, zilch, nada...

 

Now, in case you haven't spotted it, the problem with the Squeezebox approach ( if you're using the SPDIF out of the squeezebox ) is that there is no jitter between the NAS and the Squeezebox ( which is the bit you don't care about jitter ), but then you're relying on the Squeezebox & DAC SPDIF interface where it does matter.

 

NOTE before people start getting upset, these are the theoretical facts, and people disagree wildly about how much jitter is audible, etc. so I'm not going to go into it here, but the point is that the DAC is the crucial part...

 

Additionally, many network streamers ( e.g. Linn ) have a good DAC without the SPDIF link, and this is the conundrum - do you trade off potentially higher jitter using an external DAC, or lower jitter but stuck with the same DAC forever?

 

I may return to the "in theory" bit later,

 

your friendly neighbourhood idiot

 

 

 

[Purite Audio]

Fascinating IS really, you should write a book, 'the idiots guide' perhaps? Keith.

 

[idiot_savant]

One of the reasons why I post to this forum ( and no others ) is that there is very little "evangelism" here - people realise ( I hope ) that there isn't actually a perfect way of doing anything ( i.e. the best for all possible scenarious ), so aren't hell-bent on defending their opinions, but rather are willing to be a bit more open minded - I would hope I could count myself in this category as well.

 

What I try to do is explain some of the real compromises that are taken, why they are taken, and hopefully give some idea of how much they affect the important thing, which what happens in listening rooms,

 

your friendly neighbourhood idiot

 

 

 

[Purite Audio]

IS, yes computer audiophile is extremely unusual in that respect , I hope it may long continue, do you have a name ( Bruce Wayne ) btw, please keep up the good work, interesting and informative, Keith.

 

[idiot_savant]

but if I told you....

 

 

 

 

 

 

 

 

 

 

 

 

I'd have to kill you

 

 

your friendly neighbourhood idiot

 

[Purite Audio]

Can't you tell me and kill someone else?