jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am
"High End" computer audio
Elk
Elk's picture
Offline
Last seen: 4 years 2 months ago
Joined: Dec 26 2006 - 6:32am

I share your puzzlement, especially the obsession with USB.

My guess is that USB is of interest because it is easy and every modern computer has USB ports.

Yet, installing a soundcard with a good S/PDIF out is easier than many other operations in audio. Additionally, many have spectacular sounding analog outs.

By "streaming" do you mean wireless? (All computer based audio involves streaming by my way of thinking; i.e., real time playback of a file and the generation of a data stream.)

As one that has never had any problem with hard connected digital outs (optical, coax or XLR) the only advantage I see to wireless is the lack of a need to string wires. Wireless can be quite tricky to set up (and other times is incredibly simple when everything works great right away). The receiving units have the same jitter issues as any other DAC.

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

I agree, Elk, I think many people are attracted to USB because of the simplicity; since all computers these days have USB ports it saves having to add a sound card just to get a digital music feed (although I concur this can be a pretty easy process). With USB, the degree of jitter is highly dependent on the operating mode being used. Most of the common problems are associated with isynchrounous mode where the sender is controlling the clock. For asynchronous mode where the receiver (usually the DAC) is controlling the clock this is much less of an issue.

Streaming is not synonymous with wireless however. Streaming simply refers to a prototcol that sends a file in a way that means the start of the file can start being rendered/played before the whole file is received. Streaming primarily addresses two issues, a) where the bandwidth is constrained relative to the size of the file being sent, i.e. it would take too long to transfer the whole file and b) where there the size of the buffer at the receiving end is limited relative to the size of the file being sent, i.e. there wouldn't be room to store the whole thing anyway.

Audio can be streamed over both wired and wireless networks in exactly the same way. The only difference is the amount of work the wireless protocol has to do on error correction since there are so many more issues with signal strength and interference in a wireless environment. However, since the ethernet link (whether wired or wireless) is only being used for a bulk transfer to a RAM buffer (no timing information is being sent across this link) it is inherently jitter-free. I know I am beginning to sound like a broken record (corrupted file?) but many folks seem to either fail to understand or ignore that jitter is purely a clock issue and you only need a clock at the points of ADC and DAC. In between bits are just bits and there is no time domain to worry about.

I use a computer to drive my study headphone system directly (MAYA44 soundcard to Grace m902 over Kimber's Illuminati coax cable) and this works well for me. The Grace has a PLL to deal with the jitter on the S/PDIF link (although we know that this approach is able to reduce jitter significantly but never remove it completely) and the isolation provided by the cans means I never hear the computer fan or the (noisier) network drive. The advantages of having direct access to about 400GB of music, the ability to construct playlists etc. are huge and perfectly suited to a desk environment where I don't want to be distracted by having to find or change discs all the time.

For the best digital SQ possible today however jitter-prone interface standards such as S/PDIF and isynchronous USB must be avoided. In practice this means implementing a bulk data transfer to a buffer that is local to the DAC so the DAC itself can drive the clock and coupling this with a high quality DAC, output stage etc. There are still remarkably few products out there that fit this description (e.g. Slim Transporter, Linn Klimax) but their number is growing. I am personally very interested in new products like the Minerva from Weiss, with whose more conventional digital products I have been extremely impressed.

ChiDave1
ChiDave1's picture
Offline
Last seen: 10 years 2 months ago
Joined: Jan 19 2008 - 4:50pm

It's easy and inexpensive to replace the noisy fans with ones that are virtually silent. You could always pay a store to do it for you. I'll leave the other part of your question to the other folks here. I go over silencing my PC in my post at: http://forum.stereophile.com/forum/showflat.php?Cat=0&Number=43633&an=0&page=1#Post43633

RGibran
RGibran's picture
Offline
Last seen: 2 years 12 months ago
Joined: Oct 11 2005 - 5:50pm

Tha attraction with usb or soundcards is simple.

CHEAP! Cheap, Cheap, Cheap!

RG

RGibran
RGibran's picture
Offline
Last seen: 2 years 12 months ago
Joined: Oct 11 2005 - 5:50pm


Quote:
For the best digital SQ possible today however jitter-prone interface standards such as S/PDIF and isynchronous USB must be avoided. In practice this means implementing a bulk data transfer to a buffer that is local to the DAC so the DAC itself can drive the clock and coupling this with a high quality DAC, output stage etc. There are still remarkably few products out there that fit this description (e.g. Slim Transporter, Linn Klimax) but their number is growing. I am personally very interested in new products like the Minerva from Weiss, with whose more conventional digital products I have been extremely impressed.

Is the transmission of data via Firewire the same as being streamed via an ethernet link?

Otherwise I

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

Fair point RG; I realize I didn't express myself too clearly. I believe FireWire supports both asynchronous and isynchronous transfers, similar to USB. However while I believe that FW may offer some advantages over USB my interest is not in the Minerva per se but in the development direction it points to.

The interesting aspect is not the product type but the manufacturer. Weiss is a company with serious high-end credentials; I had the opportunity to audition their Jason/Medea CD transport/DAC combo and was extremely impressed. I am delighted to see them moving in the direction of computer audio. I am hoping they have an ethernet DAC a la TP in the works!

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am

Thanks one and all for your well thought out and nicely written responses. As usual Struts, who writes better than most native English speakers, if even he may not be one himself, provide a very clearly written definition for the concept of "streaming". Sometimes what is written in response to one's initial post can be quite surprising. For example I half expected several people to come to the defense of USB or other forms of hard wired connections, as opposed to networked connections, be they wired or wireless, and yet, so far, only one has.

But don't get me wrong, I'm glad to see that several others share my enthusiasm for networked music systems. I feel that these types of systems will become much more commonplace as the major means of music distribution moves to some form of file based system, distributed electronically, whether online or via cable/satellite (as today's television signals are transmitted). Sure it's always nice to have a hard copy as backup but I really don't see how the music industry will survive without eliminating the expense of producing CDs and switching to a file based distribution system. As time goes by more and more audiophiles will come to understand just how well a networked based music system fits in with a file based music distribution system.

In the meantime I'm looking forward to my return to the US in October so that I can finish ripping my CD collection to hard drive and clearing up space in my listening room by put the CDs in storage.

RGibran
RGibran's picture
Offline
Last seen: 2 years 12 months ago
Joined: Oct 11 2005 - 5:50pm

Computer audio has a wonderful side effect in that it leads to music servers, and this is a good thing.

RG

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am


Quote:
Computer audio has a wonderful side effect in that it leads to music servers, and this is a good thing.

RG

Of course you're right, RG. But there are many media center PCs (and macs) and media streaming devices out on the market and it's up to us, as audiophiles, to help promote and guide people to the better sounding options. Which is of course the reason I started this thread: to get a general consensus on what types of computer based audio systems yield the best sound. At the present time the consensus, however small it may be, would seem to point in the direction of some type of streaming audio system. I guess the next thread should be about which one these types of systems is the best sounding, along with all the other issues that go along with these types of systems, e.g. user interface, cost, ease of setup and operation, storage options, etc.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

I simply do not understand how a hobby that insists on top quality components, expensive parts, esoteric cables, and plugs that by themselves rival the cost of a computer. How can what is essentially a junk component (the computer) provide the sound of a decent source....how can a sound card costing a few hundred rival a CD player costing thousands???

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

JIMV,

I think it is fair to say that a computer soundcard should be able to produce a low-jitter digital stream at (possibly significantly) less cost than a CD transport, but I very much doubt the same is true when comparing to a CD player, i.e. when producing a high quality line-level analogue signal.

One of the principal advantages of digital technology is that information encoded digitally is extremely robust. You can store it, move it around a network or even around the world on the internet and thanks to the way it is encoded and the various ways that you can check that what you end up with is exactly what you started with you can be damn sure it hasn't changed a bit. You could store those bits on a CD, on a PC or in a rusty bucket and it wouldn't affect the sound quality one iota.

What can and does affect the sound quality greatly is how you convert it from digital back to analogue and how you get the analogue signal to your amp. Computers are a far from ideal environment to perform either of those tasks (for a summary of the reasons see this post). For that reason most professional recording engineers use standalone professional ADCs or external sound cards to perform conversion to the digital domain and most audiophiles use audiophile (and/or professional) DACs or external soundcards to convert back to analogue.

Good soundcards costs money, good CD players cost money just as good record decks cost money. Occasionally products will come along in each category that set a new price performance benchmark. However audio engineering is an art and a science and most of the best practices (power, grounding, interference/shielding, component selection and matching etc.) apply to D2A just as they do in the purely analogue domain. Moreover, refining them to the nth degree costs money, both at the design stage and the execution stage; there are no free lunches!

RGibran
RGibran's picture
Offline
Last seen: 2 years 12 months ago
Joined: Oct 11 2005 - 5:50pm

Ahh...were you talkin' to me?

I have no argument with you. "Computer audio" is evil in my book and should be avoided.

However, "music servers" are a completely different approach. The computer merely delivers the data (via ethernet)to a music streaming device which may contain it's own high end DAC or connects to an external high end DAC.

Surely your familiar with computer servers, the things alowing you and I to communicate at this very moment? They seem to serve up all types of data day in and day out with 100% accuracy.

RG

Elk
Elk's picture
Offline
Last seen: 4 years 2 months ago
Joined: Dec 26 2006 - 6:32am


Quote:
"Computer audio" is evil in my book and should be avoided.

However, "music servers" are a completely different approach.


There's a subtle distinction.

Is a computer a music server whenever it has a DAC connected by ethernet, but just a computer when the DAC is a soundcard?

What about when the DAC is connected by an umbilical, S/PDIF, firewire or anything else rather than ethernet?

To me the only distinction is whether it sounds good and this defines whether it is a good music system - regardless of protocol, OS, layout, etc.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm


Quote:
Ahh...were you talkin' to me?

I have no argument with you. "Computer audio" is evil in my book and should be avoided.

However, "music servers" are a completely different approach. The computer merely delivers the data (via ethernet)to a music streaming device which may contain it's own high end DAC or connects to an external high end DAC.

Surely your familiar with computer servers, the things alowing you and I to communicate at this very moment? They seem to serve up all types of data day in and day out with 100% accuracy.

RG

Its the 100% accuracy claim that I have problems with...is anything a perfect copy of something else? Folk didn't even know jitter was a problem in 1983. How do we know what artifacts and distortion is in that digit stream now?

I would find all this much more believable if I ever had the opportunity to compare a good audio system with one of these good computer based ones.

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
Its the 100% accuracy claim that I have problems with...is anything a perfect copy of something else? Folk didn't even know jitter was a problem in 1983. How do we know what artifacts and distortion is in that digit stream now?

JIMV,

Rest assured that when you copy a file on your computer what you end up with is a '100% accurate' identical copy. This claim is valid and is the fundamental point of digital, a bit is either a one or a zero, a one is always a one and a zero is always a zero, there is no 'close' or 'nearly' in the world of digital. This holds true regardless of whether it is a word file or a music file. This does not imply that every single digital copy is error free, but errors are so easy to trap and correct that the overall process can be made error-free in practice.

The timing information is not stored on the disk, it is (re)introduced at the time of digital to analogue conversion by the clock controlling the DAC. So while the information is in the digital domain it is 'stripped' of its timing information and it is therefore inherently jitter-free. The 100% accuracy claim only applies to this 'static' digital information.

Any 'artifacts and distortion' in the digital samples, as stored on the CD, hard disk or whatever were put there at the point of ADC (if not before). Neither ADC nor DAC are '100% accurate' and any claims that they are should be dismissed as marketing hyperbole.

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am


Quote:
I would find all this much more believable if I ever had the opportunity to compare a good audio system with one of these good computer based ones.

Jim,

You have a very valid point there but I think that you're not quite understanding just how these computer based audio systems work.

Basically there are two types of computer based audio systems:

The first type is one where the audio is being played back via some music playback program on the computer, e.g. iTunes, Winamp, Windows Media Player, etc., and sent to amps and speakers (or powered speakers, i.e. amp built into the speakers) which are directly connected to the computer. Very similar to the setup one would have for a typical office desktop computer.

The second type is one where the audio may be being played back via some music playback program on the computer or via some type of music server, such as the SqueezeCenter (used by the Slim Devices family of products) or Sonos. The digital signal (i.e. music file) is then send either through one's network or by some other means, such as a USB or digital out connection, to a digital to analog converter and from there to one's amps and speakers.

Among the differences of these two types of systems is that with the second type of system the DAC can be incorporated into one's "main" audio system. For example, I have a Slim Devices Transporter, which is a music streaming device with a very good built in DAC, and the device serves as just another front end, similar to a CD player or turntable, within my main audio system.

Just for the record, I have done many listening tests comparing the sound of my CD player (a McCormack UDP-1) playing a CD and the same CD, well actually the ripped lossless files from the same CD) played back via the Transporter and there are no significant differences that I've been able to hear. Basically all one is comparing are the two front ends: CD player versus music streaming device. The main difference between the two front ends being that the music streaming device is only one piece of an entire computer based audio system which also includes a computer, the stored music files, the server software and a network router.

Effectively what one is doing with a streaming music system is freeing the music stored on one's computer so that it can be played back on an audio system other than that of the computer. In my case, the audio system is my main audio system, which is located several rooms away from my computer.

I hope this somewhat long winded explanation helps you to better understand what is being discussed.

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
Basically there are two types of computer based audio systems

Actually jazzfan I would say there are three. I would put 'self-contained' systems like the Olive (and maybe the Sooloos) in a category of their own. In theory they are designed and built for the sole purpose of rendering high quality two-channel audio from hard disks while supposedly shielding you from much of the nightmare-inducing computer complexity. Looking forward to reading the Sooloos review in the forthcoming September issue!

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

Sounds a lot like the claims of 'perfect sound forever' that accompanied CD's 30 years ago. Again, I have to hear it to believe. For decades the be all and end all to quality sound has been high quality expensive components yet the toy we call a computer, built to a standard below chinese, wall mart audio, is suddenly 'high end'.

This bit is interesting " a very good built in DAC"...in your computer? Why does a 'very good built in DAC' begin at about $1000 as a stand alone and add about as much to a CD player but only cost a few hundred, if that, in a computer. Another issue, what about copying those CD's to the digital server...The CD tray and reader in a computer are audio dung. The internal wiring worse than dung, yet we pretend the quality of the signal placed in that hard drive is as good as the original.

If the high end requires quality parts and construction, I am unwilling to give the digital source a pass simply because it is a computer and not a CD player or transport.

Skellum
Skellum's picture
Offline
Last seen: 12 years 3 months ago
Joined: Apr 23 2008 - 5:07pm

I had a really good post about "audio dung" and the other fluff, but I then thought it would be good to go listen to my tunes, streaming off of the computer, and have another beer. BTW, all the folks here (at my house, listening) think this is the best audio they have heard in years. Maybe more later, but then maybe I'll sleep.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

I am really not trying to be a ass, just understand why we insist on quality products everywhere else in the chain but will accept what is arguably junk as a source???

If it takes thousands to build a decent DAC, or to connect all the goodies together, but will accpet cheasy printed circuits in the server???

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

That is the whole point of digital. You can create a word document on a $500 PC and a $5000 one. Print them on the same printer and I guarantee you'll get the identical result with '100% accuracy'. Up until the point of DAC digital audio is very similar. Which part of this explanation don't you understand?

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am


Quote:
Sounds a lot like the claims of 'perfect sound forever' that accompanied CD's 30 years ago. Again, I have to hear it to believe. For decades the be all and end all to quality sound has been high quality expensive components yet the toy we call a computer, built to a standard below chinese, wall mart audio, is suddenly 'high end'.

This bit is interesting " a very good built in DAC"...in your computer? Why does a 'very good built in DAC' begin at about $1000 as a stand alone and add about as much to a CD player but only cost a few hundred, if that, in a computer. Another issue, what about copying those CD's to the digital server...The CD tray and reader in a computer are audio dung. The internal wiring worse than dung, yet we pretend the quality of the signal placed in that hard drive is as good as the original.

If the high end requires quality parts and construction, I am unwilling to give the digital source a pass simply because it is a computer and not a CD player or transport.

Okay, let's try this again.

No one is saying that audio played back on a computer audio system is high end rather what we are trying to say is that with the right combination of components a high end audio system can be assembled which uses a computer as one of its sources.

The Transporter's built in DAC is not in a computer but in a device which is situated in one's main audio system, similar to a CD player, and connected to a preamp via audio interconnects. The computer is only used to feed the digital signal to the Transporter's DAC. The build quality of the computer makes little difference to the overall sound of the audio. In the world of computers the programs being used can and often do make a bigger difference than the hardware. In the case of ripping a CD to one's hard drive the choice of "ripping" software is more important than the quality of the hardware involved. As Struts has stated, with digital it's either 100% accurate or it's not - there is no 87% or 96% accurate - it's either all or nothing.

A better made computer will have absolutely no impact on the sound coming from a Transporter. Once the digital music has been created on one's hard drive that file can be accessed, copied, moved, backed up, etc. the same as any other digital file with exactly the same degree of confidence. And this exactly what is happening when one is streaming music - what is actually being streamed is the digital file: the file is sent from the hard drive on one's computer to a receiving device and from there passed on to the DAC and finally the analog signal is sent from the DAC to one's preamp and power amp.

Hopefully someone like JA, who's expertise you might have a little more respect for, will step in here and back up what all the others and I have been trying to explain.

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
Hopefully someone like JA, who's expertise you might have a little more respect for, will step in here and back up what all the others and I have been trying to explain.


Maybe argumentum ad verecundiam will succeed where logic has failed...

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am


Quote:
Maybe argumentum ad verecundiam will succeed where logic has failed...

Okay Struts,

First I was amazed that your command of English was so good, especially if you're not a native English speaker. Plus I also had to add in the possibly that you are also fluent in Swedish, which I assume is your native language. Now on top of this I have to add a better than average knowledge of Latin.

Anyway, I had to look up just what "argumentum ad verecundiam" means, which is "argument from authority", only to find out that from the view point of logic this type of argument is not valid. So I guess that means that we are on our own when it comes to trying to teach and convince Jim V. about the high end possibilities of computer based audio systems, or more properly, the use of computer based sources within a high end audio system and the JA can continue to do what he does best: edit Stereophile.

In any event, it seems that Jim V. is like a reverse DUP in that only super expensive equipment can and should be considered "high end". In which case I guess we should be calling him PUD!

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

jazzfan,

Sorry, I was agreeing with you in a roundabout way. If JIMV won't listen to logic then maybe he will listen to JA just because he is JA.

I am afraid that since English is my 'mother tongue' (but not my 'father' one) I am not worthy of the compliment, but I accept it with gratitude nonetheless. As I am sure you are finding out in The Netherlands, speaking multiple languages is pretty much the norm in Europe now and not regarded as nearly the accomplishment it was once was.

As far as Latin is concerned it was compulsory when I went to school and for some strange reason I developed something of an affinity for it. As the old saying goes Si hoc legere scis nimium eruditionis habes!

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am


Quote:
As the old saying goes Si hoc legere scis nimium eruditionis habes!

What? ?

Okay, I cheated and looked it up:

Si hoc legere scis nimium eruditionis habes! - Essentially it says, 'if you can read this, you're overeducated.'

Since I couldn't read it (actually I could "read" it, I just couldn't understand it) I'm taking it to mean that I received the proper amount of education and not one iota more than absolutely necessary.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

It just sounds so very much like the 'perfect sound forever argument' of 1983...I am not sure 'bits is bits'.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm


Quote:

Quote:
Maybe argumentum ad verecundiam will succeed where logic has failed...

...Anyway, I had to look up just what "argumentum ad verecundiam" means, which is "argument from authority", only to find out that from the view point of logic this type of argument is not valid. So I guess that means that we are on our own when it comes to trying to teach and convince Jim V. about the high end possibilities of computer based audio systems, or more properly, the use of computer based sources within a high end audio system and the JA can continue to do what he does best: edit Stereophile.

In any event, it seems that Jim V. is like a reverse DUP in that only super expensive equipment can and should be considered "high end". In which case I guess we should be calling him PUD!

It is not the argument from authority that is the only flaw, there is the false alternative argument widely in use as well, the 'either a computer is a perfect producer of a music data stream and all that implies, or it will not produce sound' argument. I am not sure that is the case. I have enough computers to know all are not created equal. It is like saying a Rolls Royce is exactly like a Yugo because both get a passenger to the drug store if desired. Is it possible that there are digital artifacts that are present but not yet measured in one computer that are not present in another? Are there audible differences between a server and a transport?

My issue was not about the cost of the gear, though it is related, but about build quality and parts. If bits is bits then parts can well be parts, yet no audiophile would buy into that idea, except when we come to the current toy du jour, the computer.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm


Quote:
jazzfan,

Sorry, I was agreeing with you in a roundabout way. If JIMV won't listen to logic then maybe he will listen to JA just because he is JA.

I am afraid that since English is my 'mother tongue' (but not my 'father' one) I am not worthy of the compliment, but I accept it with gratitude nonetheless. As I am sure you are finding out in The Netherlands, speaking multiple languages is pretty much the norm in Europe now and not regarded as nearly the accomplishment it was once was.

As far as Latin is concerned it was compulsory when I went to school and for some strange reason I developed something of an affinity for it. As the old saying goes Si hoc legere scis nimium eruditionis habes!

There is no logic in your argument, just some assertions of fact that may or may not be true, hence my reference to 'perfect sound forever'. If all computers produce identicle bits and their manipulation of those bits is identicle in all computers, then your argument reads that a computer of diverse and cheap parts and poor build construction not only can but must sound as good as a high end transport, for, as you all insist, bits is bits and computers handle them perfectly.

That really sounds more like wishful thinking than logic or fact.

Let me rephrase...If I came to you and said my $60 phillips combo DVD/CD player when used with its TOSLINK digital out through a benchmark DAC produces the same sonic signature as the new Bryston CD player used the same way, most folk would be dismissive or skeptical at best. If I said that. because bits is bits, the data stream coming from that $60 DVD player was identicle with the one coming from the Bryston so the sound must be also perfect and identicle.

I doubt anyone would buy that BUT, we must accept that bits is bits when a computer is involved....

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am


Quote:
It just sounds so very much like the 'perfect sound forever argument' of 1983...I am not sure 'bits is bits'.

I just don't see where any of us are making claims anything like the "perfect sound forever argument" of the early 1980's. Playing digital music stored on one's computer by whatever mode, be it streamed, networked or hard wired, is still playing digitally encoded music and, as such, it comes with many of the shortcomings, for lack of a better term, of digital sound reproduction. Meaning, is it perfect? No, it's still digital but it's digital that takes advantage of the digital side of the equation and makes better use of the digital nature of the source material.

By digital nature I mean the fact that basically a CD is just a store medium for what are essentially digital data files and it is these data files that contain the music, not the CD. Ripping the CD onto to a hard drive, besides providing a perfect copy of the digital data, liberates the data files from the confines of the CD and opens up many new possibilities on how to use the files and the data contained within those files. I know that the previous sentence used the word "perfect" but still, the fact remains that files ripped from a CD to a hard drive are perfect digital copies.

High End Computer Audio, which is still in it's infancy, makes better use of these digital data files simply by treating them as data files until the last possible moment, i.e. until they reach the DAC, wherever it may be located. So ripping a CD onto a hard drive as a wav file, taking that wav file and converting it to a flac file, adding tags, reading the data back from the hard drive, sending the data wireless to a receiver and finally onto the DAC does not effect the digital data in any way. Therefore the digital data being fed into the DAC is exactly the same as the digital data being read from the original CD into a CD player's DAC. That's it, that's all that we are claiming sound wise.

However, it's the advantages, along with some minor disadvantages, of a computer based digital music storage system that really make using such a system in a high end environment so desirable. Some of the advantages are:

Very compact storage, for example I have a Drobo multi disk Raid unit with 2 terabytes of usable storage and it should hold about 4000 to 4500 CDs worth of music as flac files.

Easy cataloging and updating of music database. Since there are no physical CDs there is no having to reorder one's CD collection every time a new item is added.

Never having to put away 20 CDs after are extended listening session. And no more scratched or damaged CDs.

The ability to stream one's entire music collection to any room or area of one's home and simultaneously too!

The ability to create playlists of any length.

Disadvantages include no being no printed material to read or browse through and the entire system being much more complex and harder to maintain than a simple CD player.

Jim, it could simply be that this type of system is just for you, however, I would recommend that if you get the chance to listen to a "high end" audio system that is using a computer based source component, such as a Transporter, to please do so. I think that you will be pleasantly surprised at just how good these systems can sound.

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

JIMV,

You appear to be completely misconstruing my statements, so I am obviously doing a very poor job of trying to explain myself. Let me have one last try.

I am not claiming that digital sound is perfect or any such thing. What I am saying is that up to the point of DAC bits is bits. Any computer that is capable of maintaining the integrity of the digital information (i.e. functioning properly, no quality judgements implied) will store and transfer the information identically and therefore cannot affect the sound.

Transports do sound different when played through the same DAC when connected over any S/PDIF link, such as Toslink (although the differences between a very cheap transport and a very expensive one can be remarkably subtle). This is because the quality of the master clock and the integrity with which that information is conveyed to the DAC can and does affect the sound quality and because S/PDIF places the master clock at the sending end of the link and embeds the clock into the datastream using BMC. So at this point the clock gets recombined with the data, and it's no longer 'only about the bits'. Essentially, the transport is an inherent part of the digital to analogue conversion process and therefore influences the sound.

So I am not claiming that a cheap computer will sound the same as an expensive transport. What I am saying is that until you start the digital-to-analogue conversion process it doesn't matter how many cheap computers the signal passes through, the sound will be completely unaffected. Any component that is part of the DAC process (or comes after it) will sound better the better it is designed and implemented with respect to traditional audio engineering preoccupations such as power, grounding, shielding, etc.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

"I know that the previous sentence used the word "perfect" but still, the fact remains that files ripped from a CD to a hard drive are perfect digital copies."

So the laser itself, the electronic assembly involved, the printed circuit, the cheesy wire in the computer between the laser and the hard drive and the quality of the hard drive are all perfect in their execution? Why is this true in the internals of the computer but untrue in the internals of a CD player?

You make good points but do not ahswer the quality question.

I use 'perfect forever' because, when the claim was made, folk simply did not know any better. The greater dynamic range and quiet as well as the ease of operation overwhelmed common sense. Was the product perfect...no, no one considered jitter, wave shape, low sampling rates as audible, etc. No, it was perfect because folk we respected said so and we could hear the improvements easily. It took a while to note the harshness and other digital artifacts.

Are we sure we are not seeing the same thing with todays new toys? Does build quality really not matter. Are copies of CD's or downloads really 'perfect'?

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
So the laser itself, the electronic assembly involved, the printed circuit, the cheesy wire in the computer between the laser and the hard drive and the quality of the hard drive are all perfect in their execution?

Absolutely not, but the software catches the problems and rereads the information until it gets it right. You might try ripping a CD with a program called EAC to see this in action for yourself.

Quote:
Why is this true in the internals of the computer but untrue in the internals of a CD player?

It is not, generally CD transports use CD-ROM mechanisms anyway. The difference is that the CD Transport is reading the disc in real time, so it has far less opportunity to reread the data and apply error correction. In this sense the computer has a big advantage.

Quote:
You make good points but do not ahswer the quality question.

I believe my previous post addresses this, I hope you agree.

Quote:
Are copies of CD's or downloads really 'perfect'?

Yes they are. If the computer is unable to make a perfect copy it will tell you that the file is corrupt. You will never be in doubt.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

Well written. I have a new question, is a corrupted file one in whih a single 0 or 1 is off or is the threshold higher, say 1% or 5%. When does the data stream not work?

Elk
Elk's picture
Offline
Last seen: 4 years 2 months ago
Joined: Dec 26 2006 - 6:32am


Quote:
It just sounds so very much like the 'perfect sound forever argument' of 1983...I am not sure 'bits is bits'.


We can verify that the digital file on the CD is the same as the digital file on the computer after ripping. That is, we know at this point bit is bits.

However, there can be issues with the digital transfer from the hard drive to the DAC. For example, USB typically has lots of jitter, ethernet is reported by Stereophile and others to sound better than wireless, etc.

Elk
Elk's picture
Offline
Last seen: 4 years 2 months ago
Joined: Dec 26 2006 - 6:32am


Quote:
Maybe argumentum ad verecundiam will succeed where logic has failed...


At last! Someone who will understand why I find the name of a local vehicle service shop amusing, Kath Auto.

Pete B. should like this also.

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
I have a new question, is a corrupted file one in whih a single 0 or 1 is off or is the threshold higher, say 1% or 5%. When does the data stream not work?

A computer can check for errors in a file very quickly with a simple checksum. This can easily detect random (non-correlated) errors of even a single bit. Here is an empirical test that may further convince you and which you can easily repeat yourself. More sophisticated techniques can be used to check for correlated errors (for instance two errors that cancel each other out). The principal trade-off is the degree of certainty that there are no errors versus the computing power required (I remember designing a CRC chip as a project at University so I can go into mind-numbing detail here if required ). CD uses an error correction algorithm called 'Cross-interleaved Reed-Solomon coding' (just rolls off the tongue!) which is extremely powerful and can correct a surprising number of errors. Needless to say, a computer can be programmed to ignore even uncorrectable errors below a certain threshold if that is what is required; it simply does what it is programmed to do.

Once the samples have been correctly read from the CD in a CD player (which is a difficult and error-prone process) further bit errors are rendered extremely rare by the way the digital circuits are defined. Nevertheless they do occur and some DACs contain filters that will spot bad samples and interpolate over them. If a bad sample is passed through to the DAC it will probably manifest itself in the resulting output as a an obvious 'click' or 'pop' and not as some more subtle deterioration in fidelity. Remember that since one sample represents the audio data corresponding to one forty-four-thousandth of a second, it would take errors in quite a lot of samples to really spoil your digital day!

Maintaining the integrity of the data (samples) is considerably easier than maintaining the integrity of the clock. Again, since no clock information is stored in the digital domain (it is 'removed' at the point of ADC and reintroduced at the point of DAC) maintaining signal integrity becomes relatively easy and the 'quality' (in traditional audio engineering terms) of the equipment becomes unimportant. Again, this is not true at the points of ADC and DAC or 'outside' those points.

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
At last! Someone who will understand why I find the name of a local vehicle service shop amusing, Kath Auto.

I don't know; it's all Greek to me!

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

Apologies, I missed the second part of your question

Quote:
When does the data stream not work?

If the preamble information becomes corrupted the receiver will lose its lock. This is quite rare but when it does happen it is almost always caused by errors stamped into the CD or by a faulty CD mechanism. Admittedly this does relate to quality, but in a binary 'working' versus 'broken' way, not the quality 'scale' that I think you have in mind.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

No, what I was trying to ask is, is there some level of 'not perfect' that is seen and accepted as perfect by the computer and passed on as perfect. are a few bit erros OK. Does te machine try to correct such erors and then pass them on, as does a CD player?

Editor
Editor's picture
Offline
Last seen: 13 years 11 months ago
Joined: Sep 1 2005 - 8:56am


Quote:
is there some level of 'not perfect' that is seen and accepted as perfect by the computer and passed on as perfect.

No. EAC will flag the number of data errors it could not correct using the CD's data redundancy and Reed-Solomon encoding. No errors flagged means that every recovered byte is correct.


Quote:
are a few bit erros OK.

No.


Quote:
Does te machine try to correct such erors and then pass them on, as does a CD player?

Yes. A typical CD has perhaps 5-50 fully correctable C1 errors per second. (The maximum allowable rate according to the CD Stadard is 200/second.) Whether you play the CD on a player or rip it using EAC or iTunes set to correct errors, the correction will be identical.

What you should be concerned about are C2 errors, which is when enough data are corrupted that the correct information cannot be reconstructed and the player interpolates what its designers feel the missing data should have been. (Often by just smoothly changing the last known correct byte to the next correct one.) These are the errors that EAC will flag. Fortunately, they are extremely rare - most CDs do not have any C2 errors at all, in my experience.

John Atkinson
Editor, Stereophile

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
Yes. A typical CD has perhaps 5-50 fully correctable C1 errors per second. (The maximum allowable rate according to the CD Stadard is 200/second.) Whether you play the CD on a player or rip it using EAC or iTunes set to correct errors, the correction will be identical.


The error count is really a function of the combination of disc and player since an 'error' that doesn't cause an error is not really an error. Some laser pickups cope with CD faults like for instance uneven pit geometry better than others and a disc that causes problems for one may play fine on another. The same disc played on two different drives could give significantly different error counts, indeed a disc could be completely unplayable on one drive and play fine on another.

The CD standard specifies a BLock Error Rate of 220 which means up to 220 frames per second can contain errors (so the number of actual symbol errors could be greater). Since the CD is read at a speed of 7350 frames per second this means that essentially 3% is the correctable error threshold. As JA points out, in practice modern high-speed drives usually give a block error rate under 50 (i.e. <0.7%).


Quote:
What you should be concerned about are C2 errors, which is when enough data are corrupted that the correct information cannot be reconstructed and the player interpolates what its designers feel the missing data should have been.


Actually some errors are correctable at C2 since the data are de-interleaved between C1 and C2 splitting the errors into separate frames. There are some CD quality metrics that classify CDs according to the incidence of different types of errors. E11 measures the rate at which one bad symbol is corrected at C1, E21 two bad symbols at C1, E12 one bad symbol at C2 etc. Since C1 can correct errors in up to two symbols per frame and C2 up to four symbols per frame an E31 will get passed to C2 where it can be corrected. However not all drives can cope with an E32 error, although it is theoretically correctable, so for the purpose of broad compatibility E32s are considered uncorrectable. Finally, a BURST error is defined as seven or more contiguous C1 frames containing E21 or E31 errors.

All this said, CIRC is so powerful that most CD playback errors are actually casued by tracking errors that cause the laser to skip rather than uncorrectable error bursts.

Sorry JIMV, you did ask!

Elk
Elk's picture
Offline
Last seen: 4 years 2 months ago
Joined: Dec 26 2006 - 6:32am

Cool info, Struts!

There still remains the question however as to what errors, if any, a computer will pass on to a DAC.

Let's assume a perfect rip of a CD so that what is on a hard drive is bit-for-bit accurate.

Are there any issues of getting an accurate data stream sent to an external DAC? I see two potential issues:

1) are the bits accurate?

2) is the timing accurate?

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm

Exactly Elk, data and clock, that's it.

I am only really familiar with S/PDIF, which is generally robust as far as data is concerned but generally poor regarding the clock. I do not have any empirical data on the incidence of data (i.e. amplitude) errors over S/PDIF but since any errors in the preamble will cause the S/PDIF receiver to lose its lock this will manifest itself very obviously.

Timing errors are unfortunately a fact of life with S/PDIF. They can be addressed with phase-locked loops, although this only reduces the problem, doesn't remove it. The best architecture I am aware of is to have the sender (transport, streamer whatever) slaved to either the master clock in the DAC or an external master clock such as an Apogee Big Ben using a dedicated 75 ohm coax connection. The only problem with this is that there is no standard for it in the Red Book specification.

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

"The error count is really a function of the combination of disc and player since an 'error' that doesn't cause an error is not really an error. For instance some pickups cope with uneven pit geometry better than others and a disc that causes problems for one laser pickup may play fine on another. The same disc played on two drives could give significantly different error counts"

Impossible...bits is bits...

struts
struts's picture
Offline
Last seen: 3 years 5 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
Impossible...bits is bits...

Unfortunately it is very possible. The pits and lands on a CD are physical objects and detecting the edges with a laser is an analogue process which is used to reconstruct the symbolic (digital) data.

Maybe the answer you are looking for is that digital is not perfect all the time (although it is most of the time), however when it isn't you will know about it. A file will either copy correctly or give a checksum error, a disc will either read correctly or give a read error, etc. The point is that there is no 'close' or 'nearly', it is either right or it isn't.

Or am I missing your point entriely?

Editor
Editor's picture
Offline
Last seen: 13 years 11 months ago
Joined: Sep 1 2005 - 8:56am

An excellent clarification of my over-simplified explanation, Struts. Thank you.

I think the point is that it is rare for the data on a disc in good condition not to be retrieved with all errors corrected, whether it is played in a conventional player or ripped to hard drive using something like EAC.

John Atkinson
Editor, Stereophile

JIMV
JIMV's picture
Offline
Last seen: 4 weeks 1 hour ago
Joined: Jan 31 2008 - 1:46pm

No, you are answering my questions pretty well. I am just finding it hard to believe something as poorly made and with such cheap parts as a computer can produce the quality sound (or bit stream) as the mega buck CD player or transport. While bits may be bits, crap is crap, and if build quality and part selection are required components for quality audio gear, it sould also be for all the gear in the audio train, including the source.

If one can build a audiophile system around cheesy junk for a source, one should be able to carry that build quality throughout the system and save megabucks for media.

Editor
Editor's picture
Offline
Last seen: 13 years 11 months ago
Joined: Sep 1 2005 - 8:56am


Quote:
I am just finding it hard to believe something as poorly made and with such cheap parts as a computer can produce the quality sound (or bit stream) as the mega buck CD player or transport.

You are forgetting that even if the buts are correct, they still have to be presented to the DAC at the right time. If they are not, then sound quality will be affected. Look for the articles on "jitter" in Stereophile's free on-line archives. A lot of what you are paying for in the high-priced player is the ability to present the rights bits at the right time to the DAC.

John Atkinson
Editor, Stereophile

jazzfan
jazzfan's picture
Offline
Last seen: 5 years 8 months ago
Joined: Sep 8 2005 - 8:55am


Quote:
No, you are answering my questions pretty well. I am just finding it hard to believe something as poorly made and with such cheap parts as a computer can produce the quality sound (or bit stream) as the mega buck CD player or transport. While bits may be bits, crap is crap, and if build quality and part selection are required components for quality audio gear, it sould also be for all the gear in the audio train, including the source.

If one can build a audiophile system around cheesy junk for a source, one should be able to carry that build quality throughout the system and save megabucks for media.

Jim,

What you are stubbornly trying no to comprehend are the vast differences between things digital and things analog. In the digital world many of the issues that matter, and matter a great deal, in the analog world do not matter or are of little importance. For example, digital differs vastly from analog in the area of copying: with analog there are always generational losses when one makes a copy but with digital the copies do not suffer from any generational losses - either the copy is perfect or it fails. Black and white, on or off, just like the 1s and 0s of a digital data set.

So while things like build quality can have a tremendous impact within the analog domain, in the digital domain the build quality of one's computer, as it relates to the "sound" of a digital file stored on that computer, is of absolutely no consequence. Sure the build quality of a computer will have an impact on certain things, such as failure rate (of the computer itself, not the digital information) and longevity, but since the rate of innovation within the personal computer world is still going forward at an alarming rate, it makes little sense to invest one's money in higher build quality when that computer will still become obsolete well within the normal life span of a similar but less well built computer.

Perhaps this is why so many of the music servers built by traditional high end audio companies come up short with respect to their sound. Too much attention paid to digital build quality and not enough attention paid in the areas that truly matter, like digital to analog converters and the analog side of the unit.

So too sum up:

Not only is possible for a bottom of the heap computer to function as part of a front end for high end system but given the right DAC, this bottom of the heap front end can and will outperform traditional digital front ends (read: ultra expensive CD transports) will relative ease.

Edit: By the way, there is a plus side to all your questions since all of this back and forth has managed to turn this thread into an excellent primer on the advantages of a computer based digital music system.

Pages

Log in or register to post comments
-->
  • X