View Full Version : Optimising a Windows computer for audio playback.
I don't think this has been discussed here before, but the Computer Audio Design (http://www.computeraudiodesign.com/) website has a lot of very useful information regarding setting up and optimising a Windows 7/8 computer for audio playback.
http://www.computeraudiodesign.com/wp-content/themes/cad_wp_theme/images/head1.jpg
Computer Audio Design sells the above high end DAC which I have no experience of and if any one is interested, you should check the site out. Its certainly very stylish and I really like the design, however, this post is mainly to assist people who may want to improve their Windows audio player, but don’t really know where to start and are maybe a little intimidated about shutting down services or poking around in the registry. There are some very handy guides in this section, which you can check out: http://www.computeraudiodesign.com/computer-setup/
In addition to the guides, there is also a script which you can run on a Windows 8 machine, which shuts down and disables a large number of services and drivers in the operating system. It should be noted that unlike Fidelizer, these changes are permanent and remain on the system after a reboot and are intended for use on an audio playback only computer. You can of course go backwards and turn the services back on and there is a script on the site to do this too. If you use the script on a general purpose computer, it may not function normally, so make sure you understand this before running the script. Additionally, it is also very important you take steps to carry out some simple safeguards to your system, by making either a full backup or disk image (recommended), backing up the registry, creating a Windows system repair disk and at the very least creating a System Restore Point. After you create a system restore point, ensure you reboot your machine before running the script. If you have never created a Restore Point manually before (it's simple) there is a guide from MS here: System Restore (http://windows.microsoft.com/en-GB/windows7/products/features/system-restore)
I have mentioned it before, but with the low price of small hard drives for the O/S in a music server, it can make sense to have a spare. I have two, a primary drive and a test drive which is a clone of the original. The clone is where I do all my testing, so if I mess it up there's no harm done and I just slot my original drive in, so I can still play music. When I am happy that any changes are both worthwhile and stable, I them implement the changes to my primary drive. The script is also open source and can also be found on the website. To allay any possible fears about running a script, I have opened it and read through it and you can do the same using a program like notepad, to review all the changes being made.
I think this is fantastic work from Scott and goes a long way to helping people get the most from their Windows machine, with very little effort. It was one of the things I was in the process of doing for a blog, but he beat me to it.
Finally, just a quick thanks to CAD for doing all this work, for making it open source and for allowing me to reproduce it on AoS.
Tim
NB: Just a final reminder, please ensure you take the safeguards mentioned above before implementing any changes to your system. If you are unsure how to, just Google them, there are numerous step by step instruction guides for everything I have suggested ;)
Thanks This looks really good
kenworthy100
15-06-2013, 14:48
Thanks Tim that looks to be a very useful and informative resource
Eek! 49 pages. Linux would fit on one page Tim. Just kidding ....
I used the script a couple of days ago, I seem to be one of the few people to have problems..my system wouldn't boot. Fortunately it wasn't too traumatic to restore the system. So do please ensure you have some way back if all goes pear shaped. A spare clone drive would be a great way.
Place your hard drive on a hard surface like glass or stone and listen to your favourite track.
Now find something soft, like a new sponge or even a pile of clothes. Place the hard drive on that and listen again. A HDD that is spinning around at 7200 rpm is like a turntable. It will vibrate. Trying to reduce or absorb that vibration improves sound quality.
I read this and... stopped reading the rest. :doh:
I guess I will come back later, and constructively enumerate all of the "opinions" stated as "facts" on this paper. More soon. ;)
Rothchild
15-06-2013, 16:41
Tonight I will mostly be A/B/Xing files in a folder against files in the root directory! ;)
There's some good info in here and there's some stuff that's been handed down (via Black Viper presumably) from the ark (or at least Win2K / XP).
I recognise many of them because it's the stuff we used to do to our DAWs (Digital Audio Workstation) for music production, when processing power and disk throughput for multichannel audio with effects was still marginal. Most of these tweaks were developed to minimise processor use or the chances of other programs trying to access the HD when recording the 'killer take', generally they have very little to do with sonics. (indeed the one tweak that probably will help is played down, windows system sounds are all 8bit11khz and can cause your soundcard / interface to try switching sample rates from what you'd set it to!)
2 channel audio (even at 24/96/192) is, these days, a walk in the park for most computers available (and that probably includes phones too!) and folk will get more improvement out of attending to first principles before disabling their firewall and antivirus. Good power, short runs of audio cable, trying not to mix audio and power cabling etc will IMO yield the big wins, most of the rest is just gilding the lilly.
2 channel audio (even at 24/96/192) is, these days, a walk in the park for most computers available (and that probably includes phones too!) and folk will get more improvement out of attending to first principles before disabling their firewall and antivirus. Good power, short runs of audio cable, trying not to mix audio and power cabling etc will IMO yield the big wins, most of the rest is just gilding the lilly.
Spot on ! :cool:
Good power, short runs of audio cable, trying not to mix audio and power cabling etc will IMO yield the big wins . . .
Absolutely and its been said before by myself and Gary (Gazjam) that those elements are essential and I don't think anyone would disagree with that. This is to help those along that want to tweak a computer further, but don't have the knowledge - its a helpful guide for those new to this, who haven't been dabbling with it for the last few years. There's nothing in there that's new to me, but there will be for a lot of other people ;)
I read this and... stopped reading the rest
So you have tried it then and noticed no difference Dimitri?
So you have tried it then and noticed no difference Dimitri?
Hi Tim,
no, I haven't tried it because it is useless for me. I'm not new to computers, and I know for having used hard disks for about 33 years (which is more than 17 million minutes or 1 billion seconds) that they never ever have read different data than the ones that are already on them. Except when they are broken, but then, they don't read any data at all ! :)
Let me give you an example so that people who don't know about computers could understand:
You have a page in a forum. Let's call this forum "The Art of Sound" and the page "Optimising-a-Windows-computer-for-audio-playback".
The forum is hosted on a server in the UK in Scotland - Glasgow by Iomart Group Plc. The page is stored on a hard disk on one of their servers.
My computer is in south France, Montpellier.
I ask my favourite browser to display this page. I press Enter.
My computer is not doing just that. Right now I'm downloading Dragon's Prophet (it has been downloading for 5 hours, there are still 2 hours left, it's a big game, hope it will be a good one) , while my sever plays "Fritz Reiner and the Chicago Symphony Orchestra".
My demand goes through my wifi, then to my router.
It actually crossed tens of switches, kilometres of cable, already several computers before reaching the Montpellier main server that connects to the Internet backbone (that is, the main line). All these computers have hundreds of hard disks.
Up to then, everything is ok. :)
My demand goes through 18 different servers, still more kilometres, still more switches, still more hard disks.
It reaches Glascow. A miracle, it arrived as it should ! Not a single bit was altered (otherwise my request could not be treated) !! :eek:
But it's ok. It reached the Iomart server in Glascow. ;)
The server responds and sends back the page. The response of the server (that is, the page itself) is much bigger than my request (which was just the address of the page). The response is about 150,000 characters.
The page goes all the way back through the hundreds of kilometres of cable, the tens/hundreds of switches, the 18 servers.
The page reaches my local DSLAM (http://en.wikipedia.org/wiki/Digital_subscriber_line_access_multiplexer), it is just 6 kilometres from my home. :champagne:
It crosses the last 6km of cables and switches, reaches my modem.
The page dispatches to my computer.
My computer (still downloading Dragon's Prophet, the go-and-back of the request took less than a second) possibly swaps some of the work to disk and starts buffering the answer.
The buffering is necessary, because the answer is big (IP communication packets are smaller than the page size). The page comes in small pieces.
My computer, finally has got the last piece. It gathers all the pieces together.
By the way, it's not easy to show the answer to the screen. I invite you to try to imagine what is needed to be done before the 150,000 characters (bytes) display in a readable way to my screen. A hard work, almost as hard as getting the page.
Whatever. The page displays on my screen. :mex:
And I look very carefully through my screen. Trying to find what was altered with all this go-and-back, these thousands of kilometres of cable, these switches, tens of computers on the way with their hard disks.
Hmm... apparently nothing. Because I can still read the page, everything is in order and all the letters are in their correct place. ;)
Good. It seems to work. :cheers: And it has been working for 33 years.
Now, could I try taking out my hard disk and put it on a soft cloth ? Yes I could.
But I wonder what could it possibly additionally bring to me than the correct page I have on my screen ? An even more correct page ? :scratch:
So... some may think that all this is magic. Actually, I know it is not, because I am of the ones that have built all this. That have conceived the necessary protocols and software needed to accomplish these operations without putting any cloth around your hard disks.
This would be a bit longer to explain why it is not magic, but trust me, there's no magic in this. ;)
Now, could I try taking out my hard disk and put it on a soft cloth ? Yes I could.
But I wonder what could it possibly additionally bring to me than the correct page I have on my screen ? An even more correct page ? :scratch:
So... some may think that all this is magic. Actually, I know it is not, because I am of the ones that have built all this. That have conceived the necessary protocols and software needed to accomplish these operations without putting any cloth around your hard disks.
This would be a bit longer to explain why it is not magic, but trust me, there's no magic in this. ;)[/QUOTE]
(1) Trust you? well sure, but you are not asking forum members to trust you more than anyone else, right?
(2) Is this a longwinded 'bits = bits' argument? If so then (a) I and others download stuff from Usenet. The use of parity files is normal because bits get lost. (b) The bits may all arrive but is not the the timing of that arrival necessary for hi-fi audio?
This would be a bit longer to explain why it is not magic, but trust me, there's no magic in this
So it seems you perhaps haven't tried it then Dimitri - have you ever tried building a dedicated audio computer?
I'm also not new to computers, but I have met this type of opinion before and with respect, you're on completely the wrong track and clearly have not grasped what's at play here, which is understandable if you have not experienced it or spent time trying to understand it. It has nothing to do with packet data or a bit-perfect data stream, or MD5 hash values and what the heck DSLAM has got to do with it is beyond me? After actually building 5 dedicated audio computers myself and testing different software, a number of different laptops and tower computers, different hard drives (SSD and mechanical, I've even tested differing disk RPM!). I have found that everything can make a difference, especially where you place those hard drives in relation to the motherboard if they are mechanical. How you power them (from the computer power supply or independently) and also whether they are SSD and where you install the operating system - it can all and does make a difference sonically. You are thinking the wrong way, it's about clean power, low ripple & noise, timing, jitter, vibration, isolating some components, EMI/RFI . . . the list goes on. Its audio Dimitri, not packet data over 100's of km of data lines.
But then you wouldn't know that, as you haven't tried it and more importantly have not heard it so your comments don't really hold water I'm afraid and are frankly unrelated. For me there is only one real test -hearing it with your own ears, I don't do fancy science, paper theory or pretty graphs - let me hear it first and then I'll make up my own mind about how it sounds and offer an opinion. If you had experienced just a sample from the above you might possibly hold a different view, but at least you would be qualified to share that view?
Thanks for all the smilies though, makes for a very colourful thread ;)
Tim,
Just playing devil's advocate here. But what do you say to the person who has tried all this stuff and heard no difference. Are they just deaf, or have useless kit, or both? And hearing a difference and deciding if it's an improvement is another point open to debate.
Tim,
Just playing devil's advocate here. But what do you say to the person who has tried all this stuff and heard no difference. Are they just deaf, or have useless kit, or both? And hearing a difference and deciding if it's an improvement is another point open to debate.
Well to answer that and I can only tell you from my viewpoint, is that I have not met that person yet and a few people have been round to listen and judge for themselves.
As to differences being just different or an improvement, well that's personal opinion isn't it and you don't need to debate anything? Look at me, I'm nearly 55, half deaf and I like Americana, country and wailing females :lol: . . . and at the end of the day, what works for the individual is the only thing that's important, not how much it costs, or what it looks like, but do I like it myself is really all that matters?
Do you know what though, I have tried quite a bit lately to help out but I'm clearly in the wrong place and wasting my time. . . .
I think most of us on AOS value personal experience
If someone has a different experience then it is good to try and understand the reasons
Do you know what though, I have tried quite a bit lately to help out but I'm clearly in the wrong place and wasting my time. . . .
You not in the wrong place many here really respect your experience and knowledge
Rothchild
16-06-2013, 14:39
but is not the the timing of that arrival necessary for hi-fi audio?
Yes, but this is a function of the quality of the clock source, it 'is' packet data (not audio) all the way up to the dac chip.
Tim, your input is valid and is appreciated. Please don't let some healthy scepticism and a bit of internet banter put you off posting your experiences. Disagreeing isn't the same as not getting on.
Well to answer that and I can only tell you from my viewpoint, is that I have not met that person yet and a few people have been round to listen and judge for themselves.
As to differences being just different or an improvement, well that's personal opinion isn't it and you don't need to debate anything? Look at me, I'm nearly 55, half deaf and I like Americana, country and wailing females :lol: . . . and at the end of the day, what works for the individual is the only thing that's important, not how much it costs, or what it looks like, but do I like it myself is really all that matters?
Do you know what though, I have tried quite a bit lately to help out but I'm clearly in the wrong place and wasting my time. . . .
Tim, please don't misunderstand what I have written. The aim of my explanations was not to be pedantic, nor to minimise you efforts to transmit your experience. :o
Let alone to have any judgement on your personal opinions (on hifi or whatever). Other forum members will attest that this is not part of my character.
Nevertheless, I consider that on forums, some healthy and polite discussion is important, don't you think so ?
Let me explain myself further, if you don't mind:
Experimenting, means that you change one (or more) parameters on a process, expecting to get different results. You know, the old human activity of causes and results.
From the experimenter's point of view there are two extreme cases:
- The "black box" point of view, and
- The "master" point of view.
When an experimenter is acting from the "black box" point of view, he considers he knows nothing about the process itself, nothing about the parameters that can be manipulated, and he can simply observe the result.
Let me give you an example of such a case: I take a dac, then place it face to the east, then place half an onion 10cm from it, and listen to the result at 10pm in my room.
Anybody sane enough, can say that I have modified some parameters: the place towards the earth magnetic filed (east orientation), the sun position (hour of the day), and, even the freshly cut onion can -potentially- modify the result.
Of course.
Unfortunately, there's a huge problem when I place myself in a "black box" condition: the number of potential parameters that can be modified is enormous (hundreds or thousands) and each one of these parameters has an important number of different values... In other words, when I place myself in a "black box" perspective, there are millions of billions of possible combinations to be modified. Trying to produce a "better" result in such a perspective is an impossible task. The statistics are against me.
On the other hand, when the experimenter is acting from the "master" point of view, he has a complete vision of all the possible parameters involved and all their values.
In this second case, the statistics are a lot on the side of the experimenter, which doesn't mean that the task of possible combinations is small. It may be big, but the possibilities of yielding a different result are much, much greater than the "black box" condition. Simple maths.
Now, of course, all of us are neither in the "black box", nor in the "master" position. We're somewhere in-between, depending of our (necessarily partial) knowledge of the process.
And this is ultimately, why I consider that forum discussion are necessary: they give us more knowledge in order to eliminate a lot of unnecessary parameters. To place us as far as possible from the "black box" condition.
So, if I seemed to you a bit harsh (which I don't think I was, but never mind), I apologise.
If you want to exchange on how hard disks and software work and which are the best parameters to look at, I can probably help.
If, on the contrary, you feel unease, no worries: I swear never to add a single comment about what you may possibly say about such matters.
But, rest assured, there was no criticism to you or to your experiments. Not before, not now.
Well to answer that and I can only tell you from my viewpoint, is that I have not met that person yet and a few people have been round to listen and judge for themselves.
I would loved to have been at chez yours to listen to a quality system for once. I would have asked you to play this wailing female (track two):
http://www.prestoclassical.co.uk/r/Avie/AV0030#listen
As to differences being just different or an improvement, well that's personal opinion isn't it and you don't need to debate anything? Look at me, I'm nearly 55, half deaf and I like Americana, country and wailing females :lol: . . . and at the end of the day, what works for the individual is the only thing that's important, not how much it costs, or what it looks like, but do I like it myself is really all that matters?
I've already reached bus pass age, and wage a never ending war with ear wax. So you can take all this with a pinch of salt. Despite my better rational self, I persued a minimalist tweaked Linux config on my kit only to find it made no real difference. Perhaps using open sound instead of ALSA was an improvement, and Linux sounded better than any Windows config I used to my ears. The only obvoius gain I've had was putting a gator board in my DACs. My conclusion was with my current kit, I should just relax and listen to what I've got. That doesn't mean I'm not interested in a better set-up (Rega plus Harbeths, I wish.) nor will I ignore the experience of others. But there is always the element of subjectivism and bandwagonism to consider. Not that I think either of these apply to yourself. Yet I don't thnk a healthy scepticism is mis-placed in the "audiophile" world.
Do you know what though, I have tried quite a bit lately to help out but I'm clearly in the wrong place and wasting my time. . . .
I don't think so and actually keep an eye out for your posts.
I would have asked you to play this wailing female (track two):
http://www.prestoclassical.co.uk/r/Avie/AV0030#listen
That's beautiful Chris and I have some of her other arias playing right now - thanks ;)
There is a big effect in action here and one that is ignored by the 'digital is digital' proponents who believe that, because the data doesn't change, no change in sound can occur. I'm afraid they are completely missing the point, due to narrow thinking.
What happens in a complex digital/analogue system is that everything digital can have an influence on everything analogue by virtue of the power supply rails and RFI/EMI. The harder the DAC works with jitter, the harder the hard disk works with vibrations, the more hash ends up on the power supply rails. Therefore, a hard disk or DAC experiencing outside influences like vibrations or jitter can put more correction-based noise on the rails. In fact, the best thing you can do for a spinning hard disk is to enable the silencing (acoustic) feature so that a lower acceleration is applied to the heads which puts fewer spikes on the power rails. Remember that all signals are derived from modulating the power rails. What you don't want is for them to be modulated by the wrong things.
Tim is right and I would have thought by now that members would know that EVERYTHING in an audio component has an effect on the final result, and that includes computer and streaming solutions.
That's beautiful Chris and I have some of her other arias playing right now - thanks ;)
"As With Rosy Steps The Morn .." is regarded by many as one of Handel's finest English language arias. I've not heard a better performance than that of Lorraine Hunt Lieberson. She departed this world in 2006 aged just 52 and is greatly missed. We are lucky that Harry Bickett had the foresight to take her into the studio in a reprise of her role as Irene in Handel's Theodora. LHL had been sensational in Peter Seller's 1996 Glyndebourne production. I don't care for the modern staging, but the music is something else. It ends in a bitter sweet love duet as the characters Didymus and Theodora face death together: http://www.youtube.com/watch?v=_O0yGpBiPdI
I won't burden you with a long list of my favourite mezzo-sopranos and sopranos but they include Sarah Connolly, Alice coote, Carolyn Sampson, Rosemary Joshua, Lucy Crowe and Emma Kirkby. Long live wailing females.
StanleyB
17-06-2013, 05:45
Let me give you an example so that people who don't know about computers could understand:
You have a page in a forum. Let's call this forum "The Art of Sound" and the page "Optimising-a-Windows-computer-for-audio-playback".
The forum is hosted on a server in the UK in Scotland - Glasgow by Iomart Group Plc. The page is stored on a hard disk on one of their servers.
My computer is in south France, Montpellier.
There is a major flaw in the example that is critical in the case of an audio signal. The time it takes for the page to load is not critical. So if the HTML code loads first, followed by any pictures a few seconds later, it does not affect the final displayed page and no errors will be visible.
In audio however, time is critical. Each audio bit has to be in time and in place, otherwise drop out errors will be audible. Preventing or cutting down on obstacles that can combat drop outs is part of the exercise in this thread.
There is a major flaw in the example that is critical in the case of an audio signal. The time it takes for the page to load is not critical. So if the HTML code loads first, followed by any pictures a few seconds later, it does not affect the final displayed page and no errors will be visible.
In audio however, time is critical. Each audio bit has to be in time and in place, otherwise drop out errors will be audible. Preventing or cutting down on obstacles that can combat drop outs is part of the exercise in this thread.
Stan, the way you put it, is as if the HTML page and all its components (pictures etc) were transmitted as individual mono-blocks.
But it doesn't happen like this, unfortunately. The HTML page and all its components are transmitted, stored locally and processed on the low-level components as a series of individual bytes. The order of their arrival is as important as their values. Much like in audio. If one single byte is out of order (or "arrives out of time") or its value altered, the page cannot be possibly displayed correctly.
I see what you mean, of course, but the components that use streaming protocols (without aknowledgment and error retry procedures) are only reserved for specific parts in audio. Hard disks don't use such protocols, for instance.
StanleyB
17-06-2013, 06:52
I see what you mean, of course, but the components that use streaming protocols (without aknowledgment and error retry procedures) are only reserved for specific parts in audio. Hard disks don't use such protocols, for instance.
Hard disks read the table of content, before gong away to fetch the various fragmented data. Defragmenting the data on an audio hard disk is hardly ever mentioned as one important consideration. I actually use O&O Defrag to keep my audio HD free from broken up music files.
Defragmenting the data on an audio hard disk is hardly ever mentioned as one important consideration. I actually use O&O Defrag to keep my audio HD free from broken up music files.
Indeed, big head seeks create large spikes on the power rails. Defragmenting a rotating HD (not an SSD) is important on a music server.
StanleyB
17-06-2013, 07:25
That's one reason why I like the SB Touch/PC combination.
I found out a while ago that the SBT will execute a resend request to the PC if the data is not properly read or fetched the first time round. The SBT uses its own buffer to take up any slack so that no breaks are heard. But of course if there is a lot of data resending required then you'll hear a buffer delay eventually.
The data resend request was extreme when I used the same router to direct the SBT data as well as any other network communication. I have since then fitted a separate N router that is wired directly to the SBT. My laptop is N compliant so I can push out audio data at a far higher wireless network speed compared to before.
Hard disks read the table of content, before gong away to fetch the various fragmented data. Defragmenting the data on an audio hard disk is hardly ever mentioned as one important consideration. I actually use O&O Defrag to keep my audio HD free from broken up music files.
If the HD in question is only used to store music files, in a non volatile state, i.e. one's only ever adding new files, and the disk has sufficient overhead of contiguous free space, and is less than 70% full, one should rarely get any fragmentation IMHO
But as a general principle file fragmentation is indeed your enemy...:)
If the HD in question is only used to store music files, in a non volatile state, i.e. one's only ever adding new files, and the disk has sufficient overhead of contiguous free space, and is less than 70% full, one should rarely get any fragmentation IMHO
But as a general principle file fragmentation is indeed your enemy...:)
Yes, just checked my NTFS formatted drive and theres virtually no fragmentation. Keep your audio files on a separate drive and try to only add files. Of course you could go to Linux and ext4 or Mac OSX HFS+ which are both more resilient to fragmentation ;)
tannoydmt
17-06-2013, 08:13
Well I downloaded the CAD Windows 8 Opimization script and ran it and it certainly disables a lot of services including my firewire connected TC Electronics 48 soundcard so couldn't playback any music, so glad I did a system restore point before running the script so I could return all my normal settings.
I don't intend to get involved in any lengthy discussions about this articles content, but it does make for interesting reading and I certainly agree with its content. Something I have tried many times to get across here (and frequently failed) is one of the final comments made by Scott and its this;
The key is to stop thinking about the computer as a ‘computer’ and think of it as an audio component. The sound quality you can get from a properly set-up computer and a good DAC is seriously good.Now the naysayers will no doubt say 'well he would say that as he's selling a DAC', but I have been saying it (and Welder (http://theartofsound.net/forum/member.php?2841-Welder)) for some considerable time and I'm not selling anything ;)
Anyway, enjoy.
Interview with Scott Berry designer of the CAD DAC, winner of best sound at show - Bristol 2013 (http://www.hifiplus.com/articles/meet-your-maker-hi-fi-visits-cad-computer-audio-design/?utm_campaign=Hi-Fi%2B+Weekly+Emails&utm_medium=email&utm_source=email-134)
Rothchild
08-08-2013, 21:21
What does buffer delay sound like?
Personally I think this is taking the pursuit of great sounds a little too far....
This is a Hydra in many ways and I would agree with the quote above if you intend to play direct from the PC. The power supply is such a critical component in this situation as is the s/w, however, if you separate the PC from the 'head' end then I would disagree...the requirements of the PC to be 'tuned' for audio is totally unnecessary and a waste of time and money.
walpurgis
08-08-2013, 21:44
There no limits or boundaries when it comes to high quality audio and the evolution of technology has to be respected.
Looking back, there have always been naysayers for each new media format. Each to there own, nobody is forced to use any particular source of music, you can even use wax cylinders on a nineteenth century phonograph should you choose.
I personally don't have experience of audio via computer (apart from ripping & burning CDs), but my impression is that things are advancing all the time and judging by many comments I've read on AOS (and elsewhere), it appears many people find it satisfactory. I'm sure the day will come that I will try this. At the moment I have no need to.
There no limits or boundaries when it comes to high quality audio and the evolution of technology has to be respected.
Looking back, there have always been naysayers for each new media format. Each to there own, nobody is forced to use any particular source of music, you can even use wax cylinders on a nineteenth century phonograph should you choose.
I personally don't have experience of audio via computer (apart from ripping & burning CDs), but my impression is that things are advancing all the time and judging by many comments I've read on AOS (and elsewhere), it appears many people find it satisfactory. I'm sure the day will come that I will try this. At the moment I have no need to.
A very balanced and sensibly reasoned reply Geoff.
I'm not a vinyl man despite growing up with LP's and a PL12D and there are, for a lot of people, many understandable reasons why they prefer to stick with CD as a digital playback format, be that as a single purchase to rip to a file based audio system (my method), or the simplest way, to use them as intended as a playback medium in their own right. I totally understand computer audio is a very daunting prospect for many, not least the task of ripping a large collection, which needs a substantial investment in time. I started accidentally by testing a DAC and comparing it to my CDP when I was considering an upgrade to the CDP, now I have gone over wholesale and I no longer even own a CD player, no need really. But what really confuses me are the many closed minds to something which is the most cost effective audio (sound quality wise) I have ever encountered and as Marco has very intelligently posted elsewhere, the serious music lover should not deny themselves of any platform.
Anyway, be that as it may, some people will never be convinced that file based audio when done well can be the most financially rewarding sound per pound you can get. It is also potentially foolish to ignore a medium which is fast become part of the landscape of modern audio. Its here and its not going away and there is no reason it should be a threat to vinyl (or CD) lovers, but don't dismiss it guys without really investigating what it can do ;)
Audioman
09-08-2013, 00:00
Can I point out that CAD have introduced a start up product for £6900. Apart from the fact it is cheaper than most of Linn's music servers this is a highly optimistic exercise. No matter how good it sounds and how well designed it is the material cost ratio to asking price is astronomical. Another company trying to make at least initial easy profits based on enormous profit margins per unit on low production runs. Looks like a one man band initially so I wonder how many he can sell at this price before he goes under.
Frankly no DAC is that expensive to design and build. To make a mark on the market and ensure a long term business they would have to invest in a bit of production engineering with a capacity to meet a larger market based on a much lower selling price. The purist approach of one USB input only will also put off the majority of potential buyers at this level that require the ability to connect network devices, disc transports etc. Frankly Tim this is not an example to champion in the cause of encouraging the take up of computer audio.
Frankly Tim this is not an example to champion in the cause of encouraging the take up of computer audio.
Its not intended to be that Paul, far from it really, just thought the article might be an interesting read after some here thought it fun to trash CAD - just trying to level the playing field, but I really am flogging a dead horse, especially compared to a 'Should I get a turntable?' thread which has now run to 189 posts . . . . just buy one FFS! :lol: Anyway, if I gave that impression apologies, just sharing some knowledge. I'm not really trying to champion anything, just attempting to educate those that claim computers are incapable of producing a good sound. They don't have to go down that road themselves, as its each to their own, but maybe just open their minds a little and bear it in mind when they make claims about it being rubbish and that vinyl or CD is the only way - it isn't. Here's my favourite remark made here at AoS about the merits of computer audio;
I take the sonic superiority of CD and vinyl for grantedThe trouble is when challenged it turned out that person had never actually heard a real file based audio system and was very clearly married to the romance of vintage gear and vinyl. Nothing wrong with that at all, in fact its great if you can find a medium that suits and better still, embrace them all, then you have all the bases covered for any type of music you find. I intend on buying a TT myself when I move and finally settle down, as I still have an LP collection and buy them at gigs, even though I cannot play them yet. I'm a music lover at heart and want to be able to play my collection, which for me ultimately means at least two sources.
I'm not, nor ever would be, in the market for a DAC costing that much, its more than twice what my entire system cost and at the moment I'm using my HRT Music Streamer II+ the most and may even stick with it TBH - that cost me £250.00. You also don't have to spend 600 on a USB cable to get good sound, there are plenty of alternatives.
Anyway, I said I wasn't going to get into a debate, so this will be my last post in this thread ;)
tannoydmt
10-08-2013, 12:55
Install Windows Server 2012 without any virus program with some basic configuration changes re max performance settings that should sound fine
I am fairly new to flac but have found that ripping to flac with eac and playing it through a hrt streamer 2 is in my opinon at least as good as a cd player and dac whih i also use....not sure that you really need to o anything else as it sounds bloody god to me.
Install Windows Server 2012 without any virus program with some basic configuration changes re max performance settings that should sound fine
Unnecessary. Since Windows Server 2012 shares its core with Windows 8, and the former is designed for virtualisation, you should stick with W8 for audio streaming purposes. It'll be a lot cheaper, too.
Grant,
try converting your Flacs to AIFF format.
Sounds better imo.
If you want to nip across and hear my digital setup (with valve amp natch :)) your welcome to.
I've recently upgraded my Dac and dac power supply so sounding great.
Gaz.
hi gary, never heard of aiff, will have a look and maybe do a comparison....am using Wasapi out currently with foobar.is aiff much different?
and thanks for the offer, I will do that sometime
Rothchild
14-08-2013, 09:44
AIFF is the same as .WAV but with the header file written the 'other way around' (one is 'big endian' one is 'little endian') but they're both just PCM files.
From my minor experiments I'd say that wav/aiff/flac are analogous. Take a track as wav (Or AIFF) and load it in to an audio sequencer, take the same track converted to FLAC and load it in to the next track of the same sequencer (ensuring the start points are the same) and invert the phase of that track, you will near nothing. This means the files are the same.
sq225917
14-08-2013, 10:08
Can I point out that CAD have introduced a start up product for £6900. Apart from the fact it is cheaper than most of Linn's music servers this is a highly optimistic exercise. No matter how good it sounds and how well designed it is the material cost ratio to asking price is astronomical. Another company trying to make at least initial easy profits based on enormous profit margins per unit on low production runs. Looks like a one man band initially so I wonder how many he can sell at this price before he goes under.
Frankly no DAC is that expensive to design and build. To make a mark on the market and ensure a long term business they would have to invest in a bit of production engineering with a capacity to meet a larger market based on a much lower selling price. The purist approach of one USB input only will also put off the majority of potential buyers at this level that require the ability to connect network devices, disc transports etc. Frankly Tim this is not an example to champion in the cause of encouraging the take up of computer audio.
The usual mark up from parts cost to retail is 10:1 you'd expect 5:1 for a direct to market vendor with no distribution level in their sales path. So frankly, given that he isn't making millions of them and does most of the assembly by hand in person it's pretty much spot on for what you would expect for a totally handbuilt, artisan product.
The question is do you think a dac that you can almost build yourself with a Xmos input board and a cheap chinese TDA1734 multi-dac-chip pcb from China is worth all that money?
The files are the same when measured that way Marc,sure, but the way the PC processes them is slightly different?
Flac requires more processing (albeit only slightly) and AIFF is processed differently.
Its easy to try in your own system and judge for yourself. :)
I converted my 2TB Flac library to AIFF.
the backups still in Flac though to make use of the smaller file size.
AIFF is the same as .WAV but with the header file written the 'other way around' (one is 'big endian' one is 'little endian') but they're both just PCM files.
From my minor experiments I'd say that wav/aiff/flac are analogous. Take a track as wav (Or AIFF) and load it in to an audio sequencer, take the same track converted to FLAC and load it in to the next track of the same sequencer (ensuring the start points are the same) and invert the phase of that track, you will near nothing. This means the files are the same.
Not used Foobar in years Grant, not sure how it plays AIFF to be honest.
Had a try of Jriver Media Centre?
Its the best sounding windows playback software imo by a long shot.
Theres a 30 day trial, good video overview here:
https://www.youtube.com/watch?v=n6YYsYX41e4
hi gary, never heard of aiff, will have a look and maybe do a comparison....am using Wasapi out currently with foobar.is aiff much different?
and thanks for the offer, I will do that sometime
Rothchild
15-08-2013, 14:24
The files are the same when measured that way Marc,sure, .
And what that measurement demonstrates is that there is no sonic difference between them (if there was they would not null).
Do you have a reference to demonstrate that the PCM file referenced by the .AIFF header is processed differently to the PCM file reference by the .WAv header. The extent which they are 'processed' differently is that which end of the line the big number is at. It's amusing that there are sides in the .WAV / .AIFF discussion because their only significant difference is the basis of one of the original internet 'holy wars': http://catb.org/jargon/html/H/holy-wars.html
Not being funny mate, but how can measurement demonstrate a sonic difference? :scratch:
(I do understand your inverting the phase as a test)
yup, lots of bun fighting over this...I just go with what sounds best to me in my system and encourage folk to try it for themselves.
And what that measurement demonstrates is that there is no sonic difference between them (if there was they would not null).
Do you have a reference to demonstrate that the PCM file referenced by the .AIFF header is processed differently to the PCM file reference by the .WAv header. The extent which they are 'processed' differently is that which end of the line the big number is at. It's amusing that there are sides in the .WAV / .AIFF discussion because their only significant difference is the basis of one of the original internet 'holy wars': http://catb.org/jargon/html/H/holy-wars.html
Rothchild
15-08-2013, 15:58
Not being funny mate, but how can measurement demonstrate a sonic difference? :scratch:
(I do understand your inverting the phase as a test)
I was just echoing your phrasing when you said that the files were the same when tested 'that way'. I was just concluding that by 'the measure' of nulling when phase inverted it was demonstrable that the files are sonically the same (because if they weren't they wouldn't null when one was inverted against the other), have you tried a phase inversion test yourself?
Rothchild
15-08-2013, 17:17
And for anyone who wants to try it themselves (or doesn't but wants to see it proved) here's a little video I just knocked up to demonstrate total null between WAV / AIFF and FLAC files. (Apologies in advance about the video quality, it's just a phone cam but it does what it needs to do to demonstrate the test at work)
https://www.youtube.com/watch?v=SJuQabY0i4M
I was just echoing your phrasing when you said that the files were the same when tested 'that way'. I was just concluding that by 'the measure' of nulling when phase inverted it was demonstrable that the files are sonically the same (because if they weren't they wouldn't null when one was inverted against the other), have you tried a phase inversion test yourself?
No, and I dont think I ever be likely to. :)
I know where your coming from though with the phase inversion test, and in that context your absolutely right.
I'd say though the test demonstrates that the files have the same structure rather than sound the same, but hey ho.
As I said though...I always struggle with how measurement can indicate how something sounds. :scratch:
Its your ears that do that surely?
Don't want to argue the point mate, your not wrong in what your saying.
I just say try AIFF vs WAV for yourself (or not - also cool) and see what you think in your own system with your own ears.
Personally Im not into measurements except when Im putting a multimeter near some DIY I'm doing.
best.
gaz.
Rothchild
15-08-2013, 17:46
What about a measurement that requires you to use your ears?
As you can see in the demo above that's what I'm doing, when my ears say 'can't hear anything' that means 'there is nothing' (and therefore no difference)
I'm a bit put out that you keep encouraging me and others to try your version of the test but you won't try mine ;-)
(just to acknowledge, I'm glad you're taking this in the good humour it's meant to be, I'm only shooting the breeze and hoping to learn something along the way here)
What about a measurement that requires you to use your ears?
As you can see in the demo above that's what I'm doing, when my ears say 'can't hear anything' that means 'there is nothing' (and therefore no difference)
I'm a bit put out that you keep encouraging me and others to try your version of the test but you won't try mine ;-)
(just to acknowledge, I'm glad you're taking this in the good humour it's meant to be, I'm only shooting the breeze and hoping to learn something along the way here)
Touche! :ner:
My lug'oles are the only test equipment I'll ever need.
I thought you were meaning though, that when you inverted phase and showed that the structure of the two files was the same...that meant that they must sound the same throught the speakers?
Subjectively to you (using your own ears) it did, which I get and was all I was saying that I trust my ears over measurements.
Its all subjective imo.
I've actually done all this testing stuff (including the phase test) a few years back, when I was using Foobar funnily enough.
Had several versions of the same album (Vangelis Blade Runner Soundtrack) and spent half a day listening to all the different versions I had (I have 8!) and comparing higher bitrate versions with lower, comparing different "editions" of the album (if your a Vangelis fan you'll know the troubled background to this soundtrack) with others both in CuBase and by ear.
I'll never get that time back - Strewth! :mental:
The idea was that I would identify the "best" version and lose the rest.
Turns out I preferred the different versions for different reasons, which I hadnt factored in!
Some have more ambience in the tracks, others more hifi sounding, others have longer edits of some of the tracks...that kinda thing.
I realised then that for objective measurement has nothing to do with how I enjoy music, its completely subjective.
Genuinely, I got to the point where I got fed up obsessing about it and just enjoyed using my ears instead of computer measurement.
The classic anti-measurebator line is "How do you measure how good a glass of wine is?"
You taste it.
How do I know what sounds better to me in my system?
I use my ears. :)
which is (getting back to the where I started) why I prefer AIFF over Wav.
Its subtle and I guess maybe you need a transparent system to hear it, dunno, but its not just me thats heard it in my setup.
But hey ho.
Happy here using the analogue measuring equipment strapped to the side of my head!
Happy here using the analogue measuring equipment strapped to the side of my head!
Ah but Gary, have you downloaded the latest patch for that? :D
Rothchild
15-08-2013, 19:17
Touche! :ner:
I thought you were meaning though, that when you inverted phase and showed that the structure of the two files was the same...that meant that they must sound the same throught the speakers?
I fear my test might be being misunderstood here. I'm not comparing the sound of the files played A/B/C against each other I'm showing that inverting the phase of one against the other results in silence (as confirmed by my ears). What this means is that they are the same regardless of what I might like to think I can hear in terms of differences between them.
I find with wine (much like hifi) it comes with a 'numerical quality guide' (the price list!) down the side of the page :lol:
Like everything in Hifi test for yourself and make up your own mind as if there is a difference between file types. Measurements do not tell us everything yet and the brain can be fooled, in the end its your money and personal enjoyment that matters
What about a measurement that requires you to use your ears?
As you can see in the demo above that's what I'm doing, when my ears say 'can't hear anything' that means 'there is nothing' (and therefore no difference)
I'm a bit put out that you keep encouraging me and others to try your version of the test but you won't try mine ;-)
(just to acknowledge, I'm glad you're taking this in the good humour it's meant to be, I'm only shooting the breeze and hoping to learn something along the way here)
Looking at your test here, with phase inversion, which is interesting, but IMO not relevant;
When loading a FLAC file into the workspace, the file is decompressed; i.e. it IS a WAV-file and if you'ld seen any differences you would've proved that FLAC is not bit-perfect... Which will not happen...
The only relevant phase inversion test can be made using recordings of real time playback of the different formats, which makes it a lot more tricky...
Feel free to disagree, but in my setup the differences between FLAC and AIFF are obvious when played back...
On the other hand a file converted into FLAC and then back again to AIFF is identical to the original AIFF. But, thats not the same as playing the file as an audio stream, where the timing-factor rears it's ugly head.
Anyway each to their own, i can deffo see advantages with FLAC's also, and to be quite honest, the differences are small enough to accept :)
Regards /Mike
Rothchild
19-08-2013, 08:14
Looking at your test here, with phase inversion, which is interesting, but IMO not relevant;
When loading a FLAC file into the workspace, the file is decompressed; i.e. it IS a WAV-file and if you'ld seen any differences you would've proved that FLAC is not bit-perfect... Which will not happen...
No, Reaper un-encodes the FLAC file 'on the fly', so this is a very relevant test because all of that 'different' or additional processing (which is supposed to make it sound different) is taking place right there in the program at the master buss where the tracks are being summed.
...are you sure about that? To see the PCM wave form on-screen and available for playback the FLAC file will have to be decoded first and buffered in memory or even written to a temporary file. You cant 'view' the contents of a FLAC file without decoding first...
I know some disagree and I have said this before, but when I first undertook the mammoth task of ripping my CD collection I spent a lot of time A/B testing file formats before committing to FLAC. I personally could not detect a difference between FLAC, WAV and AIFF - YMMV, but test them and pick what works for you :)
Regarding Flac, my take is that its the decompressing back to WAV (with its additional system overhead) that accounts for any perceived diffrence in sound quality.
Tested this on my work PC and even on on a high performance machine (dual xeon CPU's, 32GB Ram) the difference to me is audible.
As Tim said, its not night and day and might not even be audible to some.
So to me, its not the file format itself that is the issue, its all to do with having one less process in the chain before your digital music file bits reach your Dac.
I use Flac format for my music backup though, as I can cram a lot more music in that way becasue of smaller file sizes.
Compressed Falc vs Uncompressed?
Try converting a DSD x2 file to uncompressed Wav..an even bigger contrast, but at silly file sizes...naa.
Interesting though. :)
As Tim says, try it and see what you think.
Putting aside any perceived or real differences between file type playback for a moment, when playing a FLAC file it gets decoded into a PCM stream not a WAV file. Also, the real processing churn for FLAC is in the encoding, the higher the compression level the more CPU cycles it takes. The load on the processor during decoding is really minimal and the same for all compression levels...in my own tests and other reports on the Web show about 1% and maybe 2% extra CPU utilization when playing FLAC vs WAV.
Its also interesting to note with streamers like the SqueezeBox and reports of it sounding better when playing WAV vs decoding FLAC on the SB when the reality is the processor is under more load playing WAV due to the higher network activity...now that's not disputing what people perceive they hear but its very doubtful that higher processing utilization is the reason.
Hi Neil,
Yeah know the PCM thing, mentioned it as WAV in the sense that in most folks scenarios, for file format comparisons, the data would need to be put in a header of sorts, whether AIFF, WAV or whatever.
Decoding Flacs doesnt take up much horsepower at all, your spot on, but wierdly it still sounds different to me and thats even running Jriver on a higher spec machine. Other folk have heard it in my system too. Go figure!
I suspects theres SOMETHING going on, just not sure what exactly...and to me the fun is in finding out.
Ex Squeezebox user here too, I always set Squeezecentre Settings to decode Flac at the Server end to take the load off the Touch's limited resources.
Found that to sound better, especially when messing around with the Edo mods for USB output.
Must admit with the Squeezebox setup I found it didn't really matter what the server was as long as it was good ENOUGH.
Ran an old Dell Optiplex office machine for a while, £45 off ebay, windows XP and it worked a treat. :)
Putting aside any perceived or real differences between file type playback for a moment, when playing a FLAC file it gets decoded into a PCM stream not a WAV file. Also, the real processing churn for FLAC is in the encoding, the higher the compression level the more CPU cycles it takes. The load on the processor during decoding is really minimal and the same for all compression levels...in my own tests and other reports on the Web show about 1% and maybe 2% extra CPU utilization when playing FLAC vs WAV.
Its also interesting to note with streamers like the SqueezeBox and reports of it sounding better when playing WAV vs decoding FLAC on the SB when the reality is the processor is under more load playing WAV due to the higher network activity...now that's not disputing what people perceive they hear but its very doubtful that higher processing utilization is the reason.
Its also interesting to note with streamers like the SqueezeBox and reports of it sounding better when playing WAV vs decoding FLAC on the SB when the reality is the processor is under more load playing WAV due to the higher network activity...now that's not disputing what people perceive they hear but its very doubtful that higher processing utilization is the reason.
The issue is that the Squeezebox Touch has very limited processor power and works hard to decode FLAC. Receiving more WAV data without decoding is a relatively light load.
...... I always set Squeezecentre Settings to decode Flac at the Server end to take the load off the Touch's limited resources.....
The issue is that the Squeezebox Touch has very limited processor power and works hard to decode FLAC. Receiving more WAV data without decoding is a relatively light load.
....but thats the thing guys, its the other way 'round, the higher I/O activity is more processor intensive than the FLAC decode for the SB, you can see this when you run TOP from the CLI...
Ah ok, real data. I hadn't run TOP from the CLI as I simply forgot it can run those Linux commands.
I'll do a few more listening tests as it could be something else going on.
Rothchild
19-08-2013, 19:33
Ah ok, real data. I hadn't run TOP from the CLI as I simply forgot it can run those Linux commands.
I'll do a few more listening tests as it could be something else going on.
In the spirit of where this thread has gone, HTOP is better (pretty colours to boot) :eyebrows:
I think I'll just listen to some music, much more fun :D
Were you guys running Squeezecentre in Linux?
Windows here.
Running Windows here Gary. I was trying to get a screen capture to show but I'll need to have another look at it as I was getting inconsistent results with a lot of CPU peaking...thought I could quickly use the wireless connection to my server but for some reason it was playing up and kept dropping out.
And for anyone who wants to try it themselves (or doesn't but wants to see it proved) here's a little video I just knocked up to demonstrate total null between WAV / AIFF and FLAC files. (Apologies in advance about the video quality, it's just a phone cam but it does what it needs to do to demonstrate the test at work)
Marc
Rather convincing test. It is good to have these measurements. If I may make an analogy: a 60% 60.5% and 59.5% triangle may look equilateral but once you have measured it you know it is not. I would trust the measurements rather than trust my eyes.
Putting aside any perceived or real differences between file type playback for a moment, when playing a FLAC file it gets decoded into a PCM stream not a WAV file.
I'll second that, which also applies to ALL formats of audio in the digital domain IMO. :)
In the case of FLAC, it has to be decoded to PCM data to be played or edited
Yup, thats the reason for Rotchild to not identify any differences between his three files. :)
All audio is based on PCM in the computer audioworld today (with the exception of DSD, which most of the time is at some stage converted into PCM)
Therefore all other formats are containers for PCM audio, AIFF, FLAC and even lossy formats like OGG, MP3 and similar hideous versions of fidelity...
But this is the quintessence of the question/opinion:
I would trust the measurements rather than trust my eyes.
I feel the other way around, if my ears tell me there is a difference, there is a difference. I might not be able to explain it or measure it, but that just means ii'm using the wrong tools... ;)
Of course, this is all in my opinion, based on gathered knowledge, epxeriments and mindset etc.. :eyebrows:
Best regards /Mike
Rothchild
20-08-2013, 07:07
I'll second that, which also applies to ALL formats of audio in the digital domain IMO. :)
Yup, thats the reason for Rotchild to not identify any differences between his three files. :)
The reason I didn't identify any sonic difference is because there isn't any.
Doubters may obfuscate all they like about containers / formats etc. The bottom line is that the output delivered by the three file types is the same, if I were to throw an mp3 in to the mix too it wouldn't null in the same way (despite the fact that it too is being decoded to PCM for the D/A to output) because... it's not the same.
realysm42
20-08-2013, 07:23
If that's what they are (the same), why are there so many versions of it?
I fear my test might be being misunderstood here. I'm not comparing the sound of the files played A/B/C against each other I'm showing that inverting the phase of one against the other results in silence (as confirmed by my ears). What this means is that they are the same regardless of what I might like to think I can hear in terms of differences between them.
I find with wine (much like hifi) it comes with a 'numerical quality guide' (the price list!) down the side of the page :lol:
Not trying to be funny Marc, but your earlier post here kinda indicates that you aren't comparing sound quality between different formats, only the null test.
I feel your test wasn't identifying a sonic difference, just comparing bit structure of both files which (of course) are the same because they are using same PCM data?
What your comparing here is exactly the same PCM data, so of course its going to cancel out.
MP3 wouldn't because as you say its different.
Personally I don't think all files ARE outputted the same as there are factors which differentiate them, header type, the need to decode to PCM first...etc.
There are arguments on both sides whether these factors actually make an audible difference - I believe they do and this is bourne out in my system to my (and others) ears.
Its not night and day by any means, the very least is that you can tell they are different.
As I said, its subjective. Its how our brains work.
Sometimes what you hear and what you measure don't always add up.
If you go by measurements how do you know your measuring whats important, are you measuring everything of relevance? Is your methodology set up correctly?
...Has it been peer reviewed? ;)
If someone goes by ear (like me), perhaps you favor that kind of distortion or it works better with the synergy in your system?
Maybe its possible over analysis can get in the way of musical enjoyment?
That said, its possible I'm delusional, OCD, deaf or just plain mad. :)
Were you guys running Squeezecentre in Linux?
Windows here.
Windows here too, Gary. It's the Touch that runs Linux.
Windows here too, Gary. It's the Touch that runs Linux.
Not familiar with TOP or HTOP Martin, wondered if it was something you ran on a linux machine.
Made me wonder if you guys were running squeezecentre on a Linux box. :)
TOP shows CPU performance and running services live on a Linux machine. It's great for monitoring our firewall at work.
like Task Manager for Linux?
Cool.
Regarding the SB touch and Flac, I take back what I wrote earlier. Spent way to much time on it but observing playback using TOP with Flac and Wav at 16/44 and 24/96 I get the impression that WAV uses less total processor cycles vs Flac, however, what I did observe or seemed to observe as I didn't chart it and compare is that Wav causes slightly higher peaks of CPU activity which seem to coincide with Network access. Overall CPU % use at 24/96 was Flac 35%, Wav 25% and at 16/44 Flac 13% and Wav 11%
What surprised me is how many times and length the CPU hits 100% running 24/96 on both Flac and Wav...
With regard to FLac vs Wav payback, this is an interesting read:
http://www.computeraudiophile.com/blogs/mitchco/flac-vs-wav-part-2-final-results-155/
Yep - I think the Touch's CPU is just good enough for the job. I'll leave my settings of decoding FLAC at the server and sending a simple stream alone then. It certainly tallies with what I've heard, which is that the Touch with EDO and doing the least work sounds best (a nice external PSU helps, too). Disabling the screen on playback helps, too.
Maybe its possible over analysis can get in the way of musical enjoyment?
File formats - the new cable wars :doh:
As I said before, I'm off to play some music and have some fun, you crazy kids play nicely now :lol:
Oi! It's your thread. You can't just go, you know!
Oi! It's your thread. You can't just go, you know!
:lol:
Not much for me to add Martin, I did extensive A/B testing via my main system and through my headphone system before embarking on my mammoth ripping task and I couldn't detect a difference, so went with FLAC due to its file size. Admittedly I don't compress my FLAC files by that much, so a heavily compressed one will obviously need more processing power, but even an Atom CPU barely ticks over on my server whilst playing a FLAC file. I guess its possible to detect a difference on some systems and having an unsuitable configuration in a streamer could perhaps come into play during the processing, but for me there was no discernible difference, so I am clearly a sceptic and not convinced on this occasion by peoples reported subjective listening and as I'm from the digital world, I'm happy to accept the science when it comes to file format discussions. That's not to say I'm right, just that in my case the science equals my real world subjective listening experience - at 54 I doubt my ears are as good as a twenty somethings either?
Anyway, at the end of the day I'm more than happy with what I have, so I'm tending to shy away from these circular arguments discussions these days.
I think everyone needs to try it themselves, as I always like to do and go with what you believe is best. Once you make your mind up to something, not many AoS folk are often swayed towards an opposing position, me included ;)
Originally Posted by MartinT
Oi! It's your thread. You can't just go, you know!
:D
Tim, As long as we respect each others opinions and keep it civil there's no reason to not have a good 'ole debate about it...Flac decoding overhead is the same no matter what compression level was used, its the encoding that chews processor cycles.
Has anybody read the thread I linked to above...
I'm a regular reader of the CA forum so I have read that Neal, very good article indeed - lots of good discussion over there.
I totally agree there's nowt wrong with a good 'ole debate if that's what floats your boat and please don't anyone take this the wrong way, as its not meant as any kind of sleight, but I'm bored with it to be brutally frank. What floats my boat is music itself and the means to play it is secondary once I find what I need, which I have done. I'm inclined to be more a leveller than a sharpener ;)
Are You a Sharpener or a Leveler? (http://www.stereophile.com/asweseeit/are_you_a_sharpener_or_a_leveler/index.html)
Rothchild
20-08-2013, 18:11
I feel your test wasn't identifying a sonic difference, just comparing bit structure of both files which (of course) are the same because they are using same PCM data?
What your comparing here is exactly the same PCM data, so of course its going to cancel out.
MP3 wouldn't because as you say its different.
The null test is exactly about comparing the sound of the files, if they don't sound the same (even by 0.1 of a dB at a single, tightly notched frequency - which I challenge anyone to be able to reliably hear) then they don't null.
I'm trying to get my head around how you can maintain a claim that essentially states that there are things that are both the same, and different (wav/flac/aiff) and some things that are both different and different (MP3).
If you can understand that an MP3 won't null because it sounds different (which we all have no trouble acknowledging) how can it be that the stuff that does null is somehow because of something more complicated than the simplest possible explanation (that they sound the same, exactly the same)?
None of this gets in the way of any matter of personal preference, perceived difference (and preference) etc etc but there's such a clear logic flaw here (things that are the same are different but things that are different are different too) that I just want to understand a bit more.
I know you're only yanking my chain but I'd love for someone to 'peer review' this test. Anyone can download the free, un-expiring, demo of Reaper (www.cockos.fm) rip three files and do the null test that I've demoed, I'd be very surprised for anyone to find anything significantly different to what I've done. As for the relevance of the test it's fairly easy to demonstrate, as long as you understand that 1 - 1 = 0 then the use of the test is readily apparent.
As have been observed 'ALL' audio files (including MP3s) are rendered to PCM for delivery to the DAC so this aspect is a red herring, as we've acknowledged the PCM data coming out of an MP3 type file is different to the PCM data coming out of the 3 file types I've compared (which is the same between all three, which makes them the same)
Rothchild
20-08-2013, 18:19
:D
Has anybody read the thread I linked to above...
Yeah, I've had a look through, looks like a more involved version of what I've done in a quick and dirty way. I quite like this line as it sort of sums up my take on it (I've also been involved in music production for some time):
"So do we not trust our ears? I am not saying that. What I am doing is bringing a balance of both subjective and objective thoughts together so we can correlate what we hear with what we measure and vice versa. Again, when performing ABX listening tests, if you are hearing an audible difference, then the waveform must have changed. If the waveform has changed then we can measure the difference."
its...the...same...PCM...data... :)
of course its gonna null!
I dont think an MP3 doesn't null because it sounds different, it doesn't null because the DATA is different.
Could be the same container, the same need/no need for decoding etc, but it wont null becasue the data is different.
Nothing wrong with wanting to understand more, same here!
Aren't you measuring PCM data though, hence like for like?
I've done all this testing before Marc, it proves nothing in the scenario where you sit down in front of your speakers and hit play.
IMO of course.
The null test is exactly about comparing the sound of the files, if they don't sound the same (even by 0.1 of a dB at a single, tightly notched frequency - which I challenge anyone to be able to reliably hear) then they don't null.
I'm trying to get my head around how you can maintain a claim that essentially states that there are things that are both the same, and different (wav/flac/aiff) and some things that are both different and different (MP3).
If you can understand that an MP3 won't null because it sounds different (which we all have no trouble acknowledging) how can it be that the stuff that does null is somehow because of something more complicated than the simplest possible explanation (that they sound the same, exactly the same)?
None of this gets in the way of any matter of personal preference, perceived difference (and preference) etc etc but there's such a clear logic flaw here (things that are the same are different but things that are different are different too) that I just want to understand a bit more.
I know you're only yanking my chain but I'd love for someone to 'peer review' this test. Anyone can download the free, un-expiring, demo of Reaper (www.cockos.fm) rip three files and do the null test that I've demoed, I'd be very surprised for anyone to find anything significantly different to what I've done. As for the relevance of the test it's fairly easy to demonstrate, as long as you understand that 1 - 1 = 0 then the use of the test is readily apparent.
As have been observed 'ALL' audio files (including MP3s) are rendered to PCM for delivery to the DAC so this aspect is a red herring, as we've acknowledged the PCM data coming out of an MP3 type file is different to the PCM data coming out of the 3 file types I've compared (which is the same between all three, which makes them the same)
Rothchild
20-08-2013, 19:15
it doesn't null because the DATA is different.
And what does the DAC use to generate the analogue signal? The Data.
If the data is the same (which it seems, at least, that we're in agreement that it is) how or why would the DAC generate a different analogue signal? (Surely it can't, because the data is the same.)
I give up, honestly I wish it was that simplistic. (no offence)
with that...I'm out. :)
With regard to FLac vs Wav payback, this is an interesting read:
http://www.computeraudiophile.com/blogs/mitchco/flac-vs-wav-part-2-final-results-155/
It is interesting indeed. I can't comment on it just yet because I think I need to read it at least three more times to do justice to the poster. Thanks for the link!
Rothchild
20-08-2013, 22:02
I give up, honestly I wish it was that simplistic. (no offence)
with that...I'm out. :)
None taken, shame but thanks for playing ;-)
Regarding the SB touch and Flac, I take back what I wrote earlier. Spent way to much time on it but observing playback using TOP with Flac and Wav at 16/44 and 24/96 I get the impression that WAV uses less total processor cycles vs Flac, however, what I did observe or seemed to observe as I didn't chart it and compare is that Wav causes slightly higher peaks of CPU activity which seem to coincide with Network access. Overall CPU % use at 24/96 was Flac 35%, Wav 25% and at 16/44 Flac 13% and Wav 11%
What surprised me is how many times and length the CPU hits 100% running 24/96 on both Flac and Wav...
Neal,
I did this experiment some time ago, and replicated your initial findings, i.e. network IO workload gets a LOT higher at 24/96 with server-side PCM decoding vs. decoding FLAC on the SB Touch. I put it down to the fact that the SBT's SoC NIC doesn't have any hardware acceleration, and therefore many more CPU cycles were needed. Your results may be different, depending on whether you are using the factory settings, or frigged buffer sizes or other tweaks.
I am now using the excellent EDO mod for the SBT, and it manages 24/192 FLAC playback fine... may do some more testing to see what the CPU is doing (if anyone still cares!).
A link to the discussion - see SBT red book and hi-res FLAC and WAV decoding (http://theartofsound.net/forum/showthread.php?15594-File-based-audio-playback-compressed-vs-uncompressed&p=291108#post291108)
Cheers,
Alex
I am now using the excellent EDO mod for the SBT
It's mandatory for the best quality. Next is a good external linear PSU. Both of these transform the Touch from an also-ran to a contender.
It's mandatory for the best quality. Next is a good external linear PSU. Both of these transform the Touch from an also-ran to a contender.
Got one - an MCRU PSU. I was very pleased with the improvement in SQ.
Cheers,
A.
Yup, I had the prototype of that PSU, made a good improvement.
Nick Gorham knows his onions.
realysm42
25-10-2013, 15:12
I've just tried this guide, after having created a backup etc.
The only thing I've not done is use Jplay (I'm aware this is meant to have a large impact).
Tbh, I think it sounded lots better with the script having been run, I've just run the system restore and everything sounds flatter and less interesting. I wasn't expecting this to work at all, so if I had expectation bias it was negative.
Very interesting.
Try MQn its free and most people prefer it to Jplay
Wee bit off-topic(ish) but worth mentioning I think.
You can get good results with software, but I think its its well worth looking at how your server's powered? Get the "infrastructure" sorted first and I think you'll be in a better position to appreciate all the incremental tweaks software can bring.
Biggest jump for me recently was going "all linear" powered on a new server build.
My previous box doubled as a games machine so had a standard (though very high quality) Seasonic ATX supply, as well as a fanless graphics card in there.
Built a new one, all change.
Strictly music 'n movies, took the opportunity to rethink how it was powered.
Added extra feeds from the balanced mains and I'm powering it with separate linear supply's for the MB, CPU and both SSD drives. Another one is powering my usb card.
Its all hard wired fed directly from the balanced mains unit.
Nice step up, I think improving the source is really making the most of the rest of my system.
Well worth having a think about imo, get rid of those switch mode power supplies! :)
Yes Tim reported similar benefits
Powered by vBulletin® Version 4.2.3 Copyright © 2026 vBulletin Solutions, Inc. All rights reserved.