> I fall wholeheartedly into the ‘tone deaf’ part of the crowd when it comes to analyzing audio and prefer an easy-to-read number as a comparison, although I understand this is not the be-all and end-all when it comes to audio.<
The problem is once you go beyond easy-to-read numbers, audiophiles go crazy and start forming sentences that contain more fluff than an Apple press release. I acknowledge that there is difference in sound quality of across devices, but the willingness of audiophiles to hear unicorns sing is amazing. The snake oil salesmen of old have all switched to audio gear, for sure. So yeah, stick to numbers.
Well, the fire proof test I have is to use Studio Monitors (Active ones) to test if the sound card is good or not (among other things, of course). They usually detect noise and other artifacts the sound cards produce right away, so you can either cry or give the sigh of relief.
That being said. I've managed to nail down that a MoBo can be said to have a decent sound card if it posses at least an ALC892 in it. This chip/codec contains very good sound processing and options to actually enjoy anything (specially from the HTPC perspective). I also have a Creative Sound Blaster X-Fi HD USB and it very worth its price tag, even not being 5.1 or 7.1. If the sound card sucks, I attach this baby and fix everything.
I would agree, please stick to objective testing and quantifiable numbers like SNR. Subjective audio testing really can get bizarre. As soon as reviews start commenting on how "warm" or "brown" a sound card is, I'm going to start skipping the audio sections.
Honestly, I would prefer subjective audio listening for as long as comparison and description is clear and understandable... Let's say, subjectively the reviewer will compare the audio output of this motherboard's audio to a, say, Asus Xonar STX or even just the DX... of course, quality difference will be tremendous but at least from there, I can have an idea how it really sounds... numbers are just numbers...
Your analysis of the situation is born out of ignorance and your own bias. Numbers are still an inaccurate substitute for ears. We have a situation here something like the reporting of frames per second as a measure of video quality - it is a factor, but hardly tells the whole picture, as has (finally) been proven recently with the focus on measuring consistency in frames started by The Tech Report. SNR and THD are good numbers to know, but they are just a starting point for evaluating sound quality and a rather small part of the big picture.
It is no fault of Anandtech or any hardware site that better analysis isn't done, as it is expensive, time-consuming, and until the last couple of years not all that important in mainboard testing. However, since Asus and others are now claiming they have some kind of better sound solutions on their boards, it is good to have as much information as we can. We can't yet get a total package of measurements that will indicate clearly what we will hear, and a site like this getting involved in subjective impressions (meaning using their ears to measure information they can pout in a review) is difficult, so reporting the numbers that are possible is a good thing, but hardly paints a complete picture.
I'm sure there is plenty of "snake oil" in the audio business, as there is in all businesses, but it isn't to the degree you indicate. People should buy with their ears and their budget, and if an inexpensive on-board sound solution works for you, good news, you can save a lot of money. Others have better ears and want to get more out of their listening experiences, sneering at them because you can't hear what they do is childish at best.
There's no if's, and's or buts about snake oil in the audio business - it flat out exists. Numbers don't like (but can be misreported or misrecorded as this article indicates) but ears...ears *DO* lie. They lie for a number of reasons. Genetics, age but even more...what you want to hear. Spend 10k on those speakers? You're going to damn well believe they sound amazing. that 1k wooden nob that "isolates" the sound of your receivers volume? Yea, you want it sound good.
It also boils down to personal preference. Some people want the most accurate sound reproduction possible. Others want a warm, natural sound. Some just want booming bass. To each of those, their "ears" will find issues. The numbers, the science? They'll show the truth. So, don't let your own ignorance and bias color YOUR responses
IMO what really needs to be done is to find a way to quantify these aspects of audio that are currently hidden behind fancy words. Aside from SNR and THD+N, things like "attack" can actually be seen by plotting the square wave response of a DAC or amplifier.
So that's one snake-oil buzzword down, now what we need to do is try and get the rest of them.
We can render very detailed plots from frequency/levels using raw digital files. Now run the same file through the audio equipment, capture the output and see how the new plot compares to the pure digital original and we have an objective way of seeing exactly how the equipment changes the original waveform, and slowly, but almost certainly we can begin to find what quantifiable differences there are (or the lack thereof) when an audiophile reviewer says something like "airy", or "forward".
There's a good intro to DACs and digital signals from a guy at Xiph, which are the same guys that write the FLAC audio codec. It addresses some common myths about digital audio, like "stairstep" samples, bit-depth, and dithering. Even just from this ~24 mins of video, you can see a lot of silly audio rumors still floating around online today are absolutely false.
It's a little more complicated than that. After all the original was no doubt recorded (then post-processed) in a studio. You are playing that audio back in a very different environment with a different speaker setup and most likely without the kind of professional quality microphone used in the studio.
So, sure, it's that simple for the studio to evaluate and tweak their audio setup. Simply use a high-quality mic and make sure the original sound matches the playback as closely as possible at that microphone (repeat with the mic at other points you care about playback quality).
But since you don't have the original room nor the original mic this begs the question of WHAT sound you should be reproducing. Different studios are likely to take the exact same original and capture it in a slightly different manner depending on the type and number of microphones (where it's sensitive) they use and other factors. So how do you even define a standard against which your sound can be compared?
I tend to think the best way is to insist on binaural recordings (recordings made by mics placed on each ear of a dummy human head) and standardize on a particular kind of mic that should be used to compare against the original. Of course, how you weight different distortions is going to make a bunch of difference but at least you have a clearly defined notion of perfect reproduction to start from.
Attack can be quantified by slew rate and rise time. I'm happy you brought that up. I have a nice little home studio based around a relatively ancient Motu 828MKII firewire soundcard. These boxes have reasonable sound quality, and work really well with Macs. Mine is not stock, it has had the audio opamps and digital clocking upgraded to tighter specs. Ironically it actually has slightly worse SNR now. However the THD+N is much lower, and the bandwidth gained an octave on either side. There is nothing subjective about things rattling off your desk, or being able to record frequencies higher than I can hear. This brings up my next point. A lot of mobos have optical outs. Those are neat, but they are next to useless if the signal is so jittery that it sounds worse than the analog signal, or worse, is so jittery it can't communicate the audio to your AV device.
This business about wanting a warm sound and the like made sense when we were stuck in the bad old days of analog audio but it shouldn't persist.
Deliberately choosing an audio component that distorts the sound being reproduced only ensures that you can never get exactly the sound you want. Sure, it might make the sound seem more warm but when you want to play something where the input already captures that warm sound you still get that distortion on top of the already warm audio you are playing.
Now that audio files are all digital instead of choosing an audio solution less capable of reproducing the recorded sound the most accurately (I know that's a bit vague) simply reprocessing the sounds in software should suffice to imbue them with whatever effect you desire. This then would allow you to take the same audio files from one device to another and expect them to all sound very very similar.
Yes, I know that if you want your in game sound to be warm this isn't really an option but really, how good can the game be, if you are paying attention to the fact that the background music isn't as warm as you usually prefer. Ideally, you could tune these parameters in hardware (and at the high end I'm sure you can) so you could get the effect even in games.
Numbers can't be an inaccurate substitute for ears. Our machines, which represent the audio in numbers, are far far more sensitive to changes in pressure (sound) than the human ear.
I don't doubt that some people with sensitive ears get something out of some reasonably subtle effects. However, just as with wine when double blind studies are done many people are not able to reliably identify which of the experiences is better despite claiming it was clear as day when they knew what was what.
The suggestion that you just go into the store and see what you like is a recipe for wasting money. You know that people are easily subject to placebo effects here and yet you place yourself in a situation where a friendly sales guy can subtly maneuver you using your desire to appear sophisticated and knowledgeable into convincing yourself that, why yes, you can tell the difference between the cheap junk unit and the fancy one.
Also, as I've already suggested there is absolutely no need to go out and buy an audio unit tuned to your tastes. It should be possible to simply buy the one that objective analysis indicates is best able (using taste only to decide what kind of distortions matter more) to reproduce the widest range of sound at your ears and simply modify the audio in software so that the actual sound you hear is whatever you prefer.
Unfortunately while a good portion of audio does involve actual science, there is a large amount of pseudo-science, subjective opinion, and even placebo effect involved. A couple years ago, there was a bit of a debate about lossless 24-bit 192kHz audio being better than normal 16-bit 44.1 kHz, and how people with "golden ears" or "better ears" could clearly hear the larger range of frequencies. Then some people at Xiph, who develop the popular FLAC audio codec, stepped in and showed how they were all being silly. I particularly like how they compared these audiophiles to hypothetical "spectrophiles" that claim they can see X-rays and microwaves everywhere. It just doesn't happen.
Even earlier than that, years before iTunes came around, there were 2 competing MP3 player programs for Mac, Audion and SoundJam. Despite both of them implementing the MP3 decoder the same way, numerous audiophile review websites and magazines all claimed that Audion somehow sounded "better" than SoundJam, even though the Audion devs weren't doing anything special. It's also mentioned in an article about the history of Audion and iTunes on the dev's site.
I'm not familiar with these players, but MP3 decoders aren't necessarily created the same (e.g., dithering, among other things), and the rest of the playback chain can be variable as well.
I really feel all these advanced audio options on motherboards are pure marketing. If you give a damn about audio quality from your PC you use a digital output and an external DAC, and these days there are lots of very cheap options available. All of the "audiophile" components in the world are not going to give you good results if they are located in the EM quagmire inside a PC.
While it's a hard pill to swallow with all of the marketing fluff that surrounds audio and the even more flamboyant marketing around PC OEMs and motherboards there is good engineering sense behind the things they do. "THE RED LINE". Cheesy as hell but having a separate ground plane helps and paying close attention to the ground path of every component is absolutely necessary in high fidelity audio circuits. There's no shortage of reading on the characteristic of caps and in some places the quality of the cap won't affect performance but in other places close to the signal path the can. Nichicon, WIMA, Nippon, etc. are all more expensive but also _may_ result in better performance.
The marketing gimmicks alone don't make a good product though. The implimentation (ie the layout) is much more important than the components chosen. If you pull up the spec sheet for any one of these codecs you'll see absurd SNR and dynamic range numbers. Expecting those out of your speakers is like expecting to get 1.9Gbps out of your fancy 11ac router. It's just not going to happen for practical reasons. The quality of the tiny circuit around the codec is going to be largest limiting factor in most cases.
So while it may seem like marketing trash there's a good reason attention is being paid to this part of the board. There is absolutely no reason at this day and age to spend $200 on a dedicated sound card unless you are working in a professional studio. Even these fancy on board sound cards are good enough for amateur recording given a $100 mic.
I appreciate the RMAA tests that you're doing. Even if it has it's limits, it's a very useful tool, IMO.
But it seems that, when you're testing, you're always using the same onboard soundcard for the input. As you also noted, this means that the input performance can "bottleneck" the results. But I'm guessing the majority of readers don't use the line-in at all and would rather just see the clear line-out performance and the headphone performance (you can test the latter with a 3.5mm splitter and a headphone load).
RMAA lets you use different in/out devices, so why not get a good quality dedicated soundcard that has a verified good quality input with high SNR and use that for testing the output on the motherboards? (And you could also use that to test the input performance independently.)
I second that. You could get a good USB audio interface to do the test. This way even though the bottleneck would be possed on the audio interface, at least the test runs with consistency across test subject (mainboard).
The problem would be to get the proper audio interface to work with.
PS. Audio Interface is another name of the device we call sound card, usually used for professional-level AFAIK.
THIS. At least then you keep the variable quality of the input fixed when comparing outputs and you can also run things the other way - have the external reference provide the output to test the input on the motherboard. There are plenty of "pro" audio usb options you could use for this for not too much $$. e.g. E-MU 0204-USB
Hmm, the claim in the article is that RM:AA is actually much more sensitive to poor implementations when you use the same hardware for both input and output.
In particular, by feeding back the same input into an output on the same device you can sometimes create feedback effects from any signal leakage while detecting the same leakage merely by analyzing the audio out would require a very very sensitive instrument because you've moved the potential interference both closer to the output (EM leakage is more likely from the input that from random computer parts) and can amplify that weakness to a more noticeable size by creating the appropriate kind of feedback in software.
I understood that this was the argument for using the same input and output. Sure, ideally you would have a very sensitive input that lets you directly examine the output for bias (including using feedback) but lacking that you instead choose to increase the pressure on the output by running the most problematic kind of signal right next to it.
Also, one thing I noticed with some onboard chips is that they start making noise with CPU/GPU load (sometimes just scrolling in some window), so it might make sense to try to simulate that and see if it affects the results. Perhaps simply running Furmark during the test would do it...
See the comment from davidedney above, there is a good likelyhood that you are noticing EM emissions from the CPU/GPU or power supply components connected to it being induced into the analog part of the audio solution. Testing this is far from trivial, as it does not only depend on the Mainboard itself, but also on the type of CPU, GPU and PSU; as well as the case geometry and internal reflectivities for the frequencies in question.
It may be far from trivial, but this is exactly the kind of functional testing I'm looking for in a review here at AnandTech. I want to know if the solution *works* and in what ways - headphones, line out, optical out, front connection on a case (maybe this should be a part of the case tests too). What headphones can I buy that are supposed to work, and do they? Will it hum, crackle or pop? How does it work connected to an AVR?
As a reviewer, I think it's best to stick to the numbers. I think we all appreciate that you are trying to add some additional info to your testing. I think consistent testing is the key if you can't get the absolute top end testing equipment.
Audio is a tricky thing for sure, and it has a certain, well let's call them devout group of followers.
I'll add my voice to those who prefer that you stick strictly to numbers, as trying to judge something as nebulous as audio "quality" is bound to backfire.
For me, the most important aspects of any PC audio test are noise threshold and audible interference. I can't tell you how annoying it is to spend 20+ hours recording a track to have it mangled and mashed into some unrecognizable combination of squeals, beep, ticks and hisses by cheap MoBo audio on a "premium" board. If I am paying extra for enthusiast gear, I expect the audio to be accurate, distortion and interference free at the very least.
I'd be more interested in you guys exploring sound editing software and text to speech then hardware testing audio chips that have been more then good enough.
For instance one of the selling points for me getting Creative soundcards over the competition in ages past was the ability to record audio from wave output. In ages past you couldn't record audio from a wave device that was playing it at the same time internally without resorting to round about methods using the analog out into the mic/line in. It was a royal pain in the butt.
AP is certainly very respected, but they're not the only player in town. More so if you don't need super top-of-the-line performance and features.
There's the dScope Series III and R&S UPV, off the top of my head, but those are relatively expensive as well. There are cheaper analyzers too, but I don't think it's particularly that worth it for a site like AnandTech.
One thing perhaps worth considering is still using RMAA but being more careful about actually physically measuring output levels for consistency and so on. Also, rather than loopback, you could record with a high-quality (and by this I mean a couple orders of magnitude less expensive than an AP analyzer) audio interface on another machine. To avoid ground loop issues, maybe a USB-powered interface running off a laptop off the mains. Or some external interface powered via isolation transformers. It's hard to get more legitimate noise results without some effort, anyhow.
Please use an audio interface and test input and output of the motherboards separately. This will give a more consistent test environment. Please also test for the nagging noise that can creep in to the audio signal. On some Laptops it is so bad, that you can nearly count by ear the pixels the mouse is traveling ;)
So how much of a difference is there actually between an audio solution like ALC1150 and an external DAC. Also which sound is better.. motherboard or smartphone ?
It's nice that you try, but I'd say: don't bother testing onboard-audio at all!
The reason is pretty simple. The onboard-audio provides good enough sound for the genmeral crowd. Be it for watching movies, playing games or listening to MP3s and audio-streams. Those who really listen to music or like to watch films in a hometheater-setup don't use a PC for playback to start with.
The best suggestion you can make to the majority of people is to use a cheap USB-DAC like the Behringer U202, which only costs some $20 and it'll drastically improve the sound, especially when more advanced headphones are used.
So yeah... don't bother testing onboard-audio, as you'll only get shout at by the audiioophiles while the majority isn't interested at all.
On what grounds are you making the claim that a $20 USB DAC performs better than all, or even most onboard audio codecs? I'll want to see some graphs before I take a statement like that seriously.
In this day and age, who really uses the analog outputs from a computer? A computer is an inherently nasty place for audio. There are plenty of clean digital output options such as SPDIF, USB, HDMI and DisplayPort.
Maybe we need a survey of what users here are running? I have some good headphones and a Logitech 2.1 system on my desktop, both using rear analog audio on an old X58 board (Realtec something). If there is a $20 solution that would be better I'd like to know.
My Xonar STX sounds just as accomplished and accurate as a lot of dedicated external amp/DAC solutions... Although I only got it for the value proposition as an all-in-one solution (headphone amp) and for Dolby Headphone.
If you don't care for the latter then yeah, just go with a USB DAC if using headphones or HDMI/SPDIF if you have an AVR you like and a 5.1 setup (for stereo/2.1 a USB DAC is probably still the bettervalue tho, paired with something like a T-amp if on a budget).
I think most people who are critical of their audio will probably stay away from on board solutions for practical reasons as much as anything... It's hard to predict what you're gonna get with any given mobo, and if you have high end headphones or an AVR then it's kind of irrelevant either way. You're gonna need better amplification or you're gonna pass thru a digital signal.
>ASUS achieve a -108 dB SNR compared to my -101 dB SNR result due to ASUS using Audio Precision testing tools and my test using the Line In + RMAA methodology. The RMAA test also relies on a 1 kHz pulse at -60 dB for its test rather than a range of frequencies, which as a result causes the peak seen at 1 kHz.
I don't see any discrepancy. You're comparing different types of tests. Your RMAA measurement is dynamic range (ratio of a test signal to distortion+noise) while the audio precision test looks like a simple noise floor measurement. In general, the dynamic range is usually less then the noise floor.
Essentially, Asus is choosing a more favorable test to quote numbers from while you are choosing a more realistic one. No surprise really, they want to look good :)
Thank you for your further research in this issue, and I look forward to seeing more refined testing in your audio reviews. Like some of the other commentators have mentioned, please stick to numbered results. And kind of hearing tests should only be icing on the cake, not certifiable evidence, considering how the brain likes to fill in the gaps.
I've recently started looking more and more into audiophile grade equipment. I start off on a site that seems to do a great job explaining how one DAC, amp, headphones, or speaker sounds better than another, then I see them talking about how digital cables can sound different, how SATA and fan cables can make your audio better, how a black box you put your laptop on can improve audio, and that site just loses all credibility. I can't trust people who ignore science and believe in magic.
Being a musician and an audiophile (by necessity, I pay as little as I can to get by) I have to disagree with many of the blanket "snake oil" comments. There is a need for better audio outside of a studio environment. Even on a game mobo audio is distracting; fortunately the Asus STX cards are a good substitute and don't cost a fortune. Just simply going "digital out" isn't as open and shut as it sounds. I wish it wasn't so, believe me.
"Professional and prosumer audio equipment can be honed to perfection and cost an appropriate amount, where discussion on quality over $10k per meter cables can happen."
10,000$ for a cable wouldn't be an appropriate amount, but snake oil. :)
Is there really a point to this anyway? A next to nothing improvement for nearly everybody with an overpriced board on one hand, and 3.5mm jacks on any mobo which will never satisfy the golden ears crowd.
I'll stick with my former recommendation of using an affordable, but standard across all tests soundcard to separately test in- and outputs. And not to test at only a single frequency, but instead try to get at least a certain amount of key pulses to go through.
Currently there's no single fixed point in the measurement approach, no reference, nothing. By going with a reference card, you're of course limited by the quality of its inputs and outputs, but at least you can attempt to get a feel for those, and then obtain a set of comparable results.
Measurement equipment should be more precise than measured device. Typical audio input on mother board is not proper equipment. Asus even doesn't provide audio input specification on their site. But 100dB of dynamic range is a very good value for input. BTW, you seems to mix dynamic range with SNR, which are 2 different measurement in RMAA.
Yes, but it is right next to the output on the motherboard so using that line as input is going to put more stress on the design of the audio components.
You are ultimately right, however, in the sense that it would be better to *both* use the motherboard input to stress test the output while actually doing your analysis using another higher end audio solution. You could simply drive the board's audio in with some test signal to produce a harder test.
I'm only using analog outputs on my Asus mobo for el-cheapo Logitech speakers (for "Windows sounds", basically, and YouTube). Everything else goes to external DACs (USB and S/PDIF) for some serious listening.
I can say you guys are on the right track... as I often wonder why there are no comprehensive reviews out there when it comes to motherboard audio (when we know that recently OEMs are getting better and better in this segment)... Now show me some re-review of the Maximus VI Impact audio sub-system.
I'd like to see an evaluation and comparison of laptop audio. I find the older (I have not tried recent realtek audio chipsets) realtek audio and accompanying headphone amps or associated audio circuitry satisfactory with good headphones ($180 intra-aural and $500 around the ear)) headphones.
I'd like to see an evaluation and comparison of laptop audio. I find the older (I have not tried recent realtek audio chipsets) realtek audio and accompanying headphone amps or associated audio circuitry satisfactory with good headphones ($180 intra-aural and $500 around the ear)) headphones.
Hm, it looks like there's no way to edit posts. However, seeing as how there are numerous earlier mentions involving that post; headphones more negate the need for room acoustic speaker placement, as that does have a large impact on how something sounds, et al. as far as speakers go. The need for an external DAC can be more limited down to an inexpensive surround HT processor, it could have a 1/4" jack for headphones, and even lower powered amplifiers for typical speakers, for <$100 even. Eh, grammer is somewhat of a mess, but hopefully a partial point was understood :)
A very realistic and honest look at audio testing. Ian, this writing has given me confidence that your future audio tests will be really worth reading.
It's worth noting I think that 24/192 audio is good for audio testing, but is literally meaningless for playback. For example, audio with sampling rates above about 44KHz are useless -- not for "most people" or "for most speakers" or "in some expert's opinion" -- I mean mathematically provable that it has exactly zero benefit. Same goes for 24-bit audio (vs 16 bit).
- You should still test at 44.1 or 48 kHz, because that is what everyone uses. Do games even use more than 22 kHz samples now?
- And one very basic test you should add is to connect a pair of speakers with three prong (earth) power-plug. Even very expensive ("durable" class) mainboards in combination with top class power-supplies fail to provide proper grounding on *any* port. That includes onboard audio as much as PCIe, USB and Firewire connected audio solutions. This is not a simple ground loop issue, but one where the mainboards create noise depending on load (especially GPU load) on their ground lines.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
59 Comments
Back to Article
tim851 - Friday, January 31, 2014 - link
> I fall wholeheartedly into the ‘tone deaf’ part of the crowd when it comes to analyzing audio and prefer an easy-to-read number as a comparison, although I understand this is not the be-all and end-all when it comes to audio.<The problem is once you go beyond easy-to-read numbers, audiophiles go crazy and start forming sentences that contain more fluff than an Apple press release. I acknowledge that there is difference in sound quality of across devices, but the willingness of audiophiles to hear unicorns sing is amazing. The snake oil salesmen of old have all switched to audio gear, for sure.
So yeah, stick to numbers.
YukaKun - Friday, January 31, 2014 - link
Well, the fire proof test I have is to use Studio Monitors (Active ones) to test if the sound card is good or not (among other things, of course). They usually detect noise and other artifacts the sound cards produce right away, so you can either cry or give the sigh of relief.That being said. I've managed to nail down that a MoBo can be said to have a decent sound card if it posses at least an ALC892 in it. This chip/codec contains very good sound processing and options to actually enjoy anything (specially from the HTPC perspective). I also have a Creative Sound Blaster X-Fi HD USB and it very worth its price tag, even not being 5.1 or 7.1. If the sound card sucks, I attach this baby and fix everything.
Cheers!
Mr Perfect - Friday, January 31, 2014 - link
I would agree, please stick to objective testing and quantifiable numbers like SNR. Subjective audio testing really can get bizarre. As soon as reviews start commenting on how "warm" or "brown" a sound card is, I'm going to start skipping the audio sections.bim27142 - Monday, February 3, 2014 - link
Honestly, I would prefer subjective audio listening for as long as comparison and description is clear and understandable... Let's say, subjectively the reviewer will compare the audio output of this motherboard's audio to a, say, Asus Xonar STX or even just the DX... of course, quality difference will be tremendous but at least from there, I can have an idea how it really sounds... numbers are just numbers...Sabresiberian - Friday, January 31, 2014 - link
Your analysis of the situation is born out of ignorance and your own bias. Numbers are still an inaccurate substitute for ears. We have a situation here something like the reporting of frames per second as a measure of video quality - it is a factor, but hardly tells the whole picture, as has (finally) been proven recently with the focus on measuring consistency in frames started by The Tech Report. SNR and THD are good numbers to know, but they are just a starting point for evaluating sound quality and a rather small part of the big picture.It is no fault of Anandtech or any hardware site that better analysis isn't done, as it is expensive, time-consuming, and until the last couple of years not all that important in mainboard testing. However, since Asus and others are now claiming they have some kind of better sound solutions on their boards, it is good to have as much information as we can. We can't yet get a total package of measurements that will indicate clearly what we will hear, and a site like this getting involved in subjective impressions (meaning using their ears to measure information they can pout in a review) is difficult, so reporting the numbers that are possible is a good thing, but hardly paints a complete picture.
I'm sure there is plenty of "snake oil" in the audio business, as there is in all businesses, but it isn't to the degree you indicate. People should buy with their ears and their budget, and if an inexpensive on-board sound solution works for you, good news, you can save a lot of money. Others have better ears and want to get more out of their listening experiences, sneering at them because you can't hear what they do is childish at best.
argosreality - Friday, January 31, 2014 - link
There's no if's, and's or buts about snake oil in the audio business - it flat out exists. Numbers don't like (but can be misreported or misrecorded as this article indicates) but ears...ears *DO* lie. They lie for a number of reasons. Genetics, age but even more...what you want to hear. Spend 10k on those speakers? You're going to damn well believe they sound amazing. that 1k wooden nob that "isolates" the sound of your receivers volume? Yea, you want it sound good.It also boils down to personal preference. Some people want the most accurate sound reproduction possible. Others want a warm, natural sound. Some just want booming bass. To each of those, their "ears" will find issues. The numbers, the science? They'll show the truth. So, don't let your own ignorance and bias color YOUR responses
JlHADJOE - Friday, January 31, 2014 - link
IMO what really needs to be done is to find a way to quantify these aspects of audio that are currently hidden behind fancy words. Aside from SNR and THD+N, things like "attack" can actually be seen by plotting the square wave response of a DAC or amplifier.So that's one snake-oil buzzword down, now what we need to do is try and get the rest of them.
We can render very detailed plots from frequency/levels using raw digital files. Now run the same file through the audio equipment, capture the output and see how the new plot compares to the pure digital original and we have an objective way of seeing exactly how the equipment changes the original waveform, and slowly, but almost certainly we can begin to find what quantifiable differences there are (or the lack thereof) when an audiophile reviewer says something like "airy", or "forward".
ThreeDee912 - Saturday, February 1, 2014 - link
There's a good intro to DACs and digital signals from a guy at Xiph, which are the same guys that write the FLAC audio codec. It addresses some common myths about digital audio, like "stairstep" samples, bit-depth, and dithering. Even just from this ~24 mins of video, you can see a lot of silly audio rumors still floating around online today are absolutely false.http://xiph.org/video/vid2.shtml (needs Chrome or Firefox)
TruePath - Saturday, February 1, 2014 - link
It's a little more complicated than that. After all the original was no doubt recorded (then post-processed) in a studio. You are playing that audio back in a very different environment with a different speaker setup and most likely without the kind of professional quality microphone used in the studio.So, sure, it's that simple for the studio to evaluate and tweak their audio setup. Simply use a high-quality mic and make sure the original sound matches the playback as closely as possible at that microphone (repeat with the mic at other points you care about playback quality).
But since you don't have the original room nor the original mic this begs the question of WHAT sound you should be reproducing. Different studios are likely to take the exact same original and capture it in a slightly different manner depending on the type and number of microphones (where it's sensitive) they use and other factors. So how do you even define a standard against which your sound can be compared?
I tend to think the best way is to insist on binaural recordings (recordings made by mics placed on each ear of a dummy human head) and standardize on a particular kind of mic that should be used to compare against the original. Of course, how you weight different distortions is going to make a bunch of difference but at least you have a clearly defined notion of perfect reproduction to start from.
CalaverasGrande - Thursday, February 6, 2014 - link
Attack can be quantified by slew rate and rise time.I'm happy you brought that up. I have a nice little home studio based around a relatively ancient Motu 828MKII firewire soundcard. These boxes have reasonable sound quality, and work really well with Macs.
Mine is not stock, it has had the audio opamps and digital clocking upgraded to tighter specs.
Ironically it actually has slightly worse SNR now. However the THD+N is much lower, and the bandwidth gained an octave on either side.
There is nothing subjective about things rattling off your desk, or being able to record frequencies higher than I can hear.
This brings up my next point. A lot of mobos have optical outs. Those are neat, but they are next to useless if the signal is so jittery that it sounds worse than the analog signal, or worse, is so jittery it can't communicate the audio to your AV device.
TruePath - Saturday, February 1, 2014 - link
This business about wanting a warm sound and the like made sense when we were stuck in the bad old days of analog audio but it shouldn't persist.Deliberately choosing an audio component that distorts the sound being reproduced only ensures that you can never get exactly the sound you want. Sure, it might make the sound seem more warm but when you want to play something where the input already captures that warm sound you still get that distortion on top of the already warm audio you are playing.
Now that audio files are all digital instead of choosing an audio solution less capable of reproducing the recorded sound the most accurately (I know that's a bit vague) simply reprocessing the sounds in software should suffice to imbue them with whatever effect you desire. This then would allow you to take the same audio files from one device to another and expect them to all sound very very similar.
Yes, I know that if you want your in game sound to be warm this isn't really an option but really, how good can the game be, if you are paying attention to the fact that the background music isn't as warm as you usually prefer. Ideally, you could tune these parameters in hardware (and at the high end I'm sure you can) so you could get the effect even in games.
TruePath - Saturday, February 1, 2014 - link
Numbers can't be an inaccurate substitute for ears. Our machines, which represent the audio in numbers, are far far more sensitive to changes in pressure (sound) than the human ear.I don't doubt that some people with sensitive ears get something out of some reasonably subtle effects. However, just as with wine when double blind studies are done many people are not able to reliably identify which of the experiences is better despite claiming it was clear as day when they knew what was what.
The suggestion that you just go into the store and see what you like is a recipe for wasting money. You know that people are easily subject to placebo effects here and yet you place yourself in a situation where a friendly sales guy can subtly maneuver you using your desire to appear sophisticated and knowledgeable into convincing yourself that, why yes, you can tell the difference between the cheap junk unit and the fancy one.
Also, as I've already suggested there is absolutely no need to go out and buy an audio unit tuned to your tastes. It should be possible to simply buy the one that objective analysis indicates is best able (using taste only to decide what kind of distortions matter more) to reproduce the widest range of sound at your ears and simply modify the audio in software so that the actual sound you hear is whatever you prefer.
ThreeDee912 - Friday, January 31, 2014 - link
Unfortunately while a good portion of audio does involve actual science, there is a large amount of pseudo-science, subjective opinion, and even placebo effect involved. A couple years ago, there was a bit of a debate about lossless 24-bit 192kHz audio being better than normal 16-bit 44.1 kHz, and how people with "golden ears" or "better ears" could clearly hear the larger range of frequencies. Then some people at Xiph, who develop the popular FLAC audio codec, stepped in and showed how they were all being silly. I particularly like how they compared these audiophiles to hypothetical "spectrophiles" that claim they can see X-rays and microwaves everywhere. It just doesn't happen.http://xiph.org/~xiphmont/demo/neil-young.html
Even earlier than that, years before iTunes came around, there were 2 competing MP3 player programs for Mac, Audion and SoundJam. Despite both of them implementing the MP3 decoder the same way, numerous audiophile review websites and magazines all claimed that Audion somehow sounded "better" than SoundJam, even though the Audion devs weren't doing anything special. It's also mentioned in an article about the history of Audion and iTunes on the dev's site.
https://www.panic.com/extras/audionstory/
sheh - Saturday, February 1, 2014 - link
I'm not familiar with these players, but MP3 decoders aren't necessarily created the same (e.g., dithering, among other things), and the rest of the playback chain can be variable as well.Jedibeeftrix - Tuesday, February 11, 2014 - link
brilliant comment. +1 for you, sir.davidedney123 - Friday, January 31, 2014 - link
I really feel all these advanced audio options on motherboards are pure marketing. If you give a damn about audio quality from your PC you use a digital output and an external DAC, and these days there are lots of very cheap options available. All of the "audiophile" components in the world are not going to give you good results if they are located in the EM quagmire inside a PC.willis936 - Friday, January 31, 2014 - link
While it's a hard pill to swallow with all of the marketing fluff that surrounds audio and the even more flamboyant marketing around PC OEMs and motherboards there is good engineering sense behind the things they do. "THE RED LINE". Cheesy as hell but having a separate ground plane helps and paying close attention to the ground path of every component is absolutely necessary in high fidelity audio circuits. There's no shortage of reading on the characteristic of caps and in some places the quality of the cap won't affect performance but in other places close to the signal path the can. Nichicon, WIMA, Nippon, etc. are all more expensive but also _may_ result in better performance.The marketing gimmicks alone don't make a good product though. The implimentation (ie the layout) is much more important than the components chosen. If you pull up the spec sheet for any one of these codecs you'll see absurd SNR and dynamic range numbers. Expecting those out of your speakers is like expecting to get 1.9Gbps out of your fancy 11ac router. It's just not going to happen for practical reasons. The quality of the tiny circuit around the codec is going to be largest limiting factor in most cases.
So while it may seem like marketing trash there's a good reason attention is being paid to this part of the board. There is absolutely no reason at this day and age to spend $200 on a dedicated sound card unless you are working in a professional studio. Even these fancy on board sound cards are good enough for amateur recording given a $100 mic.
thewhat - Friday, January 31, 2014 - link
I appreciate the RMAA tests that you're doing. Even if it has it's limits, it's a very useful tool, IMO.But it seems that, when you're testing, you're always using the same onboard soundcard for the input. As you also noted, this means that the input performance can "bottleneck" the results. But I'm guessing the majority of readers don't use the line-in at all and would rather just see the clear line-out performance and the headphone performance (you can test the latter with a 3.5mm splitter and a headphone load).
RMAA lets you use different in/out devices, so why not get a good quality dedicated soundcard that has a verified good quality input with high SNR and use that for testing the output on the motherboards? (And you could also use that to test the input performance independently.)
mr_tawan - Friday, January 31, 2014 - link
I second that. You could get a good USB audio interface to do the test. This way even though the bottleneck would be possed on the audio interface, at least the test runs with consistency across test subject (mainboard).The problem would be to get the proper audio interface to work with.
PS. Audio Interface is another name of the device we call sound card, usually used for professional-level AFAIK.
bludragon - Friday, January 31, 2014 - link
THIS. At least then you keep the variable quality of the input fixed when comparing outputs and you can also run things the other way - have the external reference provide the output to test the input on the motherboard. There are plenty of "pro" audio usb options you could use for this for not too much $$. e.g. E-MU 0204-USBTruePath - Saturday, February 1, 2014 - link
Hmm, the claim in the article is that RM:AA is actually much more sensitive to poor implementations when you use the same hardware for both input and output.In particular, by feeding back the same input into an output on the same device you can sometimes create feedback effects from any signal leakage while detecting the same leakage merely by analyzing the audio out would require a very very sensitive instrument because you've moved the potential interference both closer to the output (EM leakage is more likely from the input that from random computer parts) and can amplify that weakness to a more noticeable size by creating the appropriate kind of feedback in software.
I understood that this was the argument for using the same input and output. Sure, ideally you would have a very sensitive input that lets you directly examine the output for bias (including using feedback) but lacking that you instead choose to increase the pressure on the output by running the most problematic kind of signal right next to it.
thewhat - Friday, January 31, 2014 - link
Also, one thing I noticed with some onboard chips is that they start making noise with CPU/GPU load (sometimes just scrolling in some window), so it might make sense to try to simulate that and see if it affects the results. Perhaps simply running Furmark during the test would do it...ShieTar - Friday, January 31, 2014 - link
See the comment from davidedney above, there is a good likelyhood that you are noticing EM emissions from the CPU/GPU or power supply components connected to it being induced into the analog part of the audio solution. Testing this is far from trivial, as it does not only depend on the Mainboard itself, but also on the type of CPU, GPU and PSU; as well as the case geometry and internal reflectivities for the frequencies in question.kwrzesien - Friday, January 31, 2014 - link
It may be far from trivial, but this is exactly the kind of functional testing I'm looking for in a review here at AnandTech. I want to know if the solution *works* and in what ways - headphones, line out, optical out, front connection on a case (maybe this should be a part of the case tests too). What headphones can I buy that are supposed to work, and do they? Will it hum, crackle or pop? How does it work connected to an AVR?YoshoMasaki - Tuesday, February 4, 2014 - link
Seriously. All I want is to stop hearing random FM radio stations through my dang speakers already :-\Braumin - Friday, January 31, 2014 - link
As a reviewer, I think it's best to stick to the numbers. I think we all appreciate that you are trying to add some additional info to your testing. I think consistent testing is the key if you can't get the absolute top end testing equipment.Audio is a tricky thing for sure, and it has a certain, well let's call them devout group of followers.
Exodite - Friday, January 31, 2014 - link
Thanks for this.I'll add my voice to those who prefer that you stick strictly to numbers, as trying to judge something as nebulous as audio "quality" is bound to backfire.
Justin TenCate - Friday, January 31, 2014 - link
For me, the most important aspects of any PC audio test are noise threshold and audible interference. I can't tell you how annoying it is to spend 20+ hours recording a track to have it mangled and mashed into some unrecognizable combination of squeals, beep, ticks and hisses by cheap MoBo audio on a "premium" board. If I am paying extra for enthusiast gear, I expect the audio to be accurate, distortion and interference free at the very least.sheh - Saturday, February 1, 2014 - link
If you're recording for hours, and mobo quality isn't satisfactory or consistent, just get a standalone or external sound card.Aizen99 - Friday, January 31, 2014 - link
I'd be more interested in you guys exploring sound editing software and text to speech then hardware testing audio chips that have been more then good enough.For instance one of the selling points for me getting Creative soundcards over the competition in ages past was the ability to record audio from wave output. In ages past you couldn't record audio from a wave device that was playing it at the same time internally without resorting to round about methods using the analog out into the mic/line in. It was a royal pain in the butt.
lever_age - Friday, January 31, 2014 - link
AP is certainly very respected, but they're not the only player in town. More so if you don't need super top-of-the-line performance and features.There's the dScope Series III and R&S UPV, off the top of my head, but those are relatively expensive as well. There are cheaper analyzers too, but I don't think it's particularly that worth it for a site like AnandTech.
One thing perhaps worth considering is still using RMAA but being more careful about actually physically measuring output levels for consistency and so on. Also, rather than loopback, you could record with a high-quality (and by this I mean a couple orders of magnitude less expensive than an AP analyzer) audio interface on another machine. To avoid ground loop issues, maybe a USB-powered interface running off a laptop off the mains. Or some external interface powered via isolation transformers. It's hard to get more legitimate noise results without some effort, anyhow.
slalomsk8er - Friday, January 31, 2014 - link
Please use an audio interface and test input and output of the motherboards separately.This will give a more consistent test environment. Please also test for the nagging noise that can creep in to the audio signal. On some Laptops it is so bad, that you can nearly count by ear the pixels the mouse is traveling ;)
vortexmak - Friday, January 31, 2014 - link
So how much of a difference is there actually between an audio solution like ALC1150 and an external DAC. Also which sound is better.. motherboard or smartphone ?jrs77 - Friday, January 31, 2014 - link
It's nice that you try, but I'd say: don't bother testing onboard-audio at all!The reason is pretty simple. The onboard-audio provides good enough sound for the genmeral crowd. Be it for watching movies, playing games or listening to MP3s and audio-streams.
Those who really listen to music or like to watch films in a hometheater-setup don't use a PC for playback to start with.
The best suggestion you can make to the majority of people is to use a cheap USB-DAC like the Behringer U202, which only costs some $20 and it'll drastically improve the sound, especially when more advanced headphones are used.
So yeah... don't bother testing onboard-audio, as you'll only get shout at by the audiioophiles while the majority isn't interested at all.
willis936 - Saturday, February 1, 2014 - link
On what grounds are you making the claim that a $20 USB DAC performs better than all, or even most onboard audio codecs? I'll want to see some graphs before I take a statement like that seriously.OoklaTheMok - Friday, January 31, 2014 - link
In this day and age, who really uses the analog outputs from a computer? A computer is an inherently nasty place for audio. There are plenty of clean digital output options such as SPDIF, USB, HDMI and DisplayPort.kwrzesien - Friday, January 31, 2014 - link
Maybe we need a survey of what users here are running? I have some good headphones and a Logitech 2.1 system on my desktop, both using rear analog audio on an old X58 board (Realtec something). If there is a $20 solution that would be better I'd like to know.jrs77 - Friday, January 31, 2014 - link
A Behringer U202 USB-DAC costs only $20, like I posted above.sheh - Saturday, February 1, 2014 - link
Would be nice if all amplifiers had some sort of digital input (other than SPDIF, which is limited).Impulses - Sunday, February 2, 2014 - link
My Xonar STX sounds just as accomplished and accurate as a lot of dedicated external amp/DAC solutions... Although I only got it for the value proposition as an all-in-one solution (headphone amp) and for Dolby Headphone.If you don't care for the latter then yeah, just go with a USB DAC if using headphones or HDMI/SPDIF if you have an AVR you like and a 5.1 setup (for stereo/2.1 a USB DAC is probably still the bettervalue tho, paired with something like a T-amp if on a budget).
I think most people who are critical of their audio will probably stay away from on board solutions for practical reasons as much as anything... It's hard to predict what you're gonna get with any given mobo, and if you have high end headphones or an AVR then it's kind of irrelevant either way. You're gonna need better amplification or you're gonna pass thru a digital signal.
saratoga3 - Friday, January 31, 2014 - link
>ASUS achieve a -108 dB SNR compared to my -101 dB SNR result due to ASUS using Audio Precision testing tools and my test using the Line In + RMAA methodology. The RMAA test also relies on a 1 kHz pulse at -60 dB for its test rather than a range of frequencies, which as a result causes the peak seen at 1 kHz.I don't see any discrepancy. You're comparing different types of tests. Your RMAA measurement is dynamic range (ratio of a test signal to distortion+noise) while the audio precision test looks like a simple noise floor measurement. In general, the dynamic range is usually less then the noise floor.
Essentially, Asus is choosing a more favorable test to quote numbers from while you are choosing a more realistic one. No surprise really, they want to look good :)
Rajinder Gill - Friday, January 31, 2014 - link
The ASUS test is dynamic range - you just aren't seeing the whole screenshot :)Rajinder Gill - Friday, January 31, 2014 - link
Sweep data:C:\Apwin\VISTA\DA\WL-Dynamic Range.at2c, 09/06/13 16:05:27
DGen.Freq, Anlr.THD+N Ampl, Anlr.THD+N Ampl, ,,,,,
Source 1, Data 1, Data 2, ,,,,,
Hz, dBr A, dBr B, ,,,,,
20000.0009537, -115.647123142, -115.36093713, ,,,,,
15886.5652084, -115.935569993, -115.649383982, ,,,,,
12619.1482544, -115.927113829, -115.640927817, ,,,,,
10023.7455368, -116.310144004, -116.023957992, ,,,,,
7962.14389801, -116.482618551, -116.196432539, ,,,,,
6324.55444336, -116.621886408, -116.335700397, ,,,,,
5023.77319336, -116.512704794, -116.226518782, ,,,,,
3990.52333832, -116.387036958, -116.100850946, ,,,,,
3169.78740692, -116.236844922, -115.95065891, ,,,,,
2517.85182953, -115.960987994, -115.674801983, ,,,,,
2000.00095367, -115.848584101, -115.56239809, ,,,,,
1588.65737915, -115.533297062, -115.247111051, ,,,,,
1261.91425323, -115.412973441, -115.126787429, ,,,,,
1002.37369537, -115.315274306, -115.029088294, ,,,,,
796.214103699, -115.273364021, -114.98717801, ,,,,,
632.454872131, -115.161721996, -114.875535984, ,,,,,
502.378463745, -115.179798775, -114.893612764, ,,,,,
399.052619934, -115.013372807, -114.727186795, ,,,,,
316.97845459, -115.123112258, -114.836926247, ,,,,,
251.784324646, -115.10771629, -114.821530279, ,,,,,
199.999809265, -114.993108255, -114.706922243, ,,,,,
158.866882324, -114.925062347, -114.638876336, ,,,,,
126.191139221, -115.19014528, -114.903959269, ,,,,,
100.238800049, -115.09490713, -114.808721119, ,,,,,
79.6222686768, -115.236857676, -114.950671665, ,,,,,
63.2457733154, -115.125680908, -114.839494896, ,,,,,
50.2367019653, -115.156564104, -114.870378092, ,,,,,
39.9055480957, -115.117977236, -114.831791224, ,,,,,
31.6972732544, -115.265528306, -114.979342294, ,,,,,
25.1798629761, -115.268139426, -114.981953414, ,,,,,
20.001411438, -115.375876845, -115.089690834, ,,,,,
test page image came from: http://imageshack.com/a/img571/1842/3ej1.png
saratoga3 - Friday, January 31, 2014 - link
What is the test signal amplitude?AggressorPrime - Friday, January 31, 2014 - link
Thank you for your further research in this issue, and I look forward to seeing more refined testing in your audio reviews. Like some of the other commentators have mentioned, please stick to numbered results. And kind of hearing tests should only be icing on the cake, not certifiable evidence, considering how the brain likes to fill in the gaps.I've recently started looking more and more into audiophile grade equipment. I start off on a site that seems to do a great job explaining how one DAC, amp, headphones, or speaker sounds better than another, then I see them talking about how digital cables can sound different, how SATA and fan cables can make your audio better, how a black box you put your laptop on can improve audio, and that site just loses all credibility. I can't trust people who ignore science and believe in magic.
TheRealAnalogkid - Saturday, February 1, 2014 - link
Being a musician and an audiophile (by necessity, I pay as little as I can to get by) I have to disagree with many of the blanket "snake oil" comments. There is a need for better audio outside of a studio environment. Even on a game mobo audio is distracting; fortunately the Asus STX cards are a good substitute and don't cost a fortune. Just simply going "digital out" isn't as open and shut as it sounds. I wish it wasn't so, believe me.sheh - Saturday, February 1, 2014 - link
"Professional and prosumer audio equipment can be honed to perfection and cost an appropriate amount, where discussion on quality over $10k per meter cables can happen."10,000$ for a cable wouldn't be an appropriate amount, but snake oil. :)
StrangerGuy - Saturday, February 1, 2014 - link
Is there really a point to this anyway? A next to nothing improvement for nearly everybody with an overpriced board on one hand, and 3.5mm jacks on any mobo which will never satisfy the golden ears crowd.Rick83 - Saturday, February 1, 2014 - link
I'll stick with my former recommendation of using an affordable, but standard across all tests soundcard to separately test in- and outputs.And not to test at only a single frequency, but instead try to get at least a certain amount of key pulses to go through.
Currently there's no single fixed point in the measurement approach, no reference, nothing. By going with a reference card, you're of course limited by the quality of its inputs and outputs, but at least you can attempt to get a feel for those, and then obtain a set of comparable results.
popej - Saturday, February 1, 2014 - link
Measurement equipment should be more precise than measured device. Typical audio input on mother board is not proper equipment. Asus even doesn't provide audio input specification on their site. But 100dB of dynamic range is a very good value for input. BTW, you seems to mix dynamic range with SNR, which are 2 different measurement in RMAA.TruePath - Saturday, February 1, 2014 - link
Yes, but it is right next to the output on the motherboard so using that line as input is going to put more stress on the design of the audio components.You are ultimately right, however, in the sense that it would be better to *both* use the motherboard input to stress test the output while actually doing your analysis using another higher end audio solution. You could simply drive the board's audio in with some test signal to produce a harder test.
madwolfa - Sunday, February 2, 2014 - link
I'm only using analog outputs on my Asus mobo for el-cheapo Logitech speakers (for "Windows sounds", basically, and YouTube). Everything else goes to external DACs (USB and S/PDIF) for some serious listening.bim27142 - Monday, February 3, 2014 - link
I can say you guys are on the right track... as I often wonder why there are no comprehensive reviews out there when it comes to motherboard audio (when we know that recently OEMs are getting better and better in this segment)... Now show me some re-review of the Maximus VI Impact audio sub-system.Theresa N - Monday, February 3, 2014 - link
I'd like to see an evaluation and comparison of laptop audio. I find the older (I have not tried recent realtek audio chipsets) realtek audio and accompanying headphone amps or associated audio circuitry satisfactory with good headphones ($180 intra-aural and $500 around the ear)) headphones.Theresa N - Monday, February 3, 2014 - link
I'd like to see an evaluation and comparison of laptop audio. I find the older (I have not tried recent realtek audio chipsets) realtek audio and accompanying headphone amps or associated audio circuitry satisfactory with good headphones ($180 intra-aural and $500 around the ear)) headphones.choirbass - Sunday, February 9, 2014 - link
External DAC's really should be a solid consideration regarding audio.choirbass - Monday, February 10, 2014 - link
Hm, it looks like there's no way to edit posts. However, seeing as how there are numerous earlier mentions involving that post; headphones more negate the need for room acoustic speaker placement, as that does have a large impact on how something sounds, et al. as far as speakers go. The need for an external DAC can be more limited down to an inexpensive surround HT processor, it could have a 1/4" jack for headphones, and even lower powered amplifiers for typical speakers, for <$100 even. Eh, grammer is somewhat of a mess, but hopefully a partial point was understood :)Sivar - Tuesday, February 11, 2014 - link
A very realistic and honest look at audio testing. Ian, this writing has given me confidence that your future audio tests will be really worth reading.It's worth noting I think that 24/192 audio is good for audio testing, but is literally meaningless for playback. For example, audio with sampling rates above about 44KHz are useless -- not for "most people" or "for most speakers" or "in some expert's opinion" -- I mean mathematically provable that it has exactly zero benefit. Same goes for 24-bit audio (vs 16 bit).
Readers see http://people.xiph.org/~xiphmont/demo/neil-young.h... or read about the Nyquist–Shannon sampling theorem.
Timur Born - Tuesday, April 15, 2014 - link
Thanks for the article! Two opinions from me:- You should still test at 44.1 or 48 kHz, because that is what everyone uses. Do games even use more than 22 kHz samples now?
- And one very basic test you should add is to connect a pair of speakers with three prong (earth) power-plug. Even very expensive ("durable" class) mainboards in combination with top class power-supplies fail to provide proper grounding on *any* port. That includes onboard audio as much as PCIe, USB and Firewire connected audio solutions. This is not a simple ground loop issue, but one where the mainboards create noise depending on load (especially GPU load) on their ground lines.