Do media players have a sound?

[Vincent Kars]

What is a media player.
It is a piece of software allowing you to navigate your collection, build play list, tag, rip etc.
But of course its primary function is to play the songs you have selected.

There are many different audio formats.
You need a Codec for each format.
There are various codes to uncompress MP3.
Although today's reports about differences most of all report no differences, it might be a source of a differences in sound.
Ok, let's play WAV or FLAC.
WAV is already PCM and for FLAC each simply use the same library as supplied by the developer.
This should eliminated possible differences due to Codecs.

The media player sends the audio to a sound card (or SPDIF out).
It might apply DSP like equalization, volume control, etc.
The quality of the DSP like algorithms used and the precision (16, 32 or 64 bits) will have an impact on the sound.
Ok, disable all enhancements, set volume to 100% in other words, bit perfect output.

If the player resample, this might again affect sound quality.
Ok play everything at its native sample rate.

Having eliminated any possible meddling with the bits, will they still sound different?

If you play 24 bits sources but your DAC accept only 16 bits, the media player will trunk and might apply dither.
Ok we play only sources with a bit depth accepted natively by the DAC.

What about the drivers?
If one media player uses Direct Sound and the other WASAPI, this is obvious not a valid comparison as the drivers differ.
Ok, set both to WASAPI

Have we eliminated all possible differences?
Are both players delivering bit perfect output?
If yes, why and how could they sound different?
If they simply use the same codec and handle the output to the same driver they should sound the same.

Barry Diament did some testing (http://www.computeraudiophile.com/content/first-test-and-some-food-audio-thought)
Recording the digital out when playing the same recording at its native sample rate using iTunes and another application. The bits where identical.
Still a lot report to hear a difference using the same apps.
The usual suspects?
- expectation bias
- improper configuration
- our lender of last resort: jitter?

 

[garylkoh]

This is something that has puzzled me as well. Once identically set up, how can two different players sound different? But I think that they do.

I set up J River and Foobar on my music server and they sound different. Foobar was noticeably cleaner - even the Foobar website says that Foobar should not sound better than other correctly set-up players. I thought that it might be jitter, but then I added a sound spectrum display in Foobar - that should eat up CPU resources and make it sound different, but it didn't. I hoping that somebody else here has some ideas.

 

[Phelonious Ponk]

Cheat Vincent! You're supposed to answer the question, not ask it!

But you'll never get one answer, and I'd be honored to step in and start the opposition: I've tried both Amarra and Pure Audio in my systems. Playing the same files at the same rates, ad infinitum etc., they sounded just like iTunes to me. But, of course, I expected them to sound the same, so consider expectation bias. But I'm skeptical not just because there is no good reason for there to be a difference, but because so many reporting these differences say a) they are obvious, not subtle, or b) subtle, but only audible on a sufficiently resolving system (ie: You can't hear it because your gear's not good enough) and c) the differences always seem to be heard in sound stage, micro dynamics, air around the instruments, musicality....ie: the same descriptors used to talk about every unmeasurable audiophile phenomenon ever recorded on a discussion board.

I vote for expectation bias. Or wishful thinking. A significant upgrade that doesn't require the expense of new speakers or the magic of getting room treatments past the wife? I could almost talk myself into that.

Tim

 

[Vincent Kars]

Some month ago I compared J River and MusicBee.
J River has its own 64 bit audio library; MusicBee uses the BASS lib by Un4seen.
Played FLAC, volume 100%, etc.
So I do think I had them both ‘bit perfect’, same drivers, etc.
Both using WASAPI.

I had the feeling they do sound slightly different, MusicBee a bit ‘lighter’ whatever this might be (no poet lost in my case).
As one has to close the aps (WASAPI exclusive) this was not exactly an unsighted test.
As the difference was so subtle I was not in need of an unsighted test, I would simply fail it. In the end, although I do think the interface of MusicBee is a very good one, I stick to JRMC, its many features and excellent sound makes it my favorite player.

Personally I do think drivers (e.g. WASAPI or DS) do make a far more substantial difference than the players (when bit perfect etc).

I did the same experiment as Gary but involuntary.
I was listening to music one day and when it stopped, there was some noise of the HD.
There was a backup running in the background!
Of course I didn’t notice the difference sound quality wise.

 

[Scott Borduin]

Obviously, there is a lot of scope for expectation bias here, and given the fundamentals in play here, I think that almost all reported incidences of "dramatic" differences between two bit-perfect playback stacks would have to be attributed to expectation bias. That said ...

I think a very important point to make here is that any sonic differences between two bit-perfect software stacks are not, as a first-order phenomenon, caused by software at all. Let's think, for a minute, about a high-res recording downloaded from a site like HDTracks. Just to get from that site to your computer, that file will be broken up into chunks (packets) which will pass through several different types of digital connections, multiple pieces of server and router hardware, each with their own software, then finally through your local area network, high and low level system software and drivers, and thence to your hard drive. And yet, the data in that file is guaranteed to be perfect to essentially infinite certainty (if that weren't true, the entire modern world would collapse pretty much instantaneously).

Now, into the picture comes our playback software. Assuming we've set things up to be bit-perfect, in comparison to the job done by all of that internet and network infrastructure, the job of our playback software is utterly trivial - transfer the data unaltered from our hard disk to the memory buffer of the sound card, on demand. The software does not change the bits, it doesn't clock the data. Conceptually, there is only one difference between the playback software and all the other software in the network infrastructure: the playback software is executing on a computer which is connected to the rest of the playback system. The unavoidable implication is: any "distortion" or imperfection related to our digital transport must arise from the connection between the computer and the playback system.

There are two broad categories of imperfection which could be related to this connection. One is "jitter" - imperfection in the shape of the digital waveform, which in turn could arise from multiple causes like clocking, modulation with fixed frequency noise, etc. The other category is electrical "noise" being transmitted to the rest of the system through interconnects or the A/C power lines. You will note: both of these imperfections are caused first of all by hardware.

Now yes, it is true that different software will exhibit different patterns of usage of hardware resources. And it is true that there are quite a number of electrically "noisy" hardware elements in a computer - power supply, CPU, memory, bus, hard drive, video driver, etc. And it is true that consumer level computers are not designed to isolate low levels of electrical noise between subsystems. And it is true that the jitter performance of a sound card can potentially be impacted by electrical noise from other sources in the system. So different usage patterns of computer resources can lead to different behavior with respect to jitter or electrical noise. How significant any of these effects might be is very much open to debate, and there has been very little attempt to measure or otherwise test for them, but they are certainly within the realm of theoretical possibility.

However, it's important to recognize that every computer is different, with a different composition of hardware elements, hardware drivers, power supply, case, A/C line and conditioners, and interactions between those elements. Given the very high number of combinatorial possibilities, it is simply not possible to design a generic piece of software which will minimize hardware issues on all systems - and any assertions about the "sound quality" of a particular piece of bit perfect software must take this into account. And BTW, that obviously goes for operating systems as well.

Beyond that, from a system design perspective, starting with inadequate piece of hardware and then attempting to indirectly patch its problems through software is an indirect, error prone, and fundamentally compromised way to approach the problem. Starting with correctly designed hardware is the obvious answer, and it is clearly achievable. After all, a high-end CD player is nothing more than a specialized audio computer, containing many of the same types of hardware as a PC, and audio designers have long since learned to appropriately isolate the subsystems in CD players.

So to be clear, and probably provocative , if you are hearing differences between two different bit-perfect software implementations, and those differences are not imaginary, then you have a hardware problem. And your first order focus ought to be on fixing the hardware, not investing a great deal of time in software optimization that could be invalid the next time you switch hardware.

 

[amirm]

That is a very well-written piece Scott. I wrote something with the same information in mind and deleted it twice because I couldn't get the message across.

I would just add this: maybe you don't want to worry about these low order effects. Get your system and enjoy it! Even though I sweat these details too, I just don't worry about them much in practice. Use the player which gives you the best interface to your music and that will provide more enjoyment than this last bit of hazy optimization .

 

[Scott Borduin]

That is a very well-written piece Scott. I wrote something with the same information in mind and deleted it twice because I couldn't get the message across.

I would just add this: maybe you don't want to worry about these low order effects. Get your system and enjoy it! Even though I sweat these details two, I just don't worry about them much in practice. Use the player which gives you the best interface to your music and that will provide more enjoyment than this last bit of hazy optimization .

100% true. By far the most important aspect of a music server system, IMO, is the way a well set-up system facilitates the rediscovery and pure enjoyment of your music collection. Using a lousy, tweaky interface just because you think it might give you a microscopic improvement in SQ defeats the whole purpose.

Of course, if you're the type who enjoys cleaning LPs and fiddling with VTA, maybe a bad interface will help fill your need for pain and deep involvement

 

[Vincent Kars]

Wondering why media players do sound different and enjoying the music is not necessary a contradictio in terminis.
In the end it is pretty much like traditional audio.
You have those who listen to the music and those who listen to components.

If it is about the music, rediscovering you collection is certainly one of the benefits.
Fooling around a little with different players is an excellent way to discover which one suites your style best.
There are a lot of things you can simply try at zero or marginal expense (except time…) and check if they improve the sound or the end user experience.

And if you are into components, forget about things like swapping cables.
Here you have a box with a myriad of parameters and you can try them all!
A tweakers paradise.

Having you audio in a computer is simple combining the best of both worlds.
The comfort of a good GUI and a good sound quality.

 

[Phelonious Ponk]

I get all of this in theory, but it doesn't come through in practice. My system comes out of a MacBook Pro's usb port into a little box that re-clocks then sends digital coax to my headphone system and optical to my speaker system. There's a bit more going on before it gets to my ears of course, but to keep this simple, let's just assume that the coax, being electrically connected to the computer, should be the one most effected by computer system activity and least effected by jitter, and the headphone system, being the most revealing for a whole lot of reasons, should make any differences more than audible enough. That being the case, when I'm sitting back in my recliner, with everything turned off but iTunes, the screen darkened, running off of battery power, all should be well. Very well. And when I'm sitting at my desk listening to the same headphone system with Safari and Mail and Vudu running all at once, receiving emails, telling you guys how great the best midfi is, and downloading a hi-res movie file, the sound quality should be dramatically different.

It's not.

In fact, if I could A/B between the two scenarios quickly (which I obviously can't), I seriously doubt I could differentiate between them. So when people on the net say they can hear unsubtle differences between iTunes and Pure Music (or Amarra or whatever...) on the same system under basically the same resource load, I roll my eyes. I feel a bit bad about it, but I roll my eyes.

Tim

 

[Vincent Kars]

Hmmm, another cheater…

Isn’t the purpose of re-clocking to remove jitter?
Maybe transformer isolated too?

 

[Scott Borduin]

And when I'm sitting at my desk listening to the same headphone system with Safari and Mail and Vudu running all at once, receiving emails, telling you guys how great the best midfi is, and downloading a hi-res movie file, the sound quality should be dramatically different.

It's not.

In fact, if I could A/B between the two scenarios quickly (which I obviously can't), I seriously doubt I could differentiate between them. So when people on the net say they can hear unsubtle differences between iTunes and Pure Music (or Amarra or whatever...) on the same system under basically the same resource load, I roll my eyes. I feel a bit bad about it, but I roll my eyes.

Tim

Actually, I'd expect any audible differences under the different conditions you mention to be extremely subtle at most, and probably nonexistent. I've tried almost every piece of playback software out there, on multiple operating systems (Mac and Windows), feeding my AlphaDAC/Spectral/Magico system. While there are major differences in UI, I've never heard an audible difference I could ascribe to software in my system, or for that matter to execution of other software on the system.

In order for the differences to be audible, I suspect the conditions would have to be "perfect": a DAC with poor input jitter rejection, shared A/C ground with high resistance to earth, poor galvanic isolation on SPDIF output/input, software which just happened to hit the weak point of the hardware, etc. But that's just speculation; as I said above, there is essentially no data supporting or disproving the theoretical possibilities I mentioned.

 

[Vincent Kars]

ScottB: ever tried DS versus WASAPI?

Tim: is it a reclocker or a protocol converter you are using?

 

[Scott Borduin]

Hmmm, another cheater…

Isn’t the purpose of re-clocking to remove jitter?
Maybe transformer isolated too?

You could call that cheating, or you could call it a sensible approach to dealing with any audiophilia nervosa you might have about the possibility your sound is being polluted by your hardware/software. In general, any computer hardware/software issues can be mitigated by simple, common sense approach to system design:

- Set up your software to be bit perfect.

- Use an asynchronous data connection, either USB or Firewire, either built into the DAC or as a converter to AES or SPDIF. Viola, jitter is now solely the problem of your interface or DAC.

- Alternatively, at least make sure your DAC has very good rejection of SPDIF or AES input jitter - most, but not all, high end DACs do.

- Pay attention to galvanic isolation. My understanding is that most SPDIF inputs are transformer coupled, but not all? Use power conditioning, or at least plug the PC into a different circuit than the main system.

There, you've just walled yourself off from the worst of the potential issues, and you can get on to listening. Or, you could do what I did, and drop significant coin on a Blue Smoke Black Box, designed from the ground up to deal with these issues. It's a terrific piece, and it's completely stopped me from having to mess with hardware anymore.

 

[Scott Borduin]

ScottB: ever tried DS versus WASAPI?

Tim: is it a reclocker or a protocol converter you are using?

I have, and honestly I can't tell the difference if the output sample rate is set to match the source. I use WASAPI or WaveOut usually, just because DS does nothing in my environment except add overhead. But at least on my Blue Smoke system, DS delivers bit perfect output when the output sample rate is set to match the source - is that true of Win 7 and Vista in general, or just a feature of the Black Box audio pipeline?

BTW, I don't even usually listen bit perfect in the last few months. I'm using the VST plugin feature of J River to support a parametric EQ plugin called ePure, which I use to correct a significant, broad ~120-180 Hz room resonance that stubbornly resisted all attempts at passive treatment. The resulting sound, unmuddied by the upper bass/lower midrange problem, is significantly more transparent than without the EQ. That's another advantage of computer-based audio, and unlike the bit-perfect scenario, the output of an EQ is definitely different between different software.

 

[Vincent Kars]

By design Vista/Win7 resamples everything to what is set in the Advanced Tab of the Audio Panel. If the source sample rate is equal to this setting no resampling will take place but every audio stream when using DS is converted to 32 bits and dithered.

WASAPI does automatic sample rate switching and as it bypasses the Win audio engine, no dither is applied.

VST.
This is where DSP is about, altering the bits to improve the sound.
I use Isone Pro when doing headphone listening
As bit imperfect as hell but an improvement

 

[Phelonious Ponk]

ScottB: ever tried DS versus WASAPI?

Tim: is it a reclocker or a protocol converter you are using?

It's one of these, Vincent:

http://www.trendsaudio.com/index.php?option=com_content&view=article&id=13&Itemid=17&lang=en

It does re-clock, runs separate analog and digital paths and is powered by USB bus or a rechargeable outboard battery power supply. Sounds good, which to me means it doesn't sound. The re-clocking is just a free option for me. I was merely looking for a way to get usb out and convert it to optical and coax. It does AES/EBU too. $225 with the external power supply. God bless the Chinese.

Tim

 

[Phelonious Ponk]

Scott, dude, here's the problem:

...feeding my AlphaDAC/Spectral/Magico system

Your system is clearly not resolving enough to reveal the differences between digital players.

Tim

 

[Phelonious Ponk]

- Pay attention to galvanic isolation. My understanding is that most SPDIF inputs are transformer coupled, but not all? Use power conditioning, or at least plug the PC into a different circuit than the main system.

Funny you should say this. I developed a nasty buzz in my headphone system a couple of years ago. I checked all the cables, changed power strips, moved things around a bit. Finally, on a tip from some web site, I plugged everything analog into one power strip and ran it to one circuit, everything digital into another and ran it to a second circuit. The buzz disappeared.

Isn't WASAPI that green stuff that comes with my sushi?

Tim

 

[Scott Borduin]

Isn't WASAPI that green stuff that comes with my sushi?

Tim

Yes. Coat the edges of your CDs with it before ripping - the improvement in air, soundstaging, and treble transparency is dramatic.

Some people claim great improvements from using it on the top and sides of the music server PC as well, but I believe those claims to be scientifically unsupported. Perhaps I just haven't listened long and hard enough due to the eye-watering vapors it gives off.

 

[Phelonious Ponk]

Yes. Coat the edges of your CDs with it before ripping - the improvement in air, soundstaging, and treble transparency is dramatic.

Some people claim great improvements from using it on the top and sides of the music server PC as well, but I believe those claims to be scientifically unsupported. Perhaps I just haven't listened long and hard enough due to the eye-watering vapors it gives off.

You're my kind of comedian, Scott.

Tim