1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

My experience with different music players.

Discussion in 'Computer Audio' started by manueljenkin, Jan 15, 2020.
  1. manueljenkin
    I realized that playback software's like foobar actually limit fidelity (even when using asio or wasapi). I switched to winyl a couple of months. Someone had measured it a while back couldn't find it now. Winyl is measurably superior to foobar.

    Don't get me started on stock players they are even worse. Microsoft's groove player has a layer of low pass on top of it.

    Played around with hqplayer demo version. My impressions after 1 hour of listening.

    1. When bit depth and sample rate are matched, winyl and hqplayer sound identical.

    2. Making bit depth higher on hqplayer without doing any processing makes it sound a little worse. Somewhat soft.

    3. Making bit depth higher and increasing sample rate to 192khz (max supported by my unit) and adding processing makes it sound different and interesting. Sinc-M and Sinc-M + couple of dithers worked fine for me. The mp (the graphs without pre ring) made it sound somewhat closer to geek out but since it retained dynamics it didn't sound as full/dark/compressed as geek out. Maybe if i added a compressor on top it could sound identical to geek out. Either way you can get the same filter on groove as you do on geek out with this processing.

    4. Most filters worked fine. Very few (the first one I told sinc-mp- long) didn't work well enough. I think they hold the most accurate rendition.

    4. All that said, I prefer winyl because I find keeping things stock has an element of accuracy I don't find otherwise. I'll try to do the higher precision filters in matlab sometime and play around.

    5. Winyl + Minorityclean is on a whole another level. I'm still trying to experiment with different variants of minorityclean but the default version 8 brought marked improvements on most fronts. Hqplayer + minorityclean could sound identical to the above you don't upsample and don't add any processing but the point of hqplayer is those features. No processing on hqplayer even comes close to what minorityclean does.

    * * * * *
    Overall hqplayer is , more like a feature packed variant of winyl, but minority clean is , just overall a higher fidelity medium. Winyl/any equivalent software + minorityclean is as good as the best any windows software could get in audio playback.

    Improvements in minorityclean is two staged. First one comes from cpu level jitter reduction by changing the compiler structure to bypass the CPUs instruction pre fetch/"optimization" and is audible the moment you run this and add your music player to the memory. The second one is complicated, it does something with respect to digital voltage referencing and I am unsure what is happening, but i can hear improvements till maybe like 15 minutes.

    If you could wait for minutes before playing music try the other software from the same guy - pinkHQ, nontallion,etc.. they do a system cooldown and voltage referencing and set the memory pointers in most optimal jitter free locations. I'm not bothering much with them coz i tried it once.. took 15 minutes to start playing one song and I was . Didn't try to find differences/improvements since i was super frustrated at that point.

    Also tried Jriver. It's fine but not as good as winyl/hqplayer/xmplay. But definitely gives a tougher fight on me when trying to compare. Foobar vs my reference players were profound. Jriver wasn't as profound. Overall jriver sounded like my apogee groove was taken to sound like geek out a little. Geek out is a bit of a compressed sounding dac, though clean and full.

    Regarding clarity on hqplayer and winyl. If music is 16 bit, Hq player at 16 bit and winyl sound same. Hq player at 24 bit for 16 bit input music without proper filter sounds worse not precisely sure why.

    Hq player at 24 bit with proper dither and resampling at higher sample rate should sound better. I've tried few things and they gave me good hope. But for now I think I'll have to learn more about the filters technically. There's 3 parameters. And each of them have like 15 options. That's 3000 combinations, some of which my cpu can't handle. Can't do that one by one. Gotta check the math of each, do analysis, and automate this.

    Also, gave a shot at macbook yesterday. Put up with it for over an hour and gave up. Was kinda horrible. Returned back to windows. Using macbook, felt legit like a hipster trying to re-learn everything that used to be intuitive otherwise. Was disappointing since I liked FreeBSD when I tried it on my virtual machine. Seems like it's not freeBSD based. Just some 20% of the code used for networking were based on BSD. Couldn't get usb audio to work properly. It made sound and everything, but didn't sound right. Felt like snakes and ladders where i try my best to climb to stage 90 and the snake pulls me down to stage 3 again - sounded similar to how windows media player sounded, if even worse. For an os that prides itself to be a professional environment I expect it to just work no excuses.
    Last edited: Jan 15, 2020
    Whitigir and Xecuter like this.
  2. manueljenkin
    A shorter crisper explanation of the whole thing in digital domain.

    1. Winyl uses BASS audio library - audio library is what handles system calls etc and the instruction sets, sequences, memory handling. BASS library is probably the best for playback.

    2. ASIO is probably the best playback kernel for windows. Developed by Steinberg. Linux has alsa which I'm not sold on yet, but I'm trying to get OSS working. OSS is apparently superior to asio and i can belive it. It's kernel doesn't have too much junk.

    3. Minority clean. Goddamn genius the guy who coded it. It's quite a simple thing. If you have repeating instructions, the x86 CPUs introduce jitter in an attempt to parallelize it (pre fetch etc). Just make an alternate compiler that spaces these repeating instructions and eliminate this instruction level jitter. Also use 64 bit registers instead of 16 bit registers to make sure all the data for a particular sample is finished in one shot.

    Chain used in comparison:
    Windows pc (surface book 2015 8gb ram 256gb SSD gt940m GPU) -> supra usb cable -> apogee groove -> burson fun + V6 vivid opamps -> shure srh 1540.

    I also used Sennheiser hd800 and urbanite xl and got same results. But I find that the shures are more resolving overall.
  3. manueljenkin
  4. Sterling2
    I've settled on iTunes over others which I've tried upon recommendation. I have one laptop running iTunes which is connected via usb to an OPPO-205's DAC up sampling all AAC, ALAC, and AIFF music to 24/192, outputting stereo analog to Pre-Pro. I am satisfied with the listening experience from this means to music. I have another laptop running iTunes which is wirelessly connected to my Pre-Pro via Airport Express optical S/PDIF 16/44.1 connection to Pre-Pro, whereby the Pre-Pro processes the digital music for analog output.This means to music entertainment is not quite as "detailed" as it is from OPPO DAC, I suspect because my 20 year old Pre-Pro's DAP is not as good as today's processors; but, this can only be discerned at high volume. Thing is, I hardly ever listen to music at what might be described as reference volume so the more convenient wireless access to iTunes music is how I mostly listen to iTunes. Overall, I like iTunes not only for presentation of music but for CD ripping and burning, iTunes Store, Playlist Creation function, Genius Playlists from my iTunes Library, and music recommendations. The bottom-line though is this: since even AAC from iTunes sounds for the most part as nice as stereo SACD, through OPPO's usb DAC, I'm not in the market for something else. My only objection to iTunes is no multi-channel music. But, with Foobar on my computer too that's not an issue.
    Last edited: Jan 16, 2020
  5. castleofargh Contributor
    I installed Winyl, played a sweep in it, did the same in foobar, ran the resulting files in Deltawave and got:
    So AFAIK, your thread's introduction seems to be a lie.
    bfreedma likes this.
  6. manueljenkin
    Umm nope. I'm comparing the data sent through USB port (captured through wireshark). Taking a while to decode, but so far nope they don't look alike at all! I've tried two different asio/wasapi plugins for foobar - one by Peter himself another by Case and both sound different and neither sound particularly good.

    Also, people have measured winyl and foobar (both in asio/wasapi in proper configurations) in oscilloscopes and have found signficant measurable difference. There were even posts in headfi about it. Will update here once I find that link. I'd fault your adc more than the players in this case. What was the adc/interface you used to measure. What is it's resolution limits. What is it's intermodulation performance?

    I'm not lying, your gear is either not having reliable performance or is not set up properly. 200% confident that anyone with an analyser or half decent measuring gear should be able to measure the difference, and it is repeatable.

    Also what kind of coverage metric is it. "I ran a sweep". No mention of any single tone comparison, no mention of intermodulation comparison, Even for the sweep, no mention of sweep duration, cut off frequencies, etc. This is like saying, hey I checked accelerating your car. Your brakes are fine.

    Are you running your gear in asio or wasapi or directsound? Does it have proper drivers installed to have usb asynchronous mode?
    Last edited: Jan 16, 2020
  7. castleofargh Contributor
    I took the digital signal before it goes out of the PC. Your statement was that the player limits fidelity so I went for the signal right out of the player. I completely agree that it's far from exhaustive, and kind of a lazy demonstration, but my attempt was to try and confirm if I could get the same output from both players to disprove your claim, which I did.
    So before trying to shift the burden of proof onto me with a diagnostic rigor that I really wish you'd have applied to yourself when testing for the content of this thread, I believe that you, the one who casually declared that bit perfect doesn't work in foobar, should do a proper fact check and bring up evidence for that big objective claim, if you have some.
    Right now, the way you contest my approach seems to suggest that you don't have a clue what(if anything) is altered by foobar. instead you're just pointing at what my lazy test might have missed. that's not improving your situation IMO.

    TBH looking at the digital output from the players was to give you half a chance. Because if you want to involve the entire chain with all possible audio paths, player settings(dither, oversampling, volume control, codecs....), DACs and their drivers, buffering options, etc, you then have to demonstrate the consistency of whatever degradation you allegedly find. And do enough measurements under enough conditions so that the cause for that given fidelity loss can confidently be attributed to foobar alone and not just one version, or one way to set it, or one computer where it's badly installed for some reason, etc. Going for this is only making your claim that much harder to substantiate IMO. And I doubt you could or would do all that work.
    It's trivial to find some measurable difference(or even better, cause them) under some circumstances. I'm pretty sure I have some old RMAA and nulls somewhere that show how with one player I "completely destroyed" the LSB because of volume control and/or the dither being applied at the output. I hope you're not referring to that magnitude of fidelity limitation, because then it would bring up an all new can of worms about how you're going to prove audibility. Something you also don't seem too concerned about in this thread.

    All in all, if you were right, if the loss of fidelity was significant(audibly so) and an unavoidable issue of foobar itself, I obviously would like to know about it and would like everybody to know about it. then most likely the issue like many found previously over the years, would be fixed.
    I really appreciate facts and shared knowledge. But you have brought up a total of zero supporting evidence. While I've seen many people use foobar as source for measurements and I have used it myself for years anytime I couldn't just loop the thing I was testing into my rig. No recurrent fidelity issue came out of those many and diverse experiments. So either we've been really unlucky(/lucky?), or foobar actually can behave like any player properly set and using a bit perfect path. With fidelity results similar to this: http://archimago.blogspot.com/2013/06/measurements-part-i-bit-perfect.html
    bfreedma and Sterling2 like this.
  8. manueljenkin
    Here is the measurements you asked for :

    Kindly refrain from calling people as liar by using measurements that don't pass resolution cut off criteria. You are really not getting anything by policing people this way, only blocking people from trying to experiment. @jude Tagging you since the attitude of "audio scientists" in this forum has been getting overboard.

    I posted measurements here only because the guy was trying to belittle my analysis. This is not audio science forum and I don't think I'm obliged to show measurements. I am an engineer with decent experience in digital systems and some signal processing math and also know to play a few instruments. I know what I'm analysing. Running a random sine wave on a random gear is no sane way of getting any reasonable coverage.

    Edit: Proper Image link: https://imgur.com/gallery/50P4hRJ

    Thanks and Regards,
    Manuel Jenkin.
    Last edited: Jan 24, 2020 at 6:41 PM
  9. bfreedma

    You have one/limited measurements and no controls to ensure they are meaningful. If this is similar to your claims about Winyl and Foobar being different, you are a long, long way from proof. For example, the measurements you cited as proof on another site don't even include an analysis of the configuration of the players. So while there is a chance the differences you measured could mean something, there's a much higher probability that something in the configuration is the root cause.

    But hey, why do the work when you can take one data point, create a scenario where unvetted correlation equals causation, and call it a day...
  10. manueljenkin
    Sorry, the same test has been repeated thrice by the same guy under three different scenarios. I refrained from posting the first two because there were chances of inconsistencies. Not now. It is 100% clear and everyone who have tried it with properly matched scenario (proper drivers, ASIO/wasapi, buffer etc) has heard it. On the other site, I have mentioned how it was configured. On a personal experience basis, I have tried virtually every plugin/setting I have been asked to try across multiple USB Dacs and the differences remain. He has tried it in different modes and settings with 1khz sine wave, 1khz square wave and 5khz square waves.

    The one who didn't describe the setup is the other guy. "Oh I ran a random sine sweep and didn't find any delta". No mention of where it was probed, no mention of the setup. And not just that, calling me a liar without any credibility on his side. I thought you were knowledgeable and had respect for you. Realized you're not and just an online bully. Proof is done, get over it. It had been done ages ago, in another post in head-fi as well. I'm not responding further till @jude can comment on this.
    Last edited: Jan 24, 2020 at 10:14 AM
  11. bfreedma

    You have 3 measurements from the same person on the same system. You don’t know the configuration of that system or the software.

    I'm sorry you consider anyone who questions your methods and claims to be a bully. It’s an odd stance to take for someone who claims to be interested in facts.

    Disappointingly, you don’t seem to understand the scientific process or, for that matter, the definition of “fact”. You seem to feel a single data point and sighted subjective opinion constitutes validation of theory - good luck with that.
  12. manueljenkin
    Oh playing shifting goalposts are we here? The other guy tells an unresolved measurement of extremely poor coverage and tells I'm a liar - he gets a pass (Oh did I mention there was not even an image of that). A friend of mine takes measurements, I state the measured conditions (multiple of them were tried with similar results), an independent measurement exists on head-fi. Numerous people hear it. And I'm told I'm wrong. Quite nice rules you keep for yourself. And again. There was no reason to provide measured proof here. You guys were given a forum "sound science" to keep shilling your half parametrized "analysis". The only reason I even bothered replying here was because he called me a liar. It has been proven there is a difference. End of story. Have fun with your "science".

    I know the next trick you'll play when Jude or someone measures the same thing in their setup. "It should not be audible" - and I reply to it here already. Who are you to say what should be audible to me or not when even scientists have zero clue on the full functioning of the auditory system? Besides, I am an 83 dimensional being with 3Thz hearing band per dimension on an average.
  13. bfreedma

    I'm sorry you take such offense at the scientific method.

    Do you really believe that something at say, -120db is audible? That's orders or magnitudes below any know human hearing capability. If so, it would be easy to prove - just setup a DBT/ABX test and see if people can consistently identify it.

    Or you can just keep making uninformed claims based on, at best, an incomplete data set. And let's stop pretending that I haven't seen the measurements and commented on the assumptions being made on the site where they originated.

    Regardless, good luck - I'm done trying to help you understand what you're trying to assess and how to properly go about it. You can continue to jump to unsubstantiated conclusions and pretending you've made some major discoveries - I'll jump back in when you publish your peer reviewed technical paper.
  14. castleofargh Contributor
    I wrote:
    You make a mountain of it.

    Then this nonsense:
    I'm tagged "Sound Science Forum Moderator" because I am it...
    the following posts keep on the anti Sound Science blablah. why? If all you can think about to defend what you consider a fact, is to stoop as low as to try and cause trouble with a strawman anti Sound Science dispute, you really should quit(or be stopped if you keep it up).

    My little test didn't need much more information than what I provided as there are not that many places where you can get bit perfect anything. So obviously the signal was captured in the digital domain. You sure complain a lot while you're the one who made the big objective claim about foobar's fidelity, while your evidence is... What is it? Apparently the pictures show how different op amps affect a square wave. Not sure if this is a master class troll or if you linked the wrong stuff? You should probably let that friend of yours come and give the technical explanation, if there is one.
  15. manueljenkin
    Sorry, your test doesn't capture anything useful. Keep calling your inhibitive brick-walled measures "science". I'm not the one who's supposed to quit, and you're not the authority to decide on that either. I made a reference to the full post in sbaf but for some reason head-fi doesn't want sbaf threads here with technical explanation. I updated the imgur link.
    Last edited: Jan 24, 2020 at 6:50 PM

Share This Page