Subjective comparison of Software Music Player

[manueljenkin]

A comparison of music players on windows, linux, android I have done.

 

Note: Most of these software are free so you can try and check for yourself. This is only relating to USB dac connected to PC directly and using a general asio/uac2 driver. If you have a network streamer or signal regenerator/reclocker or any other protocol, the inferences may not carry over. Your mileage may vary on protocol support.

 

For all of the tests/analysis, I have tried my best to make sure there is no additional zero padding, upsampling/oversampling etc is in play within the server system. I almost always used ASIO with proper buffer configurations wherever supported (almost all of them do except strawberry, mediamonkey etc). More on the technical breakdowns at the end of the analysis.

It is too early to conclude the causation of these changes but a look into USB audio protocol (asynch) gives better clue. It works more like UDP than TCP in that it doesn't guarantee a failure when packets go missing or get corrupted during transmission. There is a possibility they could just get concealed, either by any logic/fsm in the usb receiver, or the fact that a delta Sigma dac can behave as a natural packet loss concealer. I have done experiments and I have found that the more constrained and the more buffer underruns the harsher/grainer the sound gets to and extent and past that I get serious crackle noises. So yep I have experienced till there, on a system that would be shown bitperfect by any software analysis inside the computer. The key to this is to be able to probe at the i2s pins of the usb to i2s.

Gear used - surface book 2015, apogee groove, supra usb cable, burson fun, sparkos ss3601 opamps, shure srh1540, OnePlus 3. Few other headphones, dac and amp were also used to ensure coverage on other parameters, and they fit well with the same descriptions.

 

1. Windows groove player - really low fidelity. You got a pig fat low pass applied on top of the digital stream (no clue why) and very bad in overall implementation. When you scroll and play back instead of playing from start, you lose details (the only player that I've come across that does this).

 

2. Windows media player - not as terrible as groove but forgettable.

 

3. Foobar2000 - has lots of plugins and features etc. General "I'm an audio enthusiast" circle tries to push this forward but unfortunately sound isn't the best. Its quite poor to be honest, and even music players in android sound better. Outdated asio and wasapi plugins, measurable distortion and just overall sloppy implementation. Quite softened and smooth but in a very artificial and dry manner. A far cry from the best kind of fidelity you can get from windows. Nice tool for streaming from internet though, thanks to variable buffer on the input side of the player.

 

4. Winyl - the first music player software that had decent fidelity on my tries. You can hear the different textures of bass instruments and the detail/depth/resolution is insane. As my friend calls it, it is DAW level audio quality. A bit artificially sharp sounding due to some buffer management issues. Not the smoothest with bad with low res music, you want that, don't look at this. You can hear a sheen of dither noise (i assume, don't quote me) on top of voices in 16bit music as well. Not quite perfect and not ultimate resolution. I had a short trial with certain background task cleaners/audio enhancers and it did show improvements in sustain detail and left right coherency. Don't know how it uses ram but I wish it could flush the full song to ram and play back from there with high system priority. Peers to winyl are xmplay, musicbee and hqplayer both of which sound almost identical at identical settings.

 

5. Hqplayer - it's as good as winyl in equivalent settings. But it's a lot more feature rich. Lets you try custom upsampling PCM or DSD conversion options and can help get a better stream to your dac than the internal digital filters which can sometimes be low fidelity in the DACs. Also lets you try high precision fir filter convolution for a usable high precision eq.

 

6. Musicbee - the best competition to winyl. It's from the same audio library so sounds similar for the most part. But winyl has a more robust port configuration setting afaik and has less artefacts on that front. However winyl let's down in its buffer, while musicbee takes lead there. It doesn't have the winyl characteristic harshness once the buffer is set to load full song to RAM. Has almost all features of foobar, but built in and usable from the get go and actually sounds good. For some reason the player volume is at 50% by default which I recommend you to set to 100% and use the dac control panel to control the volume. Software volume control is most often poor on any player software, unless it's super sophisticated with 64bit precision and stuff like those (roon has those options).

 

7. Jriver - not as good as winyl not as bad as foobar. It sounds a bit compressed and loud in comparison to winyl. Quite clean otherwise. If you loved the geek out you may like this but it doesn't sound "correct" with usb dacs. People have had better luck with jriver sending data to network streamers.

8. Albumplayer - It was promising as it was coded from scratch and gave me options to prioritise it. I set it up at the best possible settings - asio, full song to ram, high system priority. It sounds a bit softened but not bad like foobar - this thing has depth. I don't know if this is what people call as analog. It sounds very different from winyl but very hard to compare which is better. The general opinion from my side is albumplayer is a little too midrangey and slightly muffled for the most part but with a bass punch somewhere (could just be distortion). On one song it just sounds softened and dull but on another it carries a lot of bass weight. Midrange detail is typically more visible on albumplayer but winyl produces the same with a tiny bit more edge. Most of the time albumplayer is softer but sometimes it can sound exceedingly bright or hard hitting. One thing i like though is that it can bring back that tactile feel in bass in certain songs which i mostly lost with winyl. It sounds weightier and more natural which i like but i just can't get over the resolution winyl offers typically. I wouldn't be doing further testing since I have heard players that outperform both winyl and album player by a significant margin. If the formers are a bit contrasty, edgy and in my face, album player is a bit soft and nice to listen with a different presentation of detail. Idk if it's distortion or just another way to present it or even better accuracy.

 

9. Aimp: I wanted it to sound identical. But unfortunately it didnt. Couldn't find anything bad but the winyl/musicbee kind of players have a sense of texture and aimp has a different texture and imo aimp texture is smoothed off and wrong. Lacks depth. Can't be sure if it is volume or the volume control or anything of that sort. Not a particularly bad player but I'm not particularly impressed.

 

10. Mediamonkey: didn't have an asio output so used wasapi. Didn't like it. Sounded like a low pass filter or bass boost was applied. The liveliness was missing. Can't say i noticed a difference between wasapi and asio in other players but will return to testing media monkey in asio if it improves.

 

11. Audirvana. Not bad. I like it. But winyl still better texture retrieval/detail. Winyl shows the texture and depth noticeably better. Audirvana is a little softer. Bass is even less tactile than winyl on audirvana. (Yep I did try changing settings on audirvana). It was created by an UCLA graduate I think, can't be bad (and it's not), but then again winyl is super lightweight with no bloat, runs on a rock solid audio library, and has multiple other checks to ensure better detail. I'd rate audirvana above jriver (and of course we are well beyond foobar at this point). Didn't sound like it had any compression or anything of that sort. Just not fast enough/transparent enough.

 

12. Roon: to be honest the settings I described above is sub optimal for roon. They recommend you to use your pc as a roon source and another endpoint device like raspberry Pi or ethernet streamer. I can see why. With the same pc acting as both source and endpoint fidelity suffers significantly. It sounds as if I input clipped my amplifier. Doesn't have as muddy artefacts as foobar but I don't like it since it doesn't sound transparent and has a clearly audible aberration. However if your system uses ethernet receiver or roon supported endpoints, I assume it'll be very good owing to the custom roon protocol and compliance control.

 

13. **Hysolid**: My first ever OMG moment. The difference from foobar to winyl was a little bit more than the difference between DACs I tried. The difference from winyl to hysolid was more than the difference between multiple tiers of headphones. If you had said that a year ago, I'd have laughed at you, but now that I have heard, I cannot unhear this. If I have to describe it's sound concisely, I'd say saturated and detailed. It is not the most transparent and has an obvious character that is fairly borderline warmish but without the Treble suffering anything. In fact, Treble texture was the biggest improvement from winyl - finally all the hf noise is gone but without being gooey unresolving mud like foobar. Started to hear a million shades of cymbals and digital hi hat synths (listening to mausam and escape right now). And everything else follows, vocal textures (whispers and so forth), bass instruments. Inter channel integrity still not perfect (I guess maybe coz due to it being 32 bit) but otherwise Sounds stellar. The difference in decay texture and detail with hysolid. The sound of water drops, the sound of claps, vocal textures. Feels like the first time I upgraded from hd598 to vsonic gr07. However it got a serious catch/dealbreaker. It's buggy. the architecture is in such a way that you need to have a separate control system. Android app didn't work (it broke after android 6 and never got an update I guess) so I had to use it on iOS. Running through asio to my computer.

 

14. Regarding android. Where do I even begin. What a forcibly gronked system is all I can say. A forced resampler makes every headphone sound like a hd598 <<sigh>>. Just overall low fi and I rather prefer it sending the stream as bluetooth signal and listen than to use stock android stuff. The only instance where I could get decent sound out of android is when using uapp or hiby music (and an external dac coz the internal dacs also mostly suck). I envy lg users coz lg has a reworked kernel and music player that apparently doesn't screw these things too much, apart from actually having a usable sound chip.

 

15. After all these, I tried my attempt at linux distros. First tried ubuntu without setting up alsa, just pulse audio. As expected disappointment. Then I tried ap-linux, a fork of arch Linux with audio optimizations and kernels built in. The installation process was a pain and unfortunately it didn't support my network card. I couldn't do further updates and sound quality wasn't anything to write home about in that config. I thought I had hit a dead end and had to resort to battling windows.. until.....

 

16. WTFPLAY: this sounds amaaaaazing. All sorts of good adjectives - transparency, detail, effortlessness. There is no character I can attribute to this other than those of the recordings themselves, and maybe minor effect changes by changing buffer settings or some BIOS settings (and apparently changing RAM changes sound, since different RAM will have different memory refresh properties and response times). It is a live cd, so no need to install. Just burn to pendrive, boot from the drive. It'll load to RAM and run from the RAM. When you play a song it'll be decoded and loaded to RAM and played from there, instead of buffers from storage drives. It is super focused in audio tasks and has very minimal daemons. Doesn't have any instruction cycle stealing process - no network, no mouse/mouse polling, not even a battery probe. This is what makes it great since it can respond back to the data request from the dac in a timely fashion, and also there is no sudden noise spike in USB bus owing to noise from CPU interrupts/power state shifts. It sounds great and has been my reference for the last 4 months. Haven't had the chance to compare to ethernet streamers, but within system software, this is leagues above everything available on windows, with hysolid being the only thing that can even be compared, and even that is not as transparent. Highly recommend this player.

 

Continuing some technical analysis from top.

 

ASIO communicates both the bit depth and sample rate along with volume control info in the control frames, as per my understanding. It for sure communicates sample rate that way and I can see it even in my DAC control panel. If you send something with 32khz sample rate via asio to my dac it will refuse to pick it up and throw an error. In other operating system like mac os, we have other data structure for sending the data,including left justified 8-24 inputs. 

 

I did further investigation on this topic earlier. Hqplayer lets you choose the bit depth it sends to the dac. Winyl sounds closer to hqplayer configured to have same bit depth as the music file. Which means winyl is not doing any unnecessary padding. I have actually found padding to reduce detail in my system, making it artificially soft. I can for sure confirm that padding actually reduces detail, in my system and not improve. Custom upsampling using HQplayer is a completely different story altogether, I have tried a few and have been able to see how they "change" the sound but on a pure performance level it was on par with winyl, and worse when additional functionality was being used. Also my system was not up to snuff to try the high precision improvements. There's not much way zero padding can bring in such an amount of difference as heard through the players. Also if all of them do the same thing they should all sound the same which they clearly don't.

 

Most of these improvements can be correlated to how the softwares handle the music playback at the instruction level. Which means libraries used, the stability of plugins used, memory consumption, buffering etc. I have made some investigations on what makes winyl superior to peers in bass audio library and also its one pitfall compared to its peers. It does perform optimizations with regards to usb root control as well. I was trying to see if I can make it play full song from RAM like I can do on musicbee, but after I tried hysolid, i've stopped that approach. I am 100% certain the way a library, the instruction set and features are implemented make an insane level of difference. Changing buffer and load to ram changes things. So many small changes changes things. The way their instructions are structured and the way they behave with other io has a correlatable effect on sound. In a normal music player, you need handles for i/o, trying to see where the user wants to scrub to which means your instructions will have structure in a way it keeps asking for memory pointers. The better ones do a more streamlined getting it closer to just store in a register and keep shifting it, and using it. And as I mentioned above, in my device stack the differences are bigger than different tiers of headphones.

 

Now coming to the topic of foobar sounding objectively worse. This player showed a measurable worse performance across multiple independent tests. (https://imgur.com/gallery/50P4hRJ). All the above topics, we have few players like hysolid sounding objectively better on all fronts, and other players with custom libraries like winyl,hqplayer,musicbee, etc are give an take between them but all sound better than foobar under any parameter. An accurate player will have more depth and separation for almost any song, and that foobar sorely lacks. There is some processing/added distortion happening without a hint of doubt and that is the reason for lack of detail, not accuracy. The player sounds different across different versions as well as per a report from my friend. The issue has also been reliably measured with different input tones. The aberration is closer to how a limiter/compressor would perform, and guess what, windows volume control has built in limiter and compressors. These were measured in ASIO, in WASAPI and still showed the same aberration both audibly and measurably. What does this tell? It just tells that foobar is broken,and has some processing happening even though it claims not to with the plugins. This is one huge flaw which makes it sound worse than even normal players like aimp, and apart from this, memory handling and other issues are also abundant, which as I said winyl and other BASS library based players do better and completely smack foobar out. I have tried my best to "fix" foobar. Have tried 2-3 different asio/wasapi plugins and all had different issues. Case's asio plugin literally distorted annoyingly at 100%. I'm not moving away from my conclusion at all. I've tried revisiting foobar multiple times (whenever I change something in my chain) in an attempt to see and have always been disappointed. Winyl sounds closer to professional DAWs.

 

Tl:dr: WTFPLAY >> Hysolid >>>> Winyl = Musicbee = Hqplayer = xmplay > Audirvana >> jriver = roon > Aimp > uapp on android > foobar2000 > any bundled music player with windows or non asio supported players.

[manueljenkin]

1 hour ago, fas42 said:

I've found that there is no general rule to follow - previously, foobar2000 was always knocked out because the quality was too poor - with my latest, 2nd hand laptop, foobar2000 was neck and neck with Media Monkey, my preferred player up to that point; in fact, foobar was slightly ahead in some ways.

 

Why should this be? Perhaps the latest version straightened out some kinks, perhaps there was a better match between this hardware and the way the software operated - I would say, every time you change the hardware, and every time a new version of the software comes out, double check that the ordering of subjective quality hasn't changed ... just in case, 🙂.

I have tried foobar from versions 1.3 till 1.5 . Very similar results. I don't think it has to do with the computer, though the computer by itself can impart sound. I think the more important factor would be the operating system. All my comparisons were done with windows10. Can you check if your other laptop also runs windows 10? Different versions of OS can have different kernel stacks.

[manueljenkin]

Is it on your windows 10 machine you are finding foobar to be neck and neck with musicbee?

 

I tried mediamonkey but it didn't support asio, so I moved on. Are you able to get asio working on it?

[manueljenkin]

28 minutes ago, Iving said:

... + Hysolid and maybe others only USB?

 

@manueljenkin what in your view would be the best alt. to fb2k for an ethernet/non-USB system PC [ASIO/Dante Virtual Souncard] > RedNet device > DAC - tyia

Unfortunately my experience there is limited. You can try jriver or roon as they are known to play very well with network audio. Or the same players in the list you can try if they work.

 

There are multiple linux distros other than wtfplay just for audio - daphile, Gentoo player etc.

 

Maybe the owner of Uptone Audio might have better suggestions. 

[manueljenkin]

15 minutes ago, Iving said:

 

ty

 

Commercial/subscription players are anathema to me. I am adamant about this. Your post began with "most players are free" - and I was drawn in! I like fb2k partly for its simplicity. My listening environment matters. I do not want any third party to do my thinking for me, make any suggestions whatsoever - or remind me of commerce when I am listening to music. I want to experience music alone - offline. Call me old-fashioned. I shan't care.

 

DVS requires Windows. No Linux.

 

Appreciate your OP - and thank you for your reply - saves some mucking about 🙂

Can you try smallplayer? Or ulilith? Both of them are free.

 

Or you can do this, just go to the BASS audio library development page. You'll see links to maybe a hundred music players. Try the appealing ones and report back?

[manueljenkin]

This is why you don't post on reddit: getting attacked on personal gear choices (comments section). I think I'd have got an even warmer welcome posting it at our friend forum : ASR.

 

[manueljenkin]

12 minutes ago, nonsynchro said:

What a great comparison.

I'm digging Jukebox 2112. I like its interface, the album artwork is wide form taking up half of the screen. It looks really good. I'm in the process of finding images on the net to replace the usual square jpegs. The player has inbuilt equalizer, replay gain and many more customisable settings. And it has VU meters!

N.B. it's not free at $20. I'm not affiliated, but I am a very happy user.

I'd love to know what you think of the audio quality. 

I'd love to try it at some point of time. It's a windows Store app right? I came across this but sidelined it since it was paid.

[manueljenkin]

1 hour ago, beerandmusic said:

thanks for your knowledge on wtfplay....can you confirm when you say no networking, so you have to have your music stored locally?  Also, you can't control it from remote node? (both local storage and command line interface only option of playing?)

It supports usb drives in fat32, NTFS and ext4 formats. No networking, so no remote control though you could use a wireless USB keyboard I suppose.

 

Regarding resonic. It is also based on BASS audio library similar to winyl and music bee. I tried it for a while, but since asio is not supported for the free version I couldn't make proper comparisons, though I remember it being similar to winyl, musicbee, etc as expected. My friend who has owned resonic ranked this way. A well set up musicbee > resonic > winyl. However take note that the differences are quite small and changing buffer settings could make them identical. The differences are a lot more profound swapping to an entirely different code stack/library like hysolid or wtfplay.

[manueljenkin]

3 hours ago, Norton said:

You should also add Bug Head Emperor and XXHE to your list, both of which I’d put in the top tier of players although neither really aimed at simple replay of files, “as is”.

 

The former requires a powerful PC with lots of the right memory, plus a Zen-ike approach to even finding the download (although there is no charge still I think) let alone setup and operation (think secret menus, gnomic-named routines etc) A calm acceptance that music may simply not actually play 50% of the time is also needed, but I’m sure if you consider the cherry blossom long enough you will find the path to audio enlightenment.  It does sound good, but on my “life’s to short to..” list

 

 The latter is excellent but can seem daunting at first and it’s ideal use case is serious PCM upsampling via minimised OS into a NOS DAC (ie Phasure’s own) although it sounds great into my Oppo205 at 768 too. PeterSt  the developer is on this forum.  

I havent tried XXhighEnd but I have tried bughead emperor and minority clean. I refrained from adding that to the comparsion since it is quite a hard to find/use software.

 

I liked minorityclean in conjection with winyl, but the difference was minor, mostly in sustain texture. The time requirements for bughead emperor wasn't cutting well with me, and unfortunately I didn't hear sound that Blew me away. Though they may have been better and I just got frustrated enough to not bother. They certainly didn't blow me away the way hysolid or wtfplay did but then again these can be system dependant.

 

This post was only an analysis from my point of view and I have ignored a few players I have tried that seemed too much hassle or dealbreakers like not having an asio plugin etc. I would be really happy if people take the template and improve with their opinions of these other players and more, and even more so another article for realtime network streaming players etc. Not to mention certain aberrations can swing in different ways in different systems. The same buffer underrun problem from winyl that caused harshness in my and few of my friends dacs like apogee groove, nx4, etc., it caused an artifical rounding/softening sensation in my friends system that had dangerous convert 2. And a few other players like jriver and roon which sounded rounded off to me sounded sharper to him. The aberration, based on its type can manifest in different ways in different systems. If it is just noise floor modulation or something of that sort due to noise from CPU or anything making its way through the system to the dac, the effect will be same on all dacs. And that effect is there on most systems no doubt. However the above example shows that there is some other error that is more across data integrity based that might manifest in different ways in different systems. Buffer underrun I mentioned is one example. This is why I also described the test setup including the laptop used. Even memory types can make a difference apparently. Of course in all cases, a fairly transparent one would be unanimously found easily due to lack of any noticeable aberrations, in my case wtfplay is closer to that ideal. I will try xx high end sometime, especially if i get access to a good cable and dac that can convey 768khz information properly.

[manueljenkin]

I think the word command line puts off many people as hard to use. The reality is that it is just as easy to use as mouse interaction and often way quicker and far more automatable and rule customizable. Once you go command line, you'd want to do it for almost any task except the ones where you really need tracking. Also the auto fill commands/shortcuts are just as robust so what would take multiple clicks and navigation in a custom mouse interaction would be just a few tab button presses in command line.

 

Also one more thing about wtfplay is that it doesn't have any write permissions. The OS usage corrupting your data should be in the last of worries when using wtfplay.

[manueljenkin]

On 5/8/2020 at 12:31 PM, Rexp said:

Have you tried any 'cloud players' that play files from Google Drive?

Not yet. Any recommendations?

[manueljenkin]

On 5/10/2020 at 9:41 AM, Rexp said:

I'm playing around with this one:

 

http://flacplayer.ehubsoft.net/

Thanks.. we'll check that out. However we need to be careful since we have even more variables here. Network speed, different browsers, etc

[manueljenkin]

13 hours ago, jabbr said:

 

Whether you choose to do that is your choice, just as whether to use a particular cable or power supply or really any pice of equipment.

 

The fact that the software is paid or not should not be of concern — do you expect the ISO Regen to be free, or a particular cable? 
 

Do you even use Linux or Windows? 
 

I use Linux extensively and support its use in audio. Comparing Linux audio distributions such as Audiolinux, wtfplayer, Sonic orbiter and a handful of others including NAA could be a reasonable comparison of similar functions.
 

 

There is no benefit to the typical consumer in doing anything we do here. The software is a bargain that actually does something as opposed to other things we spend our money on. 
 

In any case you ought not criticize something you admit you haven’t done and have no experience with.

 

Any frankly HQPlayer is not the only software package that upsample, for example XXHE, Roon etc ...

I refrained from adding the full comparison of hqplayer only because my pc was incapable of using it. I would really love to know your opinions and experiences on the same.

 

I was only comparing on the basis of system level timing/scheduling/noise in my post. That is one aspect of sound and I am confident that hqplayer is equal to winyl on that front.

 

Regarding filtering, not only am I interested in the filters inside hqplayer, I am also interested in custom coding filters using AI (in fact I do a few linear filters in matlab with my audio) and non linear filters like, compressors etc. On a side note I would also love to try different hardware upsampler implementations.

 

Hope that cleared it off. Hqplayer discussion is not off topic but the actual filter comparison can be made in their specific thread. I definitely agree the aberrations are different and cannot be compared oranges to oranges. Wtfplay dev is working on SOX so we could get a more comparable ground between wtfplay and hqplayer in some time.

[manueljenkin]

I hope everyone understands the entire audio sector of us is a niche. Most people never buy more than 5$ earbuds.

 

I really don't see the point of trying to pull in things like 99% don't care or whatever. This post and this place is for those who care. And the differences is there, so it is worth caring for if you are a serious music afficionado.

 

I only started this thread as a more or less a starting place to pull in music player comparisons. As long as things are well organized, I have no issue with anyone adding anything into this. Different people have different usability requirements and different softwares cater to it.

 

However make sure you clearly specify what you are comparing and based on what merits. Different aberrations have different characters.

 

I am really thankful to most people who posted here, since I did learn a thing or two. I only started this post. It's everyone's (for those who care) playground.

[manueljenkin]

2 hours ago, bluesman said:

If your PC is incapable of using it, I don't understand how you made these observations:

 

"Hqplayer - it's as good as winyl in equivalent settings. But it's a lot more feature rich. Lets you try custom upsampling PCM or DSD conversion options and can help get a better stream to your dac than the internal digital filters which can sometimes be low fidelity in the DACs. Also lets you try high precision fir filter convolution for a usable high precision eq."

Well I meant that I was only able to use the lower precision filters without issues on my pc. Since the analysis is incomplete without the higher precision ones I refrained from commenting on that aspect though I am welcome to anyone else's opinion after they have covered like @jabbr. Comparison to winyl is easy "under equivalent settings" which is no upsampling or any processing. I have tried FIR filter convolutions both on hqplayer and on MATLAB.

[manueljenkin]

Umm I did mention every detail of my chain in the original post.

 

"Gear used - surface book 2015, apogee groove, supra usb cable, burson fun, sparkos ss3601 opamps, shure srh1540, OnePlus 3. Few other headphones, dac and amp were also used to ensure coverage on other parameters, and they fit well with the same descriptions."

 

I had missed out uptone uspcb cable, so you can add that too.

[manueljenkin]

I don't know about kenrockwell but amir measured it and he didn't find any such abnormalities on the wrt distortion. Measures as good as many desktop grade dacs (Note: I'm mentioning this only because you wanted to know in terms of measurements. I am not a big fan of such limited measurements which dont capture all aspects of sound).

 

It is a current feedback amp and has a logic that controls current as per the load. It can be plugged directly into an amp without any issue. It output 5V line out at 100% so I use it at 55, or -8db for 2V line out (controlled by asio/alsa driver, nothing in software volume scaling).

 

It is interesting how the shures behave since it sounds very sharp to me and far more detailed than my hd800 and hd700, yet the measurements don't show it. Shure srh940 is the best headphone I've ever heard outside of srh1540 and the only change sonically felt between the two was a minor change to fr. Srh940 is the fastest and cleanest CSD measured at goldenears, along with good attack transients.

 

I don't want to get deeper into that again. I only mentioned the "best" analysis tools I have. Here is the rest of the list.

 

My other dacs and amps : geek out 450, topping nx4dsd. (Apart from groove and burson fun). The sparkos is used in the fun, replacing the V6 vivids it came with.

 

My cables : apogee stock cable, supra usb cable 1m, uptone uspcb cable.

 

Headphones :

Shure srh1540, Hd700, hd598, hd800 ( Hd800 is without earpads, it sounds too bassy for my tastes with earpads), Sony mdr cd1700 rev1, Sony mdr F1, Sennheiser urbanite xl, fostex thx00, akg sextett, Sennheiser hd425 vintage (dead left channel, under repair), beyer dt1350 and few other headphones I don't remember at the back of my head.

 

The changes reported were consistent across multiple mix and match of these sets, with the srh1540 showing the most easily audible intensity of changes along with hd800, urbanite xl. Others like hd700, hd598 showed a moderate change or in case of cd1700 and fostex thx00, no change.

 

I wouldn't be engaging in further discussions on the gear specs part since it will be moving off topic, and the companies aren't paying me to defend their products. I speak out what I hear, thats all. If you don't agree, fine, there are others who agree, especially audio professionals.

 

My best reference gear atm is srh1540 + apogee groove + uptone uspcb + burson fun with ss3601, and running wtfplay and that is what I use to judge products primarily.

 

Someday I'll get to own a qualia 010 or raal sr1a, but till then this will be my reference.

[manueljenkin]

33 minutes ago, manueljenkin said:

I don't know about kenrockwell but amir measured it and he didn't find any such abnormalities on the wrt distortion. Measures as good as many desktop grade dacs (Note: I'm mentioning this only because you wanted to know in terms of measurements. I am not a big fan of such limited measurements which dont capture all aspects of sound).

 

It is a current feedback amp and has a logic that controls current as per the load. It can be plugged directly into an amp without any issue. It output 5V line out at 100% so I use it at 55, or -8db for 2V line out (controlled by asio/alsa driver, nothing in software volume scaling).

 

It is interesting how the shures behave since it sounds very sharp to me and far more detailed than my hd800 and hd700, yet the measurements don't show it. Shure srh940 is the best headphone I've ever heard outside of srh1540 and the only change sonically felt between the two was a minor change to fr. Srh940 is the fastest and cleanest CSD measured at goldenears, along with good attack transients.

 

I don't want to get deeper into that again. I only mentioned the "best" analysis tools I have. Here is the rest of the list.

 

My other dacs and amps : geek out 450, topping nx4dsd. (Apart from groove and burson fun). The sparkos is used in the fun, replacing the V6 vivids it came with.

 

My cables : apogee stock cable, supra usb cable 1m, uptone uspcb cable.

 

Headphones :

Shure srh1540, Hd700, hd598, hd800 ( Hd800 is without earpads, it sounds too bassy for my tastes with earpads), Sony mdr cd1700 rev1, Sony mdr F1, Sennheiser urbanite xl, fostex thx00, akg sextett, Sennheiser hd425 vintage (dead left channel, under repair), beyer dt1350 and few other headphones I don't remember at the back of my head.

 

The changes reported were consistent across multiple mix and match of these sets, with the srh1540 showing the most easily audible intensity of changes along with hd800, urbanite xl. Others like hd700, hd598 showed a moderate change or in case of cd1700 and fostex thx00, no change.

 

I wouldn't be engaging in further discussions on the gear specs part since it will be moving off topic, and the companies aren't paying me to defend their products. I speak out what I hear, thats all. If you don't agree, fine, there are others who agree, especially audio professionals.

 

My best reference gear atm is srh1540 + apogee groove + uptone uspcb + burson fun with ss3601, and running wtfplay and that is what I use to judge products primarily.

 

Someday I'll get to own a qualia 010 or raal sr1a, but till then this will be my reference.

Most of the inferences were made from my "reference" gear, but as subset was tested with other gear to ensure there isn't a serious proxy in play.

[manueljenkin]

Yaaay. I've finally got my android system to Sound good. Turns out the battery optimisations were the culprit.

Changed that in system settings for uapp, and had to fiddle out some options regarding volume control in addition (still experimenting, some inconsistencies). Other settings as before - bit perfect output and the likes. As it stands the sound out of this setup is as good as, if not better than a well configured musicbee/winyl on windows. Linux (wtfplay) on my laptop is still noticeably better no doubt but this android setup is enjoyable too.

And it sounded good out of a phone with an old battery and couple of adapters for usb connection. Would be servicing my phone battery eventually and hopefully I'll see even better gains, since my dac gets its power from usb port.

 

I've asked a question in the cables forum, please do reply if you can. Post 1726 

 

[manueljenkin]

I have finally got around to trying PlayPCMWin, Smallplayer, Ulilith and JetAudio. There is a change to my system during this analysis. Earlier I used my apogee groove (Dac-AMP) + HD800 (nopads), or apogee groove (Dac 2v out) + Burson fun (sparkos ss3601 opamps) + SRH1540. PC remains same - surface book. This test was done using topping nx4dsd, and urbanite xl, both of which are brighter, noticeably less resolving than the formers, but I've tried my best to decouple this from the music player effects. I've personally experienced each link in the chain imparts a specific sonic footprint, regardless of changes in the other links (except when the links are tied together, like Output impedance correlations). I also did the USB selective suspend removal, and removed system sounds as per this guide from focusrite : https://support.focusrite.com/hc/en-gb/articles/207355205-Optimising-your-PC-for-Audio-on-Windows-10 . Buffer size on my dac was set at 32768 samples, the max permissible value, on all scenarios except Ulilith (reason below).

 

1. JetAudio - Unfortunately it doesnt seem to support asio/wasapi. And sq was only Okay. I'm not even sure if it supported usb asynchronous audio. Enhancements were turned off. I'd be happy to give it another try if anyone can guide me to set this up in bit perfect Asynchronous audio.

 

2. SmallPlayer - Sq in terms of texture was alright not much problem with unnatural shimmer or bad boomy or bootleggish bass, but noticeably mistimed. Pans, etc weren't so great. Not sure if it supports usb asynchronous, I doubt it considering how old it is and the sound, I seriously doubt it. I guess it would be a very nice pairing with internal cards on pcie or otherwise. I did try it on my Laptop speakers directly and it performed quite well there, but a bit quiet and extremely lacking in bass. (I guess showing the speakers with their actual properties instead of faking density using compression or other artefacts. It had noticeably more energy and speed than I'm used to with my laptop speakers). I'm gonna keep it for my laptop speaker playback.

 

3. PlayPCMWin - Very nice. I like it a lot. One of the few players that compete well against musicbee in my setup. Had to fiddle a little with settings. Settled on Exclusive mode with Pro audio priority. 16bit downscaling I left to truncation instead of other options, but it's not a deal since my dac accepts more than 16bit data streams. In terms of sound reproduction, while it is noticeably different from musicbee, neither of them feel coloured or wrong. It's the presentation, and texture around instruments where the differences pop in. PlayPCMWin sounds a lot more efforless compared to Musicbee. Musicbee has a bit of an edge to the sound and as if the instrument had an accompanying low level noise band, in comparision to PlayPCMWin where the instruments sounded fairly natural without any additional band/texture around them. However I feel that there is a problem with that too, PlayPCMWin seems to shave off the edges to the sound of instruments a little. I find transient hits to be slightly more impactful and real despite the slight noise around them on musicbee, while PlayPCMWin seems to mute it just a tiny bit away from what could be perceived real. It is tiny, but it is noticeably and makes me wanting a little more attack. It also makes bass guitars and low level echoes harder to follow than Musicbee even though it is cleaner on that aspect. I can sum it up this way. Musicbee shows good detail and hints of what could be hidden extreme detail, but is a bit distracted sounding. PlayPCMWin shows good detail in better precision but doesn't try to extract the hidden extreme details like musicbee does. PlayPCMWin is the closest I have gotten from windows to my current reference WTFPLAY linux distro, in terms of texture, effortlessness and most importantly lack of color. Hysolid, another player for windows, is more detailed than PlayPCMWin and also is just as effortless while extracting a lot of shades of shimmer, but is more denser sounding than what could be perceived real (think PlayPCM win = HD800, Hyslolid = LCD2). It's a bit of toss between hysolid and PlayPCMWin, but hysolid is buggy, PlayPCMWin is not. WTFPlay is still my reference/champ, it has the texture, and effortlessness of PlayPCMWin but also far more detail, better realistic impact, and believable sound than all of PlayPCMWin, Hysolid and Musicbee.

 

4. Ulilith - Seems to have issues with large buffer sizes, was stuttering heavily on 32768 samples on the buffer so I dropped down to 2048 samples. Works great in Wasapi. It is very different from PlayPCMWin and Musicbee. Both the star and pitfall of the show is the treble. Its sizzly, shimmery, but also aggressive. It is not distorted or fuzzy. It is very real, but very forward and in your face. Like you are standing next to a cymbal crash, with no muting off of any part of the sound. Oucch. My current setup being a bit treble fuzzy doesn't help it much either. I expect it to be a lot more enjoyable when I get back to my Srh1540 or HD800. It veers on the opposite end of PlayPCMWin if you keep Musicbee in the middle. Both Ulilith and PlayPCMWin sound cleaner than Ulilith, but both show deviations in either directions. While PlayPCMWin is mellow and laid back, Ulilith is forward and aggressive but without any of the general distorted feel which i get when eq'ing up the treble. I love the energy and shimmer of Ulilith, but I would have to be very careful of system pairing. One more point to note, I first tried buffer to Auto from 32768 which was unbearably aggressive during cymbal crashes. Setting it to 2048 made it more bearable, I think fixing it to a more optimal value would help me get it to good enjoyable levels while still maintaining that aggression.

 

I would be experimenting further on both PlayPCMWin and Ulilith with my other gear and other settings, as of now, I like PlayPCMWin a lot for windows and Ulilith as a flavor player. Musicbee still remains my go-to choice due to it being the middle of the line, despite its minor issues, but all 3 are compromises in one way or other, when compared to something like WTFplay.

 

I plan to move out to streamers where these player swaps would likely start to make lesser difference, I just made these posts to help people looking to extract the best of their existing system.

 

Tldr:
Wtfplay Linux - still the best, by a large margin
Musicbee - a good compromise land in windows, nowhere near as detailed or clean as wtfplay, but still good enough for me.
Play pcm win - closest to wtfplay in terms of texture and lack of colour/noise but also noticeably lower detail, slightly less than even musicbee. Ever so slightly muted in the attack.
Ulilith - Clean and aggressive. Opposite of playpcm win. I love the flavour but pairing can be iffy if the rest of your chain is aggressive as well.

[manueljenkin]

3 hours ago, seeteeyou said:

Awesome comparisons among so many candidates, thanks a lot for taking the time.

 

So far wtfplay seemed to be the champion and there's another comparison as follows

 

https://audiophilestyle.com/forums/topic/24966-ecdesigns/page/23/?tab=comments#comment-1070708

 

Just wondering if SqueezeCore + LMS were able to rival wtfplay by any chance?

 

https://audiophilestyle.com/forums/topic/58398-squeezecore-a-minimal-audio-os-running-squeezelite/

https://audiodigitale.eu/?p=228

 

Thank you for the kind words and the links. Unfortunately my experience stops here. I am yet to try any networked audio chain or signal regenerator. If I ever come across and make a post/comparison, I'll tag you. The sdtrans384 is something I'm interested to get when I have the budget.

[manueljenkin]

Well I guess I was a little quick to conclude PlayPCMwin lacks impact, it was actually the secondary system, mainly the dac I used that was lacking, just that it was masked by the noise in musicbee and PlayPCMwin showed it as is. Went back to my primary systems and PlayPCMwin has all the grunt and kick, and the effortless details with all headphones, including the urbanite. Not quite as good as wtfplay but it's very good. I forgot that I never tried nx4 on wtfplay - it is actually a little muted sounding on a good source.

 

Ulilith is just as expected. Forward treble hits. Just not as obnoxiously harsh as on the topping + urbanite xl but still quite forward, and cleanliness is retained. Can get fatiguing quite quickly. The threshold time counter just ticks a bit slower here.

 

Update: also I can hear noise band whenever I have content around 8khz when using PlayPCMwin, I think this also affects the low end a little, especially if the hit comes in around the 8khz band together with the bass (guessing this must be USB frame noise 1/125us). This did not happen with Wtfplay with the same system and wtfplay was all round noticeably better, so this is unlikely to be a complete fault of the dac, and has happened with all DACs I own. Wonder what is happening here! Is PlayPCMwin likely playing around with the usb output power profile too? In fact this particular issue wasn't distinctly audible even with other players despite them having a bevy of other noise. Is it something relating to how the spread of noise affects the analog section, like if it's a single noise band, it being more reactive/unstable than if it were a wide noise band? My supra usb cable also had a similar but far worse effect and caused a bootleggish sound signature. I'm assuming it's something related to a specific noise profile, might be similar or different to the above. I have tried everything right now using either stock groove cable or uptone uspcb. I'll try to pair up supra with PlayPCMwin and see if the effect compounds or nulls or modulates into something esle.

 

Is there a known way to get this fixed? PlayPCMwin seems to come eerily close to fixing windows audio for me, this particular issue aside. I'm thinking of a simple regenerator or a 8khz notch filter in the ground plane of the data line. I think going external power would also help.

[manueljenkin]

I realised PlayPCMwin developer is here. Hi @yamamoto2002. Thank you so much for this wonderful piece of software. I love it. Can you guide me if I can get rid of this 8khz issue?

 

pc - surface book 2015

dac - Apogee groove

cable - uptone uspcb.

 

Also, does it have restrictions on use of signal regenerators? My friend got this error when trying to use with his micro iusb 3.0. What I understand so far from the error is that it is looking to see if the address of the dac/slave buffer/memory (889000A, assigned by windows when plugging in) is free for taking in transactions, but since the iusb seems to be holding itself as the master for it, it doesn't seem to be free. Or is 889000A the address of iusb and it is refusing to slave to the PC and free itself?

 

Also if I were to try to make a music playback software either in Linux or windows, using full buffer to ram, are there any references you would recommend me? I am thinking of using ALSA in Linux to make one for my Allo sparky, mostly a headless one. Need guidance in what to look for when optimizing the device for lowest noise and better timing on the usb output line. Anything relating to cpu cache processes/coding to reduce even ram access?

[manueljenkin]

Another doubt I have is regarding the fetch of data from RAM. Does the 64bit version use a fetch of 64bits or more data at once (AVX, or SSE) or does it do a 32 bit or less fetch. I'm guessing the former would fetch sample for both channels in one shot (even multiple samples if more than 64 bits), while for 32 bit it would have to do a channel by channel fetch. I cannot ascertain this to any parameters yet, but curious to know if the correlation I'm thinking holds good with respect to stereo separation.

[manueljenkin]

Thank you very much @yamamoto2002. I tried experimenting with the priority settings, toning down to normal/high priority from critical and the other settings and it did help ease things quite a bit.

 

I'm still in the process of exploring audio gremlins, everything seems to affect the sound.

 

My friends also wanted to express their thanks to you for taking time and effort to write such a beautiful tool. 😊.