Dear owners of Koss Porta Pros/KSC75s: Please test this!
Oct 2, 2007 at 4:04 AM Post #16 of 50
I won't be able to answer everything in a single post... nor should I because some of this stuff is way off-topic.

Some preliminary remarks about my short experience with this setup before going into the theory:
I'm trying DH+filter right now and I think I understand why I should have done this before yacking on and on. Hopefully, I will manage to convey to you why I stubbornly insisted on deconstructing your DSP chain. You see, as an occasional DH user, I had some idea of what to expect.
First, let's get settings out of the way: I assume you're using DH1 and the light setting on the Portapros.
This is what I'm using now and it seems that some problems with DH are indeed fixed by the filter. I should really have tried it sooner but I really had no idea it could be done, however obvious it seems in hindsight.
The trouble is that not all DH issues are solved... how am I supposed to tell which instruments sound wrong then? Most of them do! Let's take an egregious example: tabla (the left one in particular). Like many drums, depending how the player hits them, there's going to be a big resonant sound after the initial hit. This is badly mangled by DH leading to a bloated, kinda under-watery sound. The filter seems to make an improvement but it's hard for me to say how much at this point because I'm not used to listen to Portapros and they manage to mangle stuff like this pretty good without the help of DSP.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
I think it should be clear that there is actually an optimum in the parameter space; which can be found


I don't think there is actually (except in a pseudo-Platonic sense maybe)... and I certainly wouldn't know how to find it.
You may be making assumptions in regard to the background and perspective of the people around here. There's no qualification or test required for a forum membership so you'll be sure to find a good number of weirdos here... very little is going to be clear for everyone.
I happen to think my somewhat unorthodox take on this matter is well-grounded and that yours is myopic but then again all crackpots do, don't they? At least I'm not alone around here.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
As I wrote in some previous post, DH must assume a certain kind of headphone. Because, if it didn't, you'd need a linear headphone and your diffuse-field equalized phones would sound like terrible. I think this implicit EQ is the biggest issue with DH, but luckily one that is present in all of its profiles and thus can be fixed.


Yes but... there is yet another problem. You're apparently assuming that headphones only differ in built-in EQ. This is counterintuitive and most certainly not true. I wonder what kind of headphone is assumed for DH.
 
Oct 2, 2007 at 4:18 AM Post #17 of 50
Oh yeah, and there's this:

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
do you know any such phones which I can directly (off-the-shelf) drive with my Audigy card?


I've no experience with that card but I assume it could drive pretty much anything... but how loud and how well? I don't know. And even if I had this card, I couldn't make such blankets statements because it would still depend on your target SPL, your standards of fidelity and your ability to hear amping issues (of which there are several, further complicating the issue).
Do you have any idea of what that card is designed to drive?

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
Price doesn't really matter b/c I'd return them after that.


At the moment I can't think of a good headphone I'm familiar with which would still be in production actually so I wouldn't be the guy who should recommend you such headphones. That said, the two top models from AKG and Beyer should be relatively safe bets.

Note that headphones do not give soundstage, no matter how expensive they are: you make it up. This isn't some kind of vain wisecrack: we really hear differently and a lot of users report changes in how they hear over time as they get used to this or that presentation.
 
Oct 2, 2007 at 10:31 AM Post #18 of 50
Wow this really makes it feel like the music is coming from outside my KSC75s, like i'm listening to real speakers. Nice work.
 
Oct 2, 2007 at 10:56 AM Post #19 of 50
I composed this using my synth a couple of years ago and it has never sounded correct on headphones. Now it almost does!

Purple Shades Lead Voice

Which plugin is doing this effect?
 
Oct 2, 2007 at 3:28 PM Post #20 of 50
Continuing my string of posts in response to post #15...
Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
I have uploaded the FR plots for the source (Koss) and target frequency response (the one that DH is likely to expect) which were used to build the current filter


How did you determine what FR DH is likely do expect? It seems odd that they would choose such a FR (essentially flat with what I assume to be the conventional treble roll-off) when no actual headphone that I'm aware of has a FR that's anything like that.
Don't you think it would be reasonable to assume that, even if the actual DH processing assumed such a FR, that some EQ would then be applied when it's packaged as a product? Maybe the DH library and wrapper we use don't do this, but what about the hardware gadgets out there? You'd assume you're supposed to plug a headphone into them without having to EQ.
 
Oct 2, 2007 at 3:30 PM Post #21 of 50
Quote:

Originally Posted by locu64 /img/forum/go_quote.gif
Which plugin is doing this effect?


There is no telling unless you can explain what makes you say it sounds correct now while it didn't before.
Or we could say that the plugins are doing this effect together.
 
Oct 2, 2007 at 10:15 PM Post #22 of 50
Good to hear that you're trying it out.
I use DH1, but the firm setting (it doesn't annoy me at all and gives me more lower bass). As for the optimum, what I mean is that there must be a setting which, on average, cannot be further improved by changing the EQ parameters. I think this is what you meant by pseudo-Platonic.
This would then be the optimal configuration for this type of audio chain (FS->DH->EQ->Koss), not necessarily the optimal audio chain among all possible audio chains. I think that optimum can be found because you can e.g. optimize the bass separately from the midrange and treble (up to a certain granularity). Then you know your bass is ok and you can continue with the other frequencies. (Technically, the global optimum in the EQ parameter space is at the same time optimum of many of its subspaces (i.e. sufficiently broad subbands).) Btw: I got myself an AKG K701 for testing.

Quote:

Originally Posted by HFat
The trouble is that not all DH issues are solved... how am I supposed to tell which instruments sound wrong then? Most of them do! Let's take an egregious example: tabla (the left one in particular). Like many drums, depending how the player hits them, there's going to be a big resonant sound after the initial hit. This is badly mangled by DH leading to a bloated, kinda under-watery sound.


Well, this is the type of comments I'd like to hear!
I added a new filter to the filters.zip where I reduced the bass < 250Hz by approx. 3db. Together with the raised treble this comes closer to the sound signature of the AKG. Actually, I'm one of those people who love intense bass, but it's still acceptable for me (though it might sound too weak with KSCs). But the AKG... well, maybe I'm too used to powerful bass.
On the other hand, it might also have to do with the recording, or do your other phones replicate that particular sound more realistically?
If they do, it would be cool if you'd try to tweak it with the CurveEQ editor until it sounds as it should or upload the sample somewhere.

Quote:

Originally Posted by HFat
How did you determine what FR DH is likely do expect? It seems odd that they would choose such a FR (essentially flat with what I assume to be the conventional treble roll-off) when no actual headphone that I'm aware of has a FR that's anything like that.
Don't you think it would be reasonable to assume that, even if the actual DH processing assumed such a FR, that some EQ would then be applied when it's packaged as a product? Maybe the DH library and wrapper we use don't do this, but what about the hardware gadgets out there? You'd assume you're supposed to plug a headphone into them without having to EQ.


I shortly explained the choice in the old thread, and I sill think that it holds.
E.g. if you want it to sound halfway good with all of the phones in this plot, you'd probably end up with a very similar FR. Do you have any other expectation?
I'd assume that the all-in-one solutions, e.g. DH-enabled headphones are specifically equalized. But for generic implementations, e.g. in an A/V receiver or our software version, one cannot make any model-specific assumtions other than that it should sound best with smooth phones which don't have excessive bass or overpowered treble. Maybe this part is actually implementation-specific (but this wouldn't bother us much since we're using a reference implementation anyways).

Quote:

Originally Posted by HFat
Yes but... there is yet another problem. You're apparently assuming that headphones only differ in built-in EQ. This is counterintuitive and most certainly not true. I wonder what kind of headphone is assumed for DH.


Concerning DH's expectation, they shouldn't have any other parameters to adapt besides frequency and phase response, because their filter is linear.
Btw, my perspective is that headphones differ mostly (but not exclusively) in their FR, the others aspects being phase response and nonlinear distortion (harmonic/intermodulation distortion). This is an exhaustive list.
There are quite a few papers/articles on the impact of headphone phase distortion ("generally negligible") and nonlinear distortion in audio systems ("specs: <0.05%"). In this thread (worth reading), they digged out a few of them.
 
Oct 3, 2007 at 1:08 AM Post #23 of 50
Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
firm setting (it doesn't annoy me at all and gives me more lower bass).


Indeed. I felt the balance was way off with the firm setting which I why I assumed you were using the light setting.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
As for the optimum, what I mean is that there must be a setting which, on average, cannot be further improved by changing the EQ parameters. I think this is what you meant by pseudo-Platonic.


Possibly... it would depend on what you mean by "on average".
Improving the parameters could mean several things but habituation is going to set in before long in any case. In practice, I think there's going to be an optimum "area" rather than an optimum "value". If you think averages point to some abstract "true" optimum, be my guest. You'd still have to find a way to randomize habituation though.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
I think that optimum can be found because you can e.g. optimize the bass separately from the midrange and treble (up to a certain granularity).


Not sure about that either... in my experience, perception works in a more holistic way. But maybe I don't know how to do this properly.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
I added a new filter to the filters.zip where I reduced the bass < 250Hz by approx. 3db. Together with the raised treble this comes closer to the sound signature of the AKG.


I think this is a definite improvement. See lower.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
On the other hand, it might also have to do with the recording, or do your other phones replicate that particular sound more realistically?


Yes on both counts. Putting the change in the position of the imaginary sound sources aside (usually a significant improvement when compared to the regular Portapro sound), some recordings sound mostly OK and others quite weird relative to a better headphone.
If I had to make a preliminary generalization, close-miked sounds alright and recordings which already have some room ambiance built in sound wrong (the instrument influences that obviously). Largish resonant instruments in general seem to have more issues. One random track I tried which sounded really bad was partly, I suspect, a recording of speakers (live amplified performance). Overall, I'd say many acoustic recordings tend to sound reverby and wrong while electric guitars, electronic music and such tends to sound good to me. Binaural also has weird reverbs (I know it's not supposed to sound right).

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
If they do, it would be cool if you'd try to tweak it with the CurveEQ editor until it sounds as it should or upload the sample somewhere.


Do you really think it's possible to solve those reverbs/room issues with EQ? how?

Do you know about this program's built-in EQ: http://www.cockos.com/wiki/index.php..._Documentation
I'd rather use that because VST in foobar seems like a PITA. Not that I'm any good at EQing anyway.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
E.g. if you want it to sound halfway good with all of the phones in this plot, you'd probably end up with a very similar FR. Do you have any other expectation?


Yes. First, it's often said that one shouldn't compare IEMs with headphones when it comes to Headroom's graphs. Second, look at how the curves evolve, not at some kind of disjointed freq-by-freq average.
I think you've already gone in the right direction by replicating the bass roll-off but you don't have the other part of the "dome" which would form the mid-bass hump and what you have above 1kHz is a smooth roll-off totally unlike the jagged curve that pretty much all headphones have... look at the big hump around 6-8KHz for example. There's a reason the curves look like that (ask the pros for details... I know next to nothing about this stuff).
It might make more sense to use supra-aural headphones as models for the target FR rather than cirucum-aurals if you're going to apply the filter to Portapros by the way.
That said, maybe the FR Dolby assumes is nothing like those headphones and looks more like the ones you imagined. Do we have any documentation, interviews or whatever to give us a clue?

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif
In this thread (worth reading), they digged out a few of them.


This has already been discussed here. Please refer to this thread (not all posts are relevant):
http://www.head-fi.org/forums/showthread.php?t=262490
If you can make your Portapros sound like the K701 you borrowed, congrats! I'm not talking about sound quality or some such: just try to get a similar perception of where the sound comes from.
 
Oct 3, 2007 at 6:22 PM Post #25 of 50
Um wow, I just followed your procedure and using my KSC35, I notice a huge difference in the soundstage/imaging. I'm in the process of trying different styles of music to see how it does.
 
Oct 4, 2007 at 6:03 AM Post #27 of 50
I invested a few hours with the CurveEQ and could significantly improve the sound. In head-to-head comparison to how it's now, the regular Koss seems to sound more often than not like some kind of toilet bowl with built-in speakers. :) Though I must admit that my music choice may be a bit biased.

The updated stuff can be found in the first post.


@HFat: Sorry, I'll answer your post later this evening; have no time right now.

@PeterDLai: Oh NO, I don't have it! And last.fm just doesn't want to come up with that song.
 
Oct 4, 2007 at 7:50 AM Post #29 of 50
Quote:

Originally Posted by PeterDLai /img/forum/go_quote.gif
If you want to hear a DRAMATIC effect, try listening to The Postal Service's "Such Great Heights" with the DSP off then turn it on.

It's tripping me out.


Keep in mind that the ears slowly adjust to the sound you are hearing, and any sudden switch may cause the new sound signature to sound alien.
 
You must log in or register to reply here.

Users who are viewing this thread

Total: 2 (members: 0, guests: 2)
Back
Top