[sound_man]

This is an attempt at making your Koss phones sound (arguably a lot) more natural. And I mean unmodded phones, because this is simply the only way to get consistent results.

I started a thread about this some time ago, but I got almost no feedback from you.
But since I'm speaking of a purely software-based thing, you don't even have to get up from your chair for this. And if "software-based" sets your alarm bells ringing, calm down: This is not the usual effects mumbo-jumbo. It is the combination of an HRTF filter (like the Headzone does it, just static) and some kind of equalization filter which fixes many of the characteristic shortcomings of your Koss (inaccurate/bloated bass, missing mids etc.)

If your are ready for the test, please do one of the following:

1. If you don't have foobar2000 0.9.x installed and just want to do a quick'n'dirty test, download this zip file and copy the foobar2000 folder from it directly to C:\ (this way you save yourself the tedious install and config procedure). The contents were recently updated (see below), so please redownload. Note: Technically, you are not allowed to use the above program without having a legal copy of PowerDVD, because one of it's DLLs is used.
2. If you don't care about your current foobar2000 installation, uninstall it and do the above. You could also try to just run the two versions in parallel.
3. If you have foobar2000 0.9.x installed or want to do it the proper way, install the required components by following this guide (look for the step-by-step part). Edit: The step-by-step guide is currently not fully up to date so please use the above complete package for now (I'll fix that).

And then listen to a lot of music (e.g. jazz, electronic, rock, classic, trance) using this player. It would make sense to compare it head-to-head to the raw sound of your Koss, e.g. by having some other player active in parallel. What's even better is to switch different signal processing setups by pressing Ctrl+P and, with DSP Manager in the left panel, select and load a setup (by pressing the Load button) from the empty drop-down box to the right. And if you have some other (top-of-the-line) headphone model lying about, please compare that, too.

Here is what I want to know from you:

• What can be improved. What frequencies seem unnaturally boosted, what frequencies seem attenuated.
• How it compares to your AKGs/Sennheisers/Beyers etc. The most interesting information is what instruments sound unnatural compared to what the other phones deliver.

I ask you this stuff because I'm a student and, like many other people, cannot (or don't want to) afford these phones/amps/etc. And because I'm searching for perfect but cheap portable sound.

And why should you try it? Because you might ultimately benefit from this, once all obvious problems are fixed (and they can be, believe me). Then you can get nice sound from your portable (you can use foobar to convert your files) without having to carry an amp around etc.

So please come on and give me some comments this time!

Edit1: obsolete

Edit2: obsolete

Edit3: I have updated the contained filters and added a few new very nice setups, e.g. the typical setting for high-end listening (called bright), several not-so-bright variants of this and two derived settings for rock music (called groove and cleanbass). Owners of KSCs should use the settings marked with KSC (currently I don't have these so I have to rely on the feedback from you), otherwise it'll be probably way too bright.
The Filters can also be found here.

 

[HFat]

No comments? Let's bump this. We wouldn't want to disappoint sound_man again.

Let's start with a couple questions: you seem to be saying that there's some kind of synergy between EQ and DH. Is there a reason for that? To me, they're very different things and they ought to be evaluated separately.
Also, what's the source for this free surround DSP? Running binaries that are laying around on some kind of free upload site isn't a good habit. Why is that DSP important in your opinion? I think DH sounds as it should on its own, with the speakers in front when playing 2 ch. material.

Then you could perhaps go in more detail into how the process you used to create the convolver files. I'm sure it would be of interest. I've seen the question "how to I EQ headphone X" pop up several times.

 

[Henmyr]

First I tried this with my porta pro through the head5. I have repaired the cable though so I do not know if they sound equal to stock. Well, I didn't like the effects at all. Much to muffled.

BUT WAIT.. I then tried the ksc75, and it sounds really good. The biggest change is the much bigger soundstage. MUCH MUCH BIGGER. Bigger than my dt880/dt990 stock. They actually sounds like a fullsized headphone using this. The effects also increased the clarity alot. Did I say it sounds really good? The midrange is unbelievable.

I will probably comment more, must listen more though. My initial impression is really WOW. The ksc75 now sounds really good with everything. Beautiful midrange, gigantic soundstage and powerful bass. I can see how some people might think that the sound is too bright though, but I am used to grados and beyers, so it's perfect for me.

EDIT: Other soundstage improving programs that I have tried have changed the sound too much, but this acutally does change the soundstage without sacrificing the sound too much. The bass is still there.

If I had known of this before I bought better headphones then I might not have bought them. Sure, this can't best the beyers or the ms2, but in pure enjoyment, I can't say it's too far behind.

EDIT2:
As for improvement. It might be a little too agressive sounding now, and kind of cold sounding. A bit warmer and leaner might be the hotspot. It sounds very good with metal though as it sounds now.

EDIT3: I can really see me using this on my portable player where I can't have my home phones.

EDIT4: I converted some music and put it on my cowon D2 and had very good results. A little tweeking to fit koss ksc75 and D2 and it will be perfect. The volume had to be turned up a lot though. From 20 which I normally listen too to 40 for the same volume.

This sounded really good listening to Katie Melua, Rammstein, Mastodon and Metallica.

 

[sound_man]

Quote:

Originally Posted by HFat /img/forum/go_quote.gif No comments? Let's bump this. We wouldn't want to disappoint sound_man again. Let's start with a couple questions: you seem to be saying that there's some kind of synergy between EQ and DH. Is there a reason for that? To me, they're very different things and they ought to be evaluated separately. Also, what's the source for this free surround DSP? Running binaries that are laying around on some kind of free upload site isn't a good habit. Why is that DSP important in your opinion? I think DH sounds as it should on its own, with the speakers in front when playing 2 ch. material. Then you could perhaps go in more detail into how the process you used to create the convolver files. I'm sure it would be of interest. I've seen the question "how to I EQ headphone X" pop up several times.

Thank your for the sane questions, HFat. I wanted to keep the technical details out of this thread, this time, and focus on the core issues instead.
I believe this, and the barrier of having to install everyting step-by-step was what kept most people from trying this out. That's why I do not consider this a "bump", as you pertly remarked. Anyway, on to the answers.

First, to the coupling of DH and EQ. DH (or sound localization in general) works because of three cues:
1. It simulates some kind of cross-feed, like it happens with all out-of-head sources (this is necessary for natural left/right imaging).
2. It also simulates the correct delay between the ears (again necessary for natural left/right imaging).
3. It simulates the frequency filtering (equalization) that your ears apply to the sound. It should be clear that this is the only reason why you can tell front from rear sounds. And it's also necessary for left/right imaging; simple example: your hear less treble on the ear which does not directly face the source.

So, if you want your music to sound as if it comes from in front of you (versus from inside your brain), you absolutely have to get the EQ part right.
But DH can EQ as much as it wants, if that cue is mixed up again by the phones you use for listening.
Of course good phones could emulate this part by themselves, without DH. And they do. That's why some phones are said to have a nicer soundstage than others. But, unfortunately, the Porta Pro seems to belong to the "others" group, and even the nice phones cannot simulate the crossfeed, delay and different EQs as they are applied to each music channel by the left/right ear.

Second, the source code for FreeSurround is here. The component discussion is here.
Whether your enable it or not is your choice. It's enabled by default because I don't see what it takes away from the experience. But I see what it adds. And this is a) one more channel in front of you (namely center). Its job is to scale down the zone in which locations have to be faked by the surrounding speakes.
And b) the range in which you can expect instruments (or voices, or echoes) is not just the 90° in front of you but instead about 230° around you, if the composer or whoever wanted to. Finally, FS does not colour the sound, it only decodes the location. So it's presence does not affect the discussion of frequency issues, but it can (my POV: will) improve the immersion/realism, so it should be more fun and easier to think about the naturalness of specific instruments etc.

(A philosophy that is common to both DH and FS is that human ears are not evolved/"designed" to interpret the raw stereo signals like they are on the CDs. What they do handle instead is spatial reality. Speakers are a trade-off, stereo is a trade-off, surround is a trade-off, but they emulate reality fairly well. Most headphones cannot. So to suit the ears well, it's necessary to simulate reality as good as we can.)

To the third point: I generated the convolver files using a Matlab script which reads two frequency response diagrams (in headroom style) and builds a linear-phase FIR filter (i.e. the core of most digital EQs). This filter makes device A sound like device B, up to diagram inaccuracies, phase distortion and nonlinearities. The code is attached, if someone is interested.

The issue of determining what kind of phone (i.e. device B) Dolby Headphone expects has already been discussed in the first thread. This, and the diagram inaccuracies are the most important problems here. Explicitly: everyone has slightly different ears, every driver sounds slightly different and, most importantly, the head model is far from perfect. This turns the identification of the exact nature of these inaccuracies into some kind of guessing game.

 

[HFat]

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif I wanted to keep the technical details out of this thread, this time, and focus on the core issues instead.

Please tell me if you would like to discuss it elsewhere but one of the reasons I brought it up is that I think that, by mixing the virtualization with the EQ, you're not putting the focus on what you want to perfect: the EQ.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif So, if you want your music to sound as if it comes from in front of you (versus from inside your brain), you absolutely have to get the EQ part right. But DH can EQ as much as it wants, if that cue is mixed up again by the phones you use for listening.

Good point, although you may be discounting habituation.

But the logical conclusion seems to me to be that we should test headphone EQ on binaural samples.
Massaging stereo recording to make them sound like speakers (or even surround speakers) on headphones would not be the ideal solution even if the DSP was perfect. Because we're tweaking the reproduction chain, we tend to forget about the recordings but the realistic immersion effect you're after was not a design goal of most stereo recordings. Stereo recordings are usually compressed, EQed and so forth to hell and designed to be played on speakers and in rooms that would destroy the spatial cues you're trying to preserve.
Am I missing something? Maybe the good "Dolby Surround / Pro Logic I/II encoded stereo" recordings... I've no experience with that.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif Of course good phones could emulate this part by themselves, without DH. And they do. That's why some phones are said to have a nicer soundstage than others. [...] So to suit the ears well, it's necessary to simulate reality as good as we can.

I disagree, but that's another topic.

 

[sound_man]

@Henmyr:
Thank you very much for your initial feedback. That seems to indicate that this is applicable to KSC75's, too. Do these phones really sound so much brighter than PPros? I'd like to know if the PPro still sounds so muffled when its directly connected to the X-Fi and all Windows EQs are turned off. Because to me, the overall treble level sounds ok (I'd hate it if the lead singer lisped or the percussion shattered my ear drums!).
But luckily, this can be changed very easily (i.e. with no fiddling). When I'm back from work, I could generate a filter with raised treble (btw: don't try to use foobar's native EQ for this since it's baaad!).
And yeah, I love the sound of Katie Melua on these, too. (And Santana, Dave Matthews Band, Autechre, Goldfrapp, Lamb, Marillion, Pink Floyd, Pothead, The Doors, Blank&Jones etc. etc.

Quote:

Originally Posted by Henmyr /img/forum/go_quote.gif As for improvement. It might be a little too agressive sounding now, and kind of cold sounding. A bit warmer and leaner might be the hotspot. It sounds very good with metal though as it sounds now.

Well... okay! I think this needs to be reformulated into what frequencies should be adapted, and how. But first, a bit more listening might be necessary. This would be an improvement only if it were slightly warmer on a majority of well-mastered albums and if this change is not outweighed by the harm that it may do to the other albums.

Quote:

Originally Posted by HFat /img/forum/go_quote.gif Please tell me if you would like to discuss it elsewhere but one of the reasons I brought it up is that I think that, by mixing the virtualization with the EQ, you're not putting the focus on what you want to perfect: the EQ.

I see your point. Let me rephrase my goal. I want natural/realistic/immersive sound from these phones. It's impossible to get that without something like DH, because too many cues would be missing. DH is not perfect (for this, you'd need to plug in your own Hi-Quality HRTFs and one could argue that 5 speakers are a bit sparse if you could have 24 arranged in a circle around you) but it's the closest we can get in the foreseeable future, so I use it from the start.
This takes us half-way to the finish line. The only link that is missing is the EQ inside this chain (DH->EQ->Koss), which depends on both DH and the Koss. Probably it's very close to how it should be because of educated guesses (headroom plus the thoughts from the first thread), so the only thing that remains to be done is to get the parameters completely right. The end result will be a chain which is as perfect as it can get, while still being ridiculously cheap.

Quote:

Originally Posted by HFat /img/forum/go_quote.gif But the logical conclusion seems to me to be that we should test headphone EQ on binaural samples. Massaging stereo recording to make them sound like speakers (or even surround speakers) on headphones would not be the ideal solution even if the DSP was perfect. Because we're tweaking the reproduction chain, we tend to forget about the recordings but the realistic immersion effect you're after was not a design goal of most stereo recordings. Stereo recordings are usually compressed, EQed and so forth to hell and designed to be played on speakers and in rooms that would destroy the spatial cues you're trying to preserve.

Yes and yes. Though my line of thought here is this:
There are some well-mastered albums. We know that these should sound nearly perfect on a very good speaker setup (e.g. B&W 700's etc.). Our chain should, by definition and philosophy of DH, simulate the sound of such a speaker setup as closely as possible, so the goal which we optimize for is that these albums should sound perfect on our chain, too. If something sounds peaked or notched, consistently over a large number of very good albums, then this is a problem in our chain and getting it right brings us closer to the sound of those speakers. If there's something we cannot fix by EQ, we'd have to live with that for now, but still, the EQ would be optimal in itself, for this purpose.
Getting the EQ of the Koss _alone_ right is another goal, and more difficult, because you'd need a wide range of very good personal binaural recordings (from concerts with very good acoustics), measured with top-notch microphones. Unfortunately, these tiny mics would probably still have more issues than the headphones themselves. And the critical part is that this is not necessarily the optimal EQ for the fully immersive chain, because that depends also on DH's inherent measurement and what-phone-is-expected issues. So you can get the Koss EQ perfectly right, but would still have to fiddle a lot to fit it into the immersive chain.

 

[HFat]

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif I'd like to know if the PPro still sounds so muffled when its directly connected to the X-Fi and all Windows EQs are turned off.

It might be DH that makes the chain sound muffled, you know. I see why one might describe it that way. That's the kind of problem you get when testing too much stuff at once.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif The end result will be a chain which is as perfect as it can get, while still being ridiculously cheap.

I don't see that happening any time soon. Certainly not with the DH we're using.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif There are some well-mastered albums. We know that these should sound nearly perfect on a very good speaker setup (e.g. B&W 700's etc.). Our chain should, by definition and philosophy of DH, simulate the sound of such a speaker setup as closely as possible, so the goal which we optimize for is that these albums should sound perfect on our chain, too.

Maybe my problem is that I don't have any or those good albums. Or maybe it's that I've never heard very good speakers in a decent room. In any case, software DH fails to sound like speakers to me... and, in my experience, speakers can't touch binaural recordings through headphones anyway.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif Getting the EQ of the Koss _alone_ right is another goal, and more difficult, because you'd need a wide range of very good personal binaural recordings (from concerts with very good acoustics), measured with top-notch microphones. Unfortunately, these tiny mics would probably still have more issues than the headphones themselves. And the critical part is that this is not necessarily the optimal EQ for the fully immersive chain, because that depends also on DH's inherent measurement and what-phone-is-expected issues. So you can get the Koss EQ perfectly right, but would still have to fiddle a lot to fit it into the immersive chain.

There may be a misunderstanding: you do NOT want to put a binaural recording through DH.

Whatever problems binaural mikes might have, I've heard binaural recordings by amateurs which are more immersive when played through my soundcard and headphones (without any DSP) than any stereo recording I've heard through any speakers in any room. So I don't get where you're coming from.
From my perspective, stereo recordings are an evil that we must live with, not an ideal we should strive for. Listening to them through headphones can indeed yield even worse results than on speakers but I think that the DH implementation we use is a poor solution that mangles the sound. Sure, I like to use it sometimes despite its problems.

 

[HFat]

But let's try to make progress on the DSP issue regardless. On the matter of Free Surround...

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif I don't see what it takes away from the experience. But I see what it adds.

I don't see what it takes away (except for my CPU cycles)... but, with the default settings, I don't see what it adds either. So what settings do you recommend?

For some reason, it fails miserably at processing the sample I downloaded from Filefront ( http://files.filefront.com/samplemp3.../fileinfo.html )... is that a settings issue?

 

[sound_man]

Quote:

Originally Posted by HFat /img/forum/go_quote.gif It might be DH that makes the chain sound muffled, you know. I see why one might describe it that way. That's the kind of problem you get when testing too much stuff at once. I don't see that happening any time soon. Certainly not with the DH we're using. Maybe my problem is that I don't have any or those good albums. Or maybe it's that I've never heard very good speakers in a decent room. In any case, software DH fails to sound like speakers to me... and, in my experience, speakers can't touch binaural recordings through headphones anyway.

Well, it seems that we're closing in on your primary concerns.
To your ears, Dolby Headphone sounds low quality and you'd go as far as to say that it muffles the music down. I accept that. It's in fact very likely that your ears do not fit the model chosen by them, so everything sounds fake and unnatural. I'm lucky that my ears do fit. To me, DH sounds just as spatial as the surround system in my living room (only better).

Actually, there's nothing inherently low-quality or terribly complex in DH; the principle is simple convolution with a kernel for each speaker that has almost certainly been recorded binaurally. And the combination of DH and EQ is 2 convolutions, nothing I would call "too much stuff at once". Again, the reason why I do it simultaneously is that you can not optimize it separately (see my previous post). Btw: DH does not muffle the sound. Otherwise, the KSC75 would not, in contrast, sound "really good" through it. And even if it did, it could be fixed by the EQ, iff both were optimized simultaneously.

Quote:

Originally Posted by HFat /img/forum/go_quote.gif There may be a misunderstanding: you do NOT want to put a binaural recording through DH. Whatever problems binaural mikes might have, I've heard binaural recordings by amateurs which are more immersive when played through my soundcard and headphones (without any DSP) than any stereo recording I've heard through any speakers in any room.

I think you got me wrong here. Of course one wouldn't send binaural recordings through DH. One would use an EQ, or linear headphones.
The main issue with binaural recordings is that there are so few of them. Sure, I've listened to the Derby traffic recordings, the nightly fireworks, the Ultrasone demo CD, you name it. It's great. If I could get my favourite music in binaural, I wouldn't even think about using Dolby Headphone. Do you know some place/site where I can get enough good albums in binaural to fill my iPod nano? I think not. The Problem is that practically all music is de facto mixed for speakers.
And because of this lack of samples I ask you: How can you assure that your EQ attempts are not biased by some particular microphone's frequency response? Or by bad mixing that was employed in the concert which has been recorded? Or by the possibly bad spot chosen by the listener? Or by the speakers used for playback? That's why I call it even more difficult than getting DH+EQ right. DH+EQ is 1 kind of speakers, 1 listening spot, 1 kind of microphones. For everyone.

To summarize it: If enough of the good music was available in binaural, this thread would be about getting the Koss EQ right, based on several well-mastered binaural recordings. But since 99.9% of the music is made for speakers, and I/we want to listen to that using headphones, I am going to put virtual speakers behind them. If you know of a better, higher-quality technology than DH, preferably with multiple listener settings, available to everyone on the street, please tell me. Btw: Some time ago I've tested the KEMAR HRTFs and they are neat and provide a great choice of subjects etc. But they do not sound nearly as good. And they are probably less universal than DH is.

Finally, I don't want this thread to deteriorate into whether EQ+DH or EQ without DH but later combined with unknown virtualization technology X is the better choice. If you have constructive sound-related feedback, please post it. If you do not see why it is perfectly ok to follow my scheme, then please read about the theory.

 

[Henmyr]

Yes to me my porta pros sound muffled always compared to my other phones. This might be because of a bad repair. Since I can't be sure I will not comment on the porta pro.

From what I can recall, ksc75 sounds just a little less bassy and a bit more natural. When the bassyness is gone from the porta pros then the hights will probably be relatively higher. To me they sound very natural. They have a very good midrange for being this cheap.

 

[HFat]

I've delayed my answer in order to allow the OP to cool down.
Post 8 remains unanswered by the way.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif And the combination of DH and EQ is 2 convolutions, nothing I would call "too much stuff at once".

It's not a technical issue. Of course it works.
What I'm saying is that, if you change several things at once, you won't know which change caused a certain change in the output.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif DH does not muffle the sound.

Muffle doesn't have a definite meaning. This is subjective. These impressions also depend on a lot of factors (including habituation).

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif And even if it did, it could be fixed by the EQ, iff both were optimized simultaneously.

Maybe the whole point of the exercise just dawned on me: so we could fix the problems with (our implementation of) DH with EQ?
Do you have a measurement that could be used as starting point (like the Portapro measurement you used)? Or maybe your current filter already takes DH into account?

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif How can you assure that your EQ attempts are not biased by some particular microphone's frequency response?

How many microphones does one need to get rid of such a bias really?

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif Or by bad mixing that was employed in the concert which has been recorded? Or by the possibly bad spot chosen by the listener? Or by the speakers used for playback?

One would listen to recordings of unamplified performances. Just like one would in any case, right?

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif If you know of a better, higher-quality technology than DH, preferably with multiple listener settings, available to everyone on the street, please tell me.

Some people here have tried inexpensive hardware DH implementations and found them to be good, better than the software implementation(s) they've tried.
I don't trust anyone's judgment by default, I'm just reporting what they said.

Quote:

Originally Posted by sound_man /img/forum/go_quote.gif You know, I have been listening to instruments squashed inside my head for long enough. And I don't really want to defend the wish to end this in every post of this thread.

There's no need to be defensive. I'm simply trying to understand the point of the exercise and what I should listen for before I don my Portapros. At first you said we ought to listen for FR issues and instruments that sound wrong but then I gathered that it's spatial positioning you're really after which is why I brought up binaural recordings: they're the only ones I know which have much in the way of non-arbitrary 3D positioning.

I'm not listening to instruments inside my head by the way... and I'm not using any DSP, exotic gadgets, or even unusual recordings.
But I'm not listening to virtual speakers either: if that's what you want, DH is indeed the only practical solution I'm aware of.

 

[Henmyr]

Quote:

Originally Posted by HFat /img/forum/go_quote.gif I've delayed my answer in order to allow the OP to cool down. Post 8 remains unanswered by the way. It's not a technical issue. Of course it works. What I'm saying is that, if you change several things at once, you won't know which change caused a certain change in the output. Muffle doesn't have a definite meaning. This is subjective. These impressions also depend on a lot of factors (including habituation). Maybe the whole point of the exercise just dawned on me: so we could fix the problems with (our implementation of) DH with EQ? Do you have a measurement that could be used as starting point (like the Portapro measurement you used)? Or maybe your current filter already takes DH into account? How many microphones does one need to get rid of such a bias really? One would listen to recordings of unamplified performances. Just like one would in any case, right? Some people here have tried inexpensive hardware DH implementations and found them to be good, better than the software implementation(s) they've tried. I don't trust anyone's judgment by default, I'm just reporting what they said. There's no need to be defensive. I'm simply trying to understand the point of the exercise and what I should listen for before I don my Portapros. At first you said we ought to listen for FR issues and instruments that sound wrong but then I gathered that it's spatial positioning you're really after which is why I brought up binaural recordings: they're the only ones I know which have much in the way of non-arbitrary 3D positioning. I'm not listening to instruments inside my head by the way... and I'm not using any DSP, exotic gadgets, or even unusual recordings. But I'm not listening to virtual speakers either: if that's what you want, DH is indeed the only practical solution I'm aware of.

Have you even tried this effect with porta pro or ksc75? You seem very doubtful without having tested it.

I have tested it for some time now, but will have to listen more to hear what could improve it. But I do think that it sound very good. Sure, it's not perfect, but with the right music it really brings out enjoyment. Listening to Folsom prison by Cash makes my foot tap more with my D2 + KSC75 + this effect than with my main rig.

Perfect for portable listening.

Keep up the good work sound_man!

 

[HFat]

I've yet to hear the whole thing because I don't know the Free Surround settings you use. But yeah, I've listened to the filter for a short while. I've also been using DH for a while now.

Don't get me wrong: it's certainly good stuff. The reason I keep bumping this is that I'd like more people to try it.
But I don't think there's a free lunch: there are downsides too. A lot depends on the recordings you're using, personal preferences and so on.
What I doubt very much is those claims about achieving a perfect chain and such. Even the idea that a flat FR is important for DH and positioning in general, although sensible in theory, doesn't explain my experiences very well.

More importantly, I'm trying to understand what impresses him so much (and you as well apparently). Sometimes, you can listen for a while and still miss the greatness because you're not listening for the right thing (or to the right thing for that matter).

 

[Henmyr]

What impresses me is the ability to have a very fun and beyerlike soundstage with my portable rig without any apparent loss of soundquality. Porta pros and ksc75 are not reference quality to begin with, so they don't hurt as much when being tampered with.

But everyones mileage may vary.

 

[sound_man]

@Henmyr: Thank you for the defending words!

Quote:

Originally Posted by HFat /img/forum/go_quote.gif I've delayed my answer in order to allow the OP to cool down. Post 8 remains unanswered by the way.

Sorry for the delay. The default parameters for FS are okay as they are, i.e. no tweaking should be necessary. Recommending any albums which sound great with it is hard, because this is a rather subjective thing, but you might want to listen to albums like Blank&Jones - Monument, Tool - 10.000 Days, RMB - Mission Horizon/Widescreen, Lamb - What Sound, or newer pop/hiphop albums if you're after some obvious surround effects.
If you only want to make sure that it works, you can download some PL2 test tracks from the net (e.g. the channeltest.mp3). Apart from this, you can play with the center setting if you're not sure whether that makes a difference.
If you want to take the discussion further, I'd say we should do this in the FS discussion thread because it's kinda orthogonal to the issues which I wanted to discuss here.
Btw: That sample is crap, indeed - I took it offline (maybe it was downmixed improperly).

Quote:

Originally Posted by HFat /img/forum/go_quote.gif It's not a technical issue. Of course it works. What I'm saying is that, if you change several things at once, you won't know which change caused a certain change in the output.

Yes, I second that. Testing strategy is well-known to me.
In our case here we start with a system (the DSP chain) and an initial guess of EQ parameters, based on measurements and assumptions about DH/Koss.
During the tweaking process, we will always change only one EQ parameter at a time (usually the one which is most likely wrong). If the initial parameters were absolutely abysmal, it would take us much longer to converge to the optimal parameter choice - but still, it should not keep us from reaching the optimum (except we give up too early). I think it should be clear that there is actually an optimum in the parameter space; which can be found, consequently. Of course this doesn't imply that it couldn't get any better by switching virtualization or headphones.

Quote:

Originally Posted by HFat /img/forum/go_quote.gif Muffle doesn't have a definite meaning. This is subjective. These impressions also depend on a lot of factors (including habituation).

Ok, that's a language problem. To me, "muffled" seems to indicate that the sound is missing the higher frequencies. Next time, I'll use a dictionary...

Quote:

Originally Posted by HFat /img/forum/go_quote.gif Maybe the whole point of the exercise just dawned on me: so we could fix the problems with (our implementation of) DH with EQ? Do you have a measurement that could be used as starting point (like the Portapro measurement you used)? Or maybe your current filter already takes DH into account?

Yes, I'd say the exercise is to fix the problems of the combination of DH and Koss. Some problems of both subsystems (i.e. wrong "EQ" values for certain frequencies) might cancel out, some others might aggregate. So, after we optimized the parameters, we cannot disentangle both anymore. But we got the optimal parameters for using DH with EQ.
Unfortunately, not all potential problems of DH can be solved via EQ, e.g. DH uses #ears * #channles EQ profiles simultaneously. But only things which are present in all the profiles can be fixed. 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.
Another issue on all profiles might be bad microphone equalization, e.g. small peaks and dips, though this is very unlikely (because they certainly know the FR of their mics and can take care of that).
There may also be non-EQ issues. Namely wrong phase delays and too much echo. Since phase delay varies mostly with head size, there will only be minimal (negligible) inter-subject variability. Therefore, DH should be as good as any other virtualizer in this respect. Echoes could be a problem, but with the DH1 setting, they are luckily kept to a minimum. What we cannot get rid of are the subtle differences between the ear types, so it will always sound bad for some people. That's it, there are no more types of processing that can be represented by the technology used in DH (except for digital noise (shaping) etc. but that's much less important).

Quote:

Originally Posted by HFat /img/forum/go_quote.gif How many microphones does one need to get rid of such a bias really?

The more, the better. If there was an 1:1 chance that any given mic has a peak or dip at frequency X, the probability that this frequency is consistently biased in all mics halves each time you add one to your pool of examples. The same goes for bad equalization in typical CD albums. Except that the chance for peaks/dips is generally not equal, which gives you a slightly slower decreasing exponential function for your probability-of-being-wrong per #-of-mics.

Quote:

Originally Posted by HFat /img/forum/go_quote.gif One would listen to recordings of unamplified performances. Just like one would in any case, right?

Yes, you're right. I was thinking of non-classical music and amateur recordings, where you usually don't have the option to attend a studio recording session.

Quote:

Originally Posted by HFat /img/forum/go_quote.gif Some people here have tried inexpensive hardware DH implementations and found them to be good, better than the software implementation(s) they've tried. I don't trust anyone's judgment by default, I'm just reporting what they said.

First off, since DH is a pure DSP technique and the software version was written by the Lake people themselves, I'd rather trust my common sense. Which tells me that there's not much scope to get an FFT wrong in any implementation. And because the impulse responses are the core intellectual property behind DH, it's very unlikely that they use different responses in different implementations. Though, anyone with a hardware DH box and a computer could verify this by playing a 1-sample click sound (one channel at a time) and comparing the results (digital out is a plus).
Anyways, I searched the forums and the only references I found were of this kind:
"I have not compared the Pioneer phones to any other surround sound headphone systems, but it is far superior to using "DolbyHeadphone" software on a computer using, e.g., PowerDVD or WinDVD. Caveat: I have not listened to movies on a computer using a dedicated headphone amp and quality headphones; only mediocre phones like the Triports through an Audigy 2 PMCIA card."
"I have a couple of probelms with that: 1-no way I use a computer sound card/stock computer/DSP as the heart of my home based sound system.Just not going to happen in my world."


Quote:

Originally Posted by HFat /img/forum/go_quote.gif There's no need to be defensive. I'm simply trying to understand the point of the exercise and what I should listen for before I don my Portapros. At first you said we ought to listen for FR issues and instruments that sound wrong but then I gathered that it's spatial positioning you're really after which is why I brought up binaural recordings: they're the only ones I know which have much in the way of non-arbitrary 3D positioning. I'm not listening to instruments inside my head by the way... and I'm not using any DSP, exotic gadgets, or even unusual recordings. But I'm not listening to virtual speakers either: if that's what you want, DH is indeed the only practical solution I'm aware of.

Okay. Maybe the introductory paragraph should have been like this:
"Hi boys'n'girls! This here is my setup for my Koss PPro, consisting of FS, DH and EQ.
I chose this setup because I'd like to enjoy a big soundstage and surround effects (and I don't see a way to get there with EQ and cheap phones alone, but that's another topic) and want you to take a look into it. The thing that I'm mostly interested in is whether you can spot any peaks and dips in the system's frequency response. These would manifest themselves in wrong sounding instruments or voices, so this is what I ask you to look for. Enjoy!
Btw: If you can change the perceived altitude (or "front-ness") of the instruments to something more pleasant by EQing, please post these findings, too."
Then it would have been clearer that I did't want to have the chain taken apart but instead looked at as an opaque system. But, actually this is not so far from the contents of the first post.

Btw: A very nice way to manipulate the EQ is to append the VST bridge plugin to the chain and tell it to load this EQ Editor. It's a very powerful thing once you've found out how to use it (but the full version is not free, unfortunately).

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, plus two newer experimental filters where I removed a peak at 1KHz and one where I additionally raised the treble a bit, both together with their plots.

Oh, and you say that your phones already give you a sufficient out-of-head soundstage. I'd like to compare some high-end phones like those to my setup, do you know any such phones which I can directly (off-the-shelf) drive with my Audigy card? I'd prefer to avoid the hassle of getting a headphone amp etc. just for a quick test. Price doesn't really matter b/c I'd return them after that.