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

    Dismiss Notice

To crossfeed or not to crossfeed? That is the question...

Discussion in 'Sound Science' started by jasonb, Oct 21, 2010.
91 92 93 94 95 96 97 98 99 100
102 103 104 105
  1. gregorio
    1. Why? Why repeat anything 100 times here, let alone 100 repetitions of utter nonsense?? For those who apparently haven't read this thread, it's utter nonsense for two reasons:
    Firstly, both music itself and commercial recordings of music are an abstract art form and therefore by definition, do NOT have to comply with reality or "natural levels".
    Secondly, even if music recordings were bound by the rules of reality, it's still nonsense because 71dB's factual assertion of what constitutes "natural levels" of ILD is false anyway! In any room (other than an anechoic chamber) we have a complex interaction between the direct sound and the various sound reflections, which causes an interference pattern and it's this interference pattern that enters our ears. This interference pattern is highly localised (particularly in the lower frequency range) because it entirely depends on the relative phase of the direct and reflected sound, and as we change position within a room, relative to the reflective surfaces and the sound source (say a speaker), this phase relationship changes. The magnitude of these changes is typically around 30dB in an untreated room and even in a very well treated room (such as a commercial recording studio control room), it's still typically around 6dB - 10dB. If you need evidence of this, simply type something like "room frequency response" into Google and click on "Images". The evidence for this being highly localised can be had from anyone who's ever spent more than few minutes in a room with a measurement mic and analysis software; simply take a measurement, move the mic say 6" or so (the average head width), measure again and the two measurements will be significantly different. Naturally occurring ILD in an average domestic room at various frequencies could be as much as 60dB or so, for example in the case of a complete null occurring where one ear is positioned but not at the other. It would be rare to experience such a large ILD in practice, although 10dB or more would be quite common.

    Where then does 71dB get his figure of 3dB ILD being the "natural level" from? One would expect around 3dB for the ILD in an anechoic chamber (where we don't have any reflections which interfere with the direct sound), so how can 71dB's ears objectively "expect ILD of about 3dB" unless he's spent his whole life living in an anechoic chamber and never been in an average domestic room? Additionally, the second reason above is literally "Room Acoustics 101" and 71dB claims to have taken 11 university courses. How is it possible to take even one course in the subject without doing room acoustics 101, let alone eleven? The only logical conclusion is that either he hasn't done any university courses or he's ignoring/discounting what he's been taught. Either way, his objective/factual assertion of "natural levels" of ILD contradicts the science and empirical evidence (that anyone with a cheap measurement mic and free software can verify for themselves) and he's presented no reliable evidence to support his assertion!!

    2. As "natural levels" of ILD in an untreated domestic room can be anywhere from 0dB to about 60dB, then you might have an objective argument for using crossfeed to reduce ILD to 60dB, provided of course that you ignore the fact that music recordings don't have to comply with "natural levels" anyway!
    2a. As we can experience "huge ILD at low frequencies" in the natural/real world (in a domestic room when listening to any sound source), 71dB is therefore effectively stating that the real world is "completely useless"!

    Round and round we go!

    Last edited: Nov 12, 2019
  2. bigshot
    If he's interested in exploring signal processing, he could at least see what's out there. Maybe he'll find something he likes better than what he's looking for.

    The silly thing about this whole argument is that the sound of speakers in a room sounds more natural because it *is* natural. The concept is that you take the canned content and wrap an envelope of real space around it. That makes the secondary depth cues more effective and adds a level of realism to the sound. The sound envelope of the room that engineers expect as a rough baseline for what they intend to create is extremely complex, and it depends on multiple acoustic principles. The sound of the room is just as important to the quality of the presentation as the sound of the speakers themselves. Remove the room from the equation and you're getting half a loaf. That may be good enough for your music listening purposes, and slathering on some signal processing might make it a little better. But it just isn't the same as speakers in a room.

    Headphones can't even come close to the sound of speakers in a room without some heavy duty processing and outboard tracking equipment. I'm not convinced that it can even be completely achieved. One thing I've learned from multichannel is that there is a fractal curve to realism. The more channels you are able to balance, the closer you'll get. But there isn't any number of channels that can completely recreate acoustic reality, and smaller differences sound bigger the further down the rabbit hole you go. In CG animation, there's a term called "uncanny valley". This refers to the difficulty of animating realistic human characters. The closer you get to "real", the more obvious the error is. The best way to deal with it is to not hammer away at realism, but instead find an alternate reality that feels real without being real. This has nothing to do with science. It's entirely psychology and it's hard wired into us as humans. We will prefer distortions that appeal to us subjectively, even when a more accurate, but slightly flawed option exists.

    There is nothing wrong with subjectivity. It's a huge part of how we listen to music. We should embrace it and cater to it. But trying to convince other people that they should share the same subjective preferences because of "science" is a complete waste of time. Subjective preference is a solipsist exercise. It's fine to suggest that other people try processing and see if they like it too, but telling people science says they should like things this particular way is wrong headed in more ways than one.
    Last edited: Nov 12, 2019
  3. gregorio
    I didn't respond to this previously because he's got me on "ignore" but my response *might* be of some use to others:
    The answer to your question is "no", there isn't such a plugin, either VST or any other type. The "stumbling block" is item #2. For such a plugin to exist (that wasn't just marketing BS), there would have to be 2 requirements:

    1. Some sort of objective measurement of the spatiality on the album (actually each track of the album, as particularly with non-acoustic genres, each track is almost certain to have at least somewhat different spatial information) AND,
    2. There would have to be some objective "value" of crossfeed, applied in a variable amount according to the results of requirement #1 (the objective measurement of spatiality on the album/track).

    NEITHER of these requirements are possible, there can be no objective measurement of the spatiality on a track/recording because the spatiality on tracks/recordings is applied subjectively and there is no objective crossfeed value because it's a subjective preference.

  4. SoundAndMotion

    I'm in a good mood now and really don't want to fight with you, but I'm also not afraid to. But, can we avoid it, please?

    Some context for you: no, I have not read all 1500 posts from this thread, nor have I read all of the 200+ from the last month. I check it out off and on, including a few dozen in the last few weeks. You don't have to read my posts, but if you did you'd see I'm not a fan/supporter of everything that 71dB posts (e.g. role of cerebral dominance in crossfeed's appeal for some).

    I don't have to read every post to see the back-and-forth between you and 71. IMHO, it is not helpful to anyone and it should stop, but how? One could ban you or 71 or both, or close the thread, but that would be overkill and I'm not a mod (everyone cheers:wink:), again IMHO. Someone can try to communicate "you're wrong, so shut up", as some have tried (got a mirror?), but we see that is hopeless, and I can't say that to either of you, because I don't believe it. If I wanted to get a bunch of readers to ROFL, I could simply ask you guys to just stop. I figured if he put you on ignore, it would, ever so slightly(!!!) raise the SNR, and help 71dB. He really looks up to you and respects you, which I can see, but also seeks your approval, which IMO is a self-destructive mistake. That motivates his continued efforts to answer you.

    When I said this:
    you responded:
    6- You know I meant in the 71dB side vs. gregorio side, one is not "correct" and the other not "incorrect". I was not referring to "subjective" vs. "objective" sides, also not "true" vs. "untrue" sides. I meant: you both make good points; you both make mistakes and say untrue things(me too); you both cherry-pick (me too); there is no winning side between you. BTW you are not the spokesman for this sub-forum.
    6A- Yes (see above).
    6B- I know that every human on the planet is incompetent in more areas than they are competent. We all hope to be competent in certain areas. I am competent in flying a small Cessna (according to the FAA), but I am incompetent flying a jet... and admit it. When there is an area where we are incompetent, but believe and act otherwise, that's DK, and that's what I said.
    6C- After writing a few posts to 71dB and bfreedma (and mentioning you), I was not asking for a response from you, but I also wasn't surprised. You peppered me with 13 questions (one of your rhetorical devices). If I counted correctly, I answered 7, told you why I would ignore 5, and yeah, I guess I left one of them kind of hanging there... I cherry-picked. Should I claim you lied (another rhetorical method of yours) because I did answer some? No, I don't believe you did. I didn't use your numeration method, and I think you missed the answers within my text.

    I'd guess you've heard of the idea that if cops wants to pull you over, they merely need to follow you for several minutes, and voila, they'll find something. If it is their goal, they might follow you a while to collect a list of infractions, so they can claim you are a dangerous driver, even if it is not true.
    If you want to know how 71dB must feel when you shred his posts, I'd be happy to take your last post to him and shred it for you. Give it the gregorio treatment. I could cherry-pick and ignore your valid points and relevant questions, number each individual mistake, and go on and on about each one. I would then use the sheer volume to call into question whether you really know anything...
    I don't claim your posts are all bad and 71's all good... I see you both having a mixture and not communicating effectively with each other.
    But shredding would be unproductive (unless it helps you to see). A more collegial way would be to cite and expand on your contribution, echo your call for evidence for the 3dB ILD (I would cite Feddersen et al, 1957, but I don't know where he got that), and give a gentle, constructive reminder about frequency-dependence. I would ignore a couple silly little things, because I know what you wanted to say (you're a native speaker though, right?)

    Well I think my goal of being non-confrontational was a miserable failure, but it was my goal... maybe you can help and we can elevate the level of discourse together.
    Hifiearspeakers likes this.
  5. castleofargh Contributor
    let go back to the basics as explaining doesn't work. your claim, your burden. where is the evidence? how many recordings have you done on people to claim objective improvement? I'm guessing you did a great many as your claim isn't limited to potential improvement(something I actually would agree with objectively, as the potential at least is there), you even argue that crossfeed is always an improvement as long as it goes in "the right direction". to be able to claim that, you must have some huge study right? you wouldn't make objective claims about stuff you have never confirmed objectively, right? not in Sound Science.

    there is no "we should". that's my point. if someone is used to headphones, maybe fixing nothing will feel more natural to him than any crossfeed. and maybe some settings won't feel right for someone, and others will. which is what most people experience, @ironmine showed that many times with his feedback on various plugins.
    I just try to demonstrate that the alleged objective argument about cherry picking one variable from speaker playback and shoving a vague version of it into headphones out of context does not necessarily lead to a subjective improvement, even if it does bring that variable closer to our reference model of speakers. just adding some amount of reverb will rapidly feel weird to most people on most musics, and IMO that disproves part of your reasoning supporting crossfeed as more than something some people will like.
    but really the non objectivity of it all is already in your post. you find the lack of left channel reaching the right ear and vice versa very objectionable, and claim that anything dealing with that in some way is an improvement. but then you just said that you like the missing room reverb on headphones. so crossfeed really just happens to serve your subjective preferences, or maybe you got so used to it that now it defines your taste?

    we have endless examples of speaker simulations, room simulation(all the surround stuff over the years for headphones, and the many crossfeed and more as VSTs). they all pick some variables and try to make them more like speaker playback. so by your logic for crossfeed's justification, all of those could be declared objective improvements. but they're all subjective tools, aimed at fooling the brain. sometimes for some people it will work pretty well, and for most people most of the time, it will just feel weird because those simulations are only taking care of some variables. and do it in a way that might end up close to someone's HRTF(lucky!), or pretty significantly far from it or whatever room the guy is used to. all the so called objective approaches to those processes result in inconsistent, and in some aspects, unpredictable impressions for the listeners. you don't seem to get how dynamic the interaction between stimuli really is when the brain creates an interpretation from them. if something is missing or if something is different enough from whatever learned "normal" to be felt, you can be pretty sure that it will affect a lot more than just the basic impact from the one variable in total isolation. and we have countless demonstrations of that.
    the most obvious example of an objective approach that simply doesn't agree with the brain despite being objectively very solid(something I can't say about your crossfeed argument),is IMO, the diffuse field compensation for frequency response graphs with headphones. they decided on what the reference sound should be, and went to see how that would look like as FR on a headphone for the average human head. that in principle, should lead a majority of people to find that a flat line in a diffuse field compensation sounds neutral to them(averaged for all direction, using calibrated speakers, all should be good, objectively when strictly considering that diffuse field model!!!). and yet, in practice almost nobody feels that diffuse field is either neutral or enjoyable on headphones. it's been annoying the headphone community for ages. so what gives? IMO the answer is obvious, even something like our impression of neutral involves all stimuli perceived by all our senses, our experiences, our expectations. the all package. and not just the FR at the eardrum based on one model of sound that nobody really gets with their speakers in the first place. making a model that doesn't count everything, cannot guaranty that it will be correct just because we follow the steps after the first few erroneous assumptions. it's true objectively, it's maybe even more true subjectively. and it should remind you of a certain crossfeed argument we're having.

    that just makes no sense to me. we agree that an album will most likely have been mastered with and for stereo speakers. so if you consider applying a different compensation on some albums for alleged objective reasons, you're treating the sound engineer's artistic intent as an objective mistake. which is nonsense, end of story.
    I get that you could prefer a different crossfeed setting on some specific tracks, just like one might want more bass on some tracks, because it feels better to you that way. but the objectivity behind it... get out of here.
  6. 71 dB
    I ignore gregorio nowadays. so I wouldn't know what he calls for. The 3 dB ILD at low frequencies is a ballpark value based on my insight of the subject. You like 4 dB better fine, but that's more or less where ILD at low frequencies are. I believe I have previously (2 years ago) mentioned it's the mean value of soundwaves from all directions. I was criticized over room modes, but it was ridiculous. I studied control of room modes for Genelec in a 4 year project and I have my name on patents regarding the issue so I have "some" insight about how low frequencies behave in a room. I'm pretty sure records are mixed in studios with acoustic treatment that keeps modes in some sort of control so arguing headphones should imitate rooms with lousy acoustics at low frequency and massive modes is totally ridiculous.

    Theoretically it's possible that if you have one ear EXACTLY at the bottom of a room mode notch and the other ear is about 17 cm from that point, you can have significant ILD, but this is ridiculous point because: Such listening position/situation means at that frequency the sound is CRAP! You are kind of supposed to do something about it. Move your head to find a better sweetspot and/or modify the acoustics of the room so that the modes are less severe. Also, when these larger ILD situations happen, the sound pressure level is very low at that frequency. The other ear hears next to nothing, and since the other ear is also near the bottom of the notch, it experiences a lowered sound pressure level also. The frequency in question is likely to be masked and also since the equal loudness curves at low frequencies are closer to each other and the threshold of hearing is high, it's likely both ears are UNDER the threshold of hearing/masked so that ILD doesn't matter at all! It is not even heard! Even if it barely heard, it's contribution to "excessive" spatiality remains more or less insignificant. Room modes are typically an issue under 200 Hz. Above that the modes become so compact they are considered reverberation rather than individual modes. Also, the room absorps more sound so the modes became less severe. At 800 Hz crossfeeders typically have a -3 dB point meaning 3 dB more of ILD and so on...
  7. 71 dB
    People hardly ever hear the music the way sound engineer's artistic intented it. What the engineer heard in the studio is what he/she intents. People have different speakers and rooms so even the speaker sound is different from what is heard in studio. Of course the rational intent is that it sounds good so engineers learn to mix so that it sounds good on different kinds of speakers and less than optimal acoustics. Headphones have always been secondary in this. It's not "making mistakes". It's traditions. Recordings have different levels of ILD. They don't look the same on Goniometer. I hear it and I have analysed recordings so I know this as a fact. On speakers this doesn't matter much, because speakers in a room are effectively ILD regulators, but headphones are not! Regulating is needed and I do it using a crossfeed with adjustable level.
    Last edited: Nov 13, 2019
  8. 71 dB
    I simply do not have evidence that is good for you, only evidence that is good for me. I am tired of fighting over my credibiity. If you think I know nothing then think that! I don't care. Do I put you on ignore too?
  9. 71 dB
    Fine, but such subjective opinion doesn't refute my objective claims that since headphones give often significantly larger ILD at low frequencies than speakers and recordings are mixed for speakers, the conclucion is that scaling ILD with crossfeed is objectively justified.
  10. 71 dB
    IF headphone spatiality was half-decent, improving it with simple methods would be challenging, but since it is complete crap (with recordings mixed for speakers) cherry picking is fine and an improvement. I am totally tired of this cherry picking thing! I have said several time I KNOW I am doing it and I know what I am ignoring and what it means! So stop it!
  11. 71 dB
    Crossfeed is VERY close to HRTF compared to no crossfeed! No crossfeed is as far from HRTF as you can get! Compared to that crossfeed is pretty close.
  12. 71 dB
    You people are constantly refuting my objective claims using subjective opinions while telling my own subjective opinion are worthless! Do we talk about subjective opinions or objective facts? You can't refute my objective cherry picked claims using (also cherry picked) subjective opinions (some people don't like crossfeed etc.)

    Seems I need to ignore 80 % of people to not lose it here! So tired.
    Last edited: Nov 13, 2019
  13. bfreedma

    The problem is, you don't have objective data that addresses the full issue and simply cherry pick for convenience to support your subjective views.

    Ignoring everyone is an option, just not one well aligned with Sound Science discussions. If you're going that route, I think a better option is to move the conversation to another sub forum.
    sonitus mirus likes this.
  14. 71 dB
    How are we supposed to discuss this issue objectively if the objective data doesn't even exist? What is the "full issue"? How do you want to define it? I have said miilion times my "cherry picking" is based on the assumption ILD is the major problem in headphone spatiality and ignoring other aspects doesn't nullify the benefits of fixing ILD.
  15. castleofargh Contributor
    you do it on your own, stop that victim crap. we're merely the reactive compound here. when you feed Sound Science with empty or false claims, we react. and if you post them over and over again, the more unstable elements of the forum will explode. it's a well known chemical reaction. anybody spending time in this section is fully aware of it. yet here you are, spamming the same claims you have no business claiming.
    all it takes to settle the matter if for you to either provide objective evidence for your objective claims, or to, wait for it, just stop posting the same empty claims in this section like a spam bot. for now 2 years, you've provided supporting evidence that you can't do either.

    PS: usually when we get posting sprees, and we see the legendary "you people" pop up, that poster is on his way out. often of his own accord, sometimes not. <= that's the modo in me talking.

    congrats, you just explained why you should not make those objective claims! \o/
    to discuss objective stuff without enough data, we have ideas and hypotheses, you can indulge in those as much as you want. we can make assumptions and wonder where it would lead us. and maybe when we experiment, we can check how that holds up. it might not be proof but it could suggest we're on the right track(or not). what we do not have, is the right to make objective claims without the objective facts to demonstrate that claim to be true. it's that simple. I say we don't have the right, but of course I mean rationally. in practice Headfi didn't give this section the rules needed to enforce the very methods of science. so people post whatever and in this section, we react.
    gregorio and sonitus mirus like this.
91 92 93 94 95 96 97 98 99 100
102 103 104 105

Share This Page