solderdude
Administrator
measureutternutter
Posts: 4,872
|
Post by solderdude on Jul 27, 2016 9:48:14 GMT
Tyll (inner fidelity) has an impressive amount of plots. I think almost all interesting headphones have been measured. His frequency plots, however, lack a strong correlation between actual plot and perceived sound signature. because his plots are widely used and he is now looking into other 'compensation' curves I thought it would be handy to show how I think Tyll's plots should be interpreted.  When you overlay the green line with Tylls plots and overlay them (use the same scale !) at the 1kHz point in the plot the plots make more sense. That is when you consider the green line as if it was 'neutral' (= a horizontal line in Rtings and my plots). Everything above the green line is a peak above 'neutral' and everything below the green line is softer than 'flat' Below, for illustrative purposes 2 known and considered 'neutral sounding' headphones (HD650 and ETHER with the green line drawn in it on the upper part of the illustration. Below it the original plots but with the 1kHz-22kHz part 'de-tilted'.  Another website that has just begun with measurements and has interesting measurements is Rtings. I did not make a typo here... the expected 'a' (as in R atings) appears missing on purpose. Do note that both Rtings as Innerfidelity use HATS for measurements. This basically means (due to resonances that are as good as impossible to 'compensate' that there are sharp peaks visbile around 10kHz that MAY or may NOT be there. almost all Innerfidelity plots show a peak in that area. Rtings shows lots of high peaks and dips above 7kHz which most likely are not there or lower in amplitude. This peak is in the artificial ear canal and get's 'exited' even when there is no peak there... the air colom inside simply resonates. So when no peak is there you may still see a peak there, unless the headphone itself has a 'dip' there. When the headphone does have a peak there or around that frequency there will also be a peak visible, maybe slightly higher in amplitude or wider in shape. This makes evaluation of the treble quite difficult (impossible ?) as the 10kHz may be there and it may not be.
|
|
Rabbit
Administrator
Posts: 7,087
|
Post by Rabbit on Jul 27, 2016 10:04:29 GMT
That's a really good explanation. Thanks Frans.
The most common thing I see on HF is people talking about rolloffs using Tyll's graphs as proof, with no mention of any compensation curves.
I do wonder though, why the graphs aren't 'straightened out' so that the compensation line IS a flat line. That would make it more logical for many people.
So from 1khz up, the graph is 'bent' upwards, representing neutral in a clearer way.
I'm not sure, but I also think that Tyll changed his compensation curve at one point so old graphs might not be in line with new.
The process of reading and properly understanding these graphs is quite difficult for most. They really have to be interpreted into 'English' or at least words to mean much to many people.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,872
|
Post by solderdude on Jul 27, 2016 12:43:26 GMT
The graphs aren't 'staightened' out because at the time Tyll started with his pdf's he was ill-advised by the guys who installed the dummy head (software) at least that's what he stated a while ago.
A dummy head with ear canal and Pinna needs to be compensated because of ear canal gain and concha gain as well as the angle under which the sound enters the ear. Free field (sounds coming from the front only) isn't the correct correction as it is for a pointsource in an anechoic room. Diffuse field also is not correct as that requires sounds coming from all directions in a certain way. A headphone is delivering the sound directly in the ear canal. There was no 'correction' for it. Perhaps they choose the wrong compensation and then erroniously added 'a specific type of room correction' making it worse.
After many measurements were done Tyll could not change it any more as he would have to redo all the pdf files (seen how many there are)
a year ago he experimented with Olive Welti correction BUT again is not correct as the OW dummy head differs from Tylls. O.W. noticed the 'known' corrections are not the correct ones for headphones and they made their own.
|
|
Rabbit
Administrator
Posts: 7,087
|
Post by Rabbit on Jul 27, 2016 13:24:53 GMT
Well, Tyll did change his 'view' on what constitutes flat quite recently as well. He also seems to like a rolled away treble I think. It's something that he has become very interested in, but it will forever revolve around exactly what 'flat' is as well I think. Perhaps we all hear 'flat' differently although something makes me feel that we possibly 'hear' flat differently but the brain compensates for however our ears work individually since there are often many correlations between people when they describe a headphone sound. So if their ears hear differently, the brain might be translating it back.
Funny thing is that I have worked with live music for a very long time, and yet, I can adapt very quickly to the crappiest sound, Many Hifi types wouldn't tolerate the crappy sound that I can actually enjoy. It has also struck me as quite funny that the many musicians I've known often use the crappiest Hifi stuff quite happily.
I can only think that the reason for this is that we are all used to working in all kinds of environments with their own individual sounds. I've done a lot of outdoor work, with no wall reflections, the Festival Hall which was space dead, the Albert Hall which is like playing in an enormous toilet bowl. (Not so bad with the ceiling mushrooms now though) Also tiny venues - even a library. Fairfield Halls in Croydon which uses electronic room feedback via speakers. Open air festivals with a massively delayed echo. They all sound vastly different and we learn to adapt as musicians to the environment and so adjust our playing.
So perhaps musicians adapt far quicker than many others? One thing that I can't adapt or create from headphones though is any real kind of soundstage. I get left right and middle and if anything, I hear things back into my head, not forward or out. I can't get past the fact that it's two mini speakers on my ears and that's what I hear!!
I guess the idea of 'flat' though is to allow the brain to do less work, which actually helps non musicians more? I can translate the most awful recording in my head to a sound that I can listen to quite easily and perhaps it's not as easy for some, which is why the search for 'flat' is so important.
Funny thing.... I think musicians probably make the worst reviewers. We can adapt more easily and appreciate the good things even from Beats rather than focus on the weaknesses more readily, so everything sounds good to us!!!
|
|
|
Post by lobalwarming on Jun 17, 2017 7:40:50 GMT
Agree, Ian, about many musicians tolerating abominable sound quality. Possibly because they are so far inside the music that mere auditory perception is a trifle. ;-) I'm the exception - as an acoustic musician (who agonizes over instruments/strings/tone from technique etc) - I can't tolerate crappy sound. Especially monitor sound. Used to hate playing the festival circuits because they wouldn't allow me to setup my tri-amped studio monitor setup - small powered studio monitors on stands and a sub. And my mixer. And my eq. And other stuff in the rack. 1 hour soundchecks for 45 minute performances is an ask too far.
If the sound was bad so was my playing - or rather my overplaying. LOL! Sadly, never got on well with in-ear monitors either. Bandmates didn't like my 'isolation', which I loved, who needs timbale poisoning ever night? And because, it was an elaborate sub mix to setup. As well worried about someone tweaking a knob and instant temporary or longer deafness. Plus accidentally pulling my head off at the end of the set walking off stage. ;-)
Conversely, with good open full-sized phones I can enjoy almost as good a soundstage as with speakers. Maybe I have spacial synesthesia. Or, just an active auditory imagination. With a good recording with some room ambiance I can zen-listen into the mix and voila! The headstage is out of my head - maybe only a foot or so, but still can perceive front-to-back depth, that's out front. Sort of like tilt shift photos. The objects are smaller sure, but the sense of space is still present. And the detail from good phones...oh my. However, if I focus on the soundstage or it's just a bad recording, the imaginary sound space collapses to the band-in-the-head mode. It's a subtle thing - can't try to perceive a out-of-head soundstage. It's a process of allowing it to spread wide and deep into my awareness. It's almost like an out-of-body experience, but my crude explanation is more of an out-of-my-mind one. I studied psycho-acoustics at university and my motto was, 'putting the psycho into acoustics' ;-)
I've mixed on phones for years with good results when the canon states that you can't get a good mix on phones. Although there's more examples of bad mixes made on speakers than good. So there's that. And since the Appletic iPoddish Revolution, more people listen on phones - mainly terrible horrible ones - than on speakers. Instead of these endless re-re-re-releases of re-re-remastered classics, why don't they (the Recording Industry), release headphone tweaked mixes as well as speaker mixes? The phone mixes would not require the ear-numbing, soul-crushing make everything-louder-than-everything-else brickwall limited semi-square waving racket. Neither would the speaker mixes, but human nature is involved in the recording arts so...
Digital encoding promised so much for music - less noise, less flutter, less rumble, less distortion with vast dynamic range. Instead it led to the Great Volume Wars, which no one won.
/rant
Thanks for letting me vent. Didn't hear that coming.
Back on topic. Thanks, Frans, for your excellent analysis of Tyll's somewhat dubious measurement process. Dummy head recordings. Dummy head measurements. And consumed by dummy consumers on the internet. We just about have it all. ;-)
Jim
|
|
solderdude
Administrator
measureutternutter
Posts: 4,872
|
Post by solderdude on Jun 17, 2017 8:37:08 GMT
Sam (At Rtings) sent me a link to a video they put up about measuring headphones using a Kemar HATS and what they found.
Explains a lot and is an interesting video worth watching (17 mins)
Fortunately for me these issues do not apply because my seal is correct and don't have to correct the treble for the alterations the earcanal creates. They do have a fake Pinna which I do not have and so final results around 2-3kHz may differ somewhat. They too have a similar conclusion as Golden Ears, Olive Welti and my own 'bass correction' I apply. Mine is very close to that of GE and Rtings, OW is a bit too much. Tyll really should step up his game to get a better correction going.
|
|
Rabbit
Administrator
Posts: 7,087
|
Post by Rabbit on Jun 17, 2017 9:15:32 GMT
For pop type things Jim, I was never even given the choice of monitors. What was set up was what you got!! I played the notes and they set up the sound and it seemed like the two never really mixed too well. Engineers seemed to take care of what everyone heard and we just made sure we played right!!! I was allowed my own iem for monitoring playback but that was always biased towards my voice since in loud stuff, you often can't hear yourself properly in order to sing in tune. I used an ie8 because the treble was rolled away for loud monitoring. Daft isn't it? So the mix sent to my receiver had me much louder than what the audience would hear. That's why we don't really know how we sound to the audiences I suppose!!!
Generally though, the tech people did a really good job and some of them are real experts in sorting out very tricky venues to at least make the sound acceptable whereas if you played a straight show with no engineering, it would probably sound pretty bad.
Orchestral stuff was also quite different for players. From an oboe perspective, I could naturally hear myself much better so all done acoustically, other than dubbing over previous recorded stuff, where you played into a headphone. For me, the volume had to be lifelike so I didn't naturally hold back on the oboe which can make it sound nasty!! So I tended to use something like a dt150 which is rolled away up top so that loud volume isn't painful.
Curiously, I often felt that the final product sounded like a clean, more distant version of what I heard. I think that's why I have a natural preference for headphones with a raised bass. For me, it brings the music a little closer than a thinner, more distant version of some headphones.
At one point, I was really keen to make sure that everything was 'balanced' and all that, but I have got to the point where I'm not quite as picky and go much more readily for a fuller sound than a clinical kind of sound. I find myself using whatever is around and tbh, I don't mind that much nowadays. Maybe I'm going deaf.
Having said that, the headphones that Frans has equalised for me to as flat as they can probably go are very good at most volumes. They are crystal clear at low volume but also go up without nastiness which some headphones don't do well. Probably because they are more even in FR.
I even enjoy the old v6 sound very often and find it perfectly good and yet so many Hi-Fi people hate its sound. I never thought it that bad and funnily enough, tonally, it sounds very good with live broadcast music. I can see why many technicians swore by them actually and don't fully understand why they get so much bashing in audio forums. Maybe I replace the actual sound with 'learned reference' sound - I'm not sure, because I do seem to be able to accept a much wider variation of headphone response than many.
|
|
fanda
contributing
Posts: 61
|
Post by fanda on Sept 30, 2017 14:54:27 GMT
I had the time to go through the full video today. Looks like they are making good progress. I like their concept of frequency response consistency, especially when it comes to how the measurements deviate on humans vs HMS. Although, in the video he states that they didn't test beyond 2 Khz on their people, he does show an example later on (HD 600 vs HD 800) measured on himself vs HMS beyond 2 Khz.
I don't know if the FR consistency graphs shown on their reviews are for 5 different HMS positions or listeners (or both), as they all seem to span the full frequency range. I have asked them that question (they estimate a response time of "4.4 business days" due to other pending questions).
In any case, if that data includes measurements on listeners, while not exactly a proxy for HRTF (which needs measurements to happen very close to the ear drum for accuracy), it could explain some of the preferential differences among different people for the same headphone.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,872
|
Post by solderdude on Sept 30, 2017 19:36:44 GMT
Sam is very thorough in his measurements. He has made quite some progress in all of this. I have been in contact with Sam about his measurements quite a few time and discussed it at length.
His consistency plots are interesting and are mostly caused by people wearing glasses (seal issues) or slightly different headshapes and possibly mic insertion depths as well. But at least you have an idea about the spread.
I think he has made more progress and done more usefull research in a year than Tyll in his whole measurement life.
|
|
fanda
contributing
Posts: 61
|
Post by fanda on Oct 1, 2017 4:35:46 GMT
Rtings: I like what they are doing. They are also clearly trying to be objective, backing their reviews with a lot of data and avoiding flowery terminology.
The consistency plots have piqued my curiosity - will need to go through them in greater detail later. In particular, their "best of" sections allow you to sort through their entire set of reviews using pretty much any column they have data for, which should throw out specific candidates for followup.
Their rating system needs some clarification. Specifically in terms of how they handle lots of peaks vs dips vs more smooth deviation in FR leading to similar deviation and error (as well as the actual range of scores given, which is narrow). This is another question I have raised with them (at the page where they explain the treble response rating).
Innerfidelity: I think he is trying to find/pursue data that fit his hypothesis rather than the other way around. Still, all that measurement data that he has collected is quite useful. I try to interpret his data by looking at differences wrt known reference headphones that he has measured (this should cancel out issues with the measurement rig for most cases), so whatever compensation curve he ends up with is likely to be irrelevant (for me).
|
|
|
Post by ronzo56 on Oct 1, 2017 12:05:32 GMT
Rtings is very thorough with their evaluations. But for the life of me I don't understand the sound ratings on the HD700, AKG701 or the DT990's. But as you have noted Frans. There is a lot of variation in what people find enjoyable, or in this case useful for critical listening. I find these three to have a treble response that is just piercing, almost painful. I owned the 701's after a great review in Stereophile, when they first came out. Gave them to a friend. The DT990's I had to return in a week. Too bright. Perhaps I am more treble sensitive than I think.
|
|
Rabbit
Administrator
Posts: 7,087
|
Post by Rabbit on Oct 1, 2017 12:28:39 GMT
I don't agree with Rting's m50 score either tbh. Reading it had me getting mine out again and to me, they really don't sound that great.
|
|
|
Post by ronzo56 on Oct 1, 2017 17:43:05 GMT
Do I sense "Rabbit and Ronzo's Ratings"? Too mature (spelling intentional) audio enthusiasts give their weekly take on why mid-range frequencies have mysteriously disappeared, headphone makers have lost the ability to accurately reproduce treble frequencies, and sub-bass is king. Get their take on the current state of the "Loudness Wars". Why all sides are losing. Tune in each week as six or seven people find out the same thing we knew last month in hi-end audio trends. (The prices are increasing for the same thing). Special guests, with no measurement experience, will join them to explain what their iPhone tells them about the sound of a headphone they have never actually listened to, as well as sharing their current grade in first year algebra. Join us. And try to stay awake!
|
|
Rabbit
Administrator
Posts: 7,087
|
Post by Rabbit on Oct 1, 2017 18:15:26 GMT
 Aah nose wareye loikes, an' aah loikes wareye nose. Argh, Jim Lad. Pyrate talk eh?
|
|
|
Post by ronzo56 on Oct 1, 2017 18:47:54 GMT
Guess I better get out my copy of Treasure Island.
|
|