CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => IEM Measurements => Topic started by: speakerphone on February 09, 2015, 05:45:11 PM
-
(http://4.bp.blogspot.com/-TuvKrt5ESgM/VJF_8a8xWTI/AAAAAAAAAMM/I936tlJZRxI/s1600/raw.JPG)
(http://1.bp.blogspot.com/-ZX8bAktMRQw/VJF_6-r0ZiI/AAAAAAAAALw/_BBA3YKpVwk/s1600/df.JPG)
(http://4.bp.blogspot.com/-rlnhVcDXjVQ/VJF_7rY214I/AAAAAAAAAMY/7Z3k4MQryuE/s1600/ow.JPG)
(http://2.bp.blogspot.com/-N2wJPYNFEnc/VJF_7ObyYkI/AAAAAAAAAL0/rv2dvP73ahc/s1600/csd.JPG)
(http://1.bp.blogspot.com/-73AJxh4WtV4/VJF_9G-rtRI/AAAAAAAAAMQ/Ya1GlKyV4f0/s1600/thd.JPG)
(http://2.bp.blogspot.com/-3TPm_Ktz4Qk/VJF_7wL666I/AAAAAAAAAMA/N_Xupw0zpAE/s1600/phase.JPG)
(http://1.bp.blogspot.com/-BejawN1tg4A/VJF_7bwdSnI/AAAAAAAAAL4/i2HhA-HHuEs/s1600/impe.JPG)
+All of my measurements are 94dB(+-0.5dB) sound pressure level(SPL) matched at 500Hz according to IEC 60268-7.
My artificial Ear (Coupler) follows IEC 60318-4 (former IEC 60711), ITU-T Recommendation P.57, ANSI S3.25 standards.
About the measuring standards
I mostly follow the IEC 60268-7 standard. I use O2 Amp for Lower the output impedance close to 0ohm. (In IEC standard, amplifier output impedance should be 120ohm.)
Also I use [1/24 octave smoothed Pink periodic noise] for frequency response analysis, and [1/6 octave stepped sine] for harmonic distortion analysis.
Diffuse Field Compensation Target follows the ISO 11904-2 standard, and Olive-Welti Target follows latest In-Room research. (Olive, Sean; Welti, Todd; McMullin, Elisabeth at AES).
But, as we know, frequency range of ISO 11904-2 standard is 20~10000Hz. So I dealt [-6dB/Oct.] at 10000Hz~20000Hz.
References
IEC 60318-4 : Electroacoustics - Simulators of human head and ear - Part 4: Occluded-ear simulator for the measurement of earphones coupled to the ear by means of ear inserts
http://webstore.iec.ch/webstore/webstore.nsf/artnum/043703!opendocument
ITU-T Rec. P.57 : Series P: Telephone transmission quality, Objective measuring apparatus : Artificial ears
http://www.itu.int/rec/T-REC-P.57
ANSI/ASA S3.25 : Occluded Ear Simulator
http://webstore.ansi.org/FindStandards.aspx?SearchString=s3.25&SearchOption=0&PageNum=0&SearchTermsArray=null%7cs3.25%7cnull
IEC 60268-7 : Corrigendum 1- Sound system equipment - Part 7: Headphones and earphones
http://webstore.iec.ch/webstore/webstore.nsf/artnum/043714!opendocument
ISO 11904-2 : Determination of sound immission from sound sources placed close to the ear
http://www.iso.org/iso/catalogue_detail.htm?csnumber=32439
Olive-Welti Target : Listener Preferences for In-Room Loudspeaker and Headphone Target Responses
http://www.aes.org/e-lib/browse.cfm?elib=17042
Feel free to ask a question about this!
-
Cool!
BTW that 9KHz spike looks painful.
-
I don't think OW correction applies for IEM measurements because part of the 'ear canal' is blocked by the IEM.
The OW curve is for on-ear/over ear headphones AFAIK.
When you sweep it by ear do you hear the resonance at 9kHz ? (the 16khz will be more difficult to hear)
A sharp 12dB peak that rings would not be pleasant to hear IF the music has those contents.
-
I don't think OW correction applies for IEM measurements because part of the 'ear canal' is blocked by the IEM.
The OW curve is for on-ear/over ear headphones AFAIK.
When you sweep it by ear do you hear the resonance at 9kHz ? (the 16khz will be more difficult to hear)
A sharp 12dB peak that rings would not be pleasant to hear IF the music has those contents.
Olive-Welti is not just for headphones. There is no reason for not using OW target for IEMs. These measurements are just a recording of eardrums, no matter that is earphone or headphone, blocked ear canal(IEM, sealed headphones) or not(opened headphones).
http://www.innerfidelity.com/content/headphone-measurements-explained-frequency-response-part-one
the new article about the frequency response in innerfidelity described it well.
And about the 9kHz ringing, it is not that audible for me. Comply foam tip reduces these spikes. (This measurement is with silicon tip)
-
(http://4.bp.blogspot.com/-9jXME2o2ksw/VJGmw-AAymI/AAAAAAAAAM8/Ngr8pxn9F50/s1600/90foam.JPG)
As you see, the spikes are gone when use Foam tips.
-
Olive-Welti is not just for headphones. There is no reason for not using OW target for IEMs. These measurements are just a recording of eardrums, no matter that is earphone or headphone, blocked ear canal(IEM, sealed headphones) or not(opened headphones).
http://www.innerfidelity.com/content/headphone-measurements-explained-frequency-response-part-one
the new article about the frequency response in innerfidelity described it well.
I disagree.
The correction curve needed for measuring coming from the FRONT of the head is VERY different from sounds coming from the side of the head (headphones).
So.... HOW can you be confident to use a correction that assumes sounds have been boucing the concha where in IEMs there isn't a Concha and the effect of that differs again with on/over ears ?
When you take measurements with different insertion depths the plots differ.
Which depth is correct for OW and have you removed the pinna flange, torso and concha influence (leaving only the, partial, ear canal)
-
I disagree.
The correction curve needed for measuring coming from the FRONT of the head is VERY different from sounds coming from the side of the head (headphones).
So.... HOW can you be confident to use a correction that assumes sounds have been boucing the concha where in IEMs there isn't a Concha and the effect of that differs again with on/over ears ?
When you take measurements with different insertion depths the plots differ.
Which depth is correct for OW and have you removed the pinna flange, torso and concha influence (leaving only the, partial, ear canal)
I think so that you are talking about the Free Field equalization. It is not a FF. Diffuse Field equalization is average of all directions of the sound. (Like in the reverbation room).
So, Diffuse Field already has those pinna, torso, etc effects.
(http://www.innerfidelity.com/images/Headphone101_InterpretingFrequencyResponse_Graph_ContributorsToTargetResponse.jpg)
And Olive-Welti Target is same as Diffuse Field + Room gain. Olive-Welti is not a special target. It's just modified Diffuse field Target.
You can reference the Rin Choi's blog.
http://rinchoi.blogspot.kr/2010/05/headphone-equalization.html?m=1
And Rin choi also uses Olive-Welti Target when compensating IEM.
-
I am aware of the theories that float around.
DF or FF cannot be used to compensate the ear canal only, no matter how you bend the theory or feel it is applicable because others use it as well.
The average (DF) differs from FF but BOTH take torso & neck, concha and Pinna in their weighting which are of NO influence in IEM.
Thus adding a correction for torso & neck, concha and Pinna while it isn't even there is a bit fishy IMO.
FF, DF OW and other room corrections are still quite different from 'only' a (partial) ear-canal.
It's one thing to own a real (calibrated) coupler but when you apply the wrong compensation it is rather useless.
Anybody know what happened to Rin btw ?... looks like he pulled a Voldemort.
-
If speakerphone is Rin all I can say is welcome! ahoy
If not, all I can say is welcome! ahoy
Here is to kool times ahead! :)p5
-
I am aware of the theories that float around.
DF or FF cannot be used to compensate the ear canal only, no matter how you bend the theory or feel it is applicable because others use it as well.
The average (DF) differs from FF but BOTH take torso & neck, concha and Pinna in their weighting which are of NO influence in IEM.
Thus adding a correction for torso & neck, concha and Pinna while it isn't even there is a bit fishy IMO.
FF, DF OW and other room corrections are still quite different from 'only' a (partial) ear-canal.
It's one thing to own a real (calibrated) coupler but when you apply the wrong compensation it is rather useless.
Anybody know what happened to Rin btw ?... looks like he pulled a Voldemort.
I can't catch your opinion... You are saying that IEM doesn't need compensation?
IEM should compensated as a Diffuse Field "BECAUSE" torso & neck, concha and Pinna aren't influence in IEM.
It is sure that DF isn't the answer, but it's the best way until now.
If speakerphone is Rin all I can say is welcome! ahoy
If not, all I can say is welcome! ahoy
Here is to kool times ahead! :)p5
Nice too meet you! Sure I'm not a Rin! He is the best reviewer I'd ever seen.
-
LOL! you copycat you.
Anyhow, IEM measures seem pretty hard to make. Do build on the work of others, but IMO, don't take it for granted.
-
I can't catch your opinion... You are saying that IEM doesn't need compensation?
Of course the raw plots need compensation ... but NOT the ones that apply to the outer ear/torso (DF, FF, OW) as that is not in the equation.
OW for on-ear/over ear headphones is more correct than DF/FF in any case but does not apply to IEM.
You will ONLY need to compensate for the earcanal/eardrum part.
That compensation should be supplied with the coupler itself.
Then again compensation also will depend on insertion depth as well.
You need different compensation for different depths (in theory) as Rin has also shown.
Its the reason why the IEM plots here are all raw b.t.w.
I have to admit that this is very confusing and has no relation to what's perceived though.
Do 'we' know how much spread there is in human ear canal widths, lengths and geometry and how a simulated coupler comes close to this ?
Can there be a 'correct' compensation if insertion depth is specified and can easily be checked ?
-
As you know, earphone's final goal is simulating stereo loudspeakers.
We use pinna for listening to music with loudspeakers. But IEM doesn't need pinna. So when listening to IEM, we have to simulate pinna on purpose(to hear the sounds like pinna does). That is what DF means.
So, we have to use DF(simulates pinna) because IEM is not using pinna (torso & neck, concha and Pinna).
Also headphones should use DF too. Because the measurement is just a recording of eardrum, so real pinna actually boosts 3kHz of headphones, and can make DF equalization.
Of course DF doesn't simulate actual pinna when measuring headphones, but simulates dummyhead's(in DF target) pinna.
-
It will depend on what you want to show with the plot(s).
If you..
A: want it to show the actual SPL that we perceive/hear
B: want to see a 'flat' line in a plot as if an 'ideal speaker in a conditioned room' would have sounded like.
If its 'B', use all the room compensation you like, as it is unrealistic and a bad representation of the average living room anyway so won't produce correct plots but will be comparable IF it is clear what compensation is used.
personally I prefer to see plots of the first kind (A)... the ones that show actual SPL as that correlates better to me on how they are perceived.
Closer than other compensations. This may have something to do with training.
I know I perceive headphones differently than your average youngster listening to popular crappy loudness warred recordings.
What sounds real to me may sound bassless, midrangy and lifeless to them.
If you want to show how it actually sounds you will have to compensate the (shortened) earcanal only in case of an IEM.
That will thus differ from the correction that comes with coupler and is made to compensate for sounds outside of the coupler.
OW is obtained by 'averaging perceived sound by multiple people with multiple headphones' and comparing/correlating those 'impressions' to ONE specific dummy head/coupler yielding a compensation curve for that specific type of dummy head.
This headphone FR spectrum will differ from the FR spectrum an IEM generates, so using OW for headphones makes sense (when derived from headphones) but NOT with IEM's
That's my POV though, others may disagree entirely..
as a reference Personal audio ru's interpretation of the Tio (but alas heavily smoothed, I much prefer less to no smoothing)
(http://personalaudio.ru/raa/otchety/naushniki/ultrasone-tio/Ultrasone_TIO_fr_impedance.png)
-
If you want to show how it actually sounds you will have to compensate the (shortened) earcanal only in case of an IEM.
Nooo... That's not the fact... I can't understand why you are thinking like that.
There is no one who compensates only ear canal response on IEM...
Then you are thinking that whole measurement sites(ex: innerfidelity, headroom, Golden Ears, rin choi etc.) are compensating the IEM wrong way?
-
Did you ever actually compare plots/measurements from Tyll (be it headroom/Innerfidelity) plots with those from Golden ears and Rin's ?
And I don't mean the raw data but the actual 'perception' plots ?
Did you ever find any quite resembling plots ?
Did these plots match within 1 or 2 dB ?
How well did any of those plots completely correlate with 'general' or your impressions ?
Do you think this may be caused by different test methods/compensation curves ?
Which one is more correct in an absolute sense ?
So... I guess the sites you mentioned don't only compensate for the ear canal alone 'cause if they did their plots might have been more similar (the raw plots seem to be).
Just as some of your plots are all derived from the same measurement but with different compensations ?
Which one would you say is closest to how you perceived them ?
Do others perceive the same items the same way ?
Do they use the same tips ? the same insertion depth ? Do they get the same seal ?
This isn't criticism on your measurements b.t.w. and think your measurements (the RAW ones) are accurate and nice.
I think its great having someone with a real coupler and this may help other DIYers a lot.
My remarks are here to make you think about it and not just follow paths others already tried, and perhaps regard as 'truth' because noted person this or that said so.
A bit like Alex (AZ) who also invented his own method. It remains to be seen who is 'righter' but at least he walked another path and thought out of the box.
Only out of the box gets us further.
I am thinking this way because, other than audio, do a lot more measurements which have to be compensated for charateristics of the transducers / circumstances and only use compensation that is needed in order to get an accurate response, not so it can match someones idea of how the results should be.
In case of the IEM only the (partial) ear canal is in the way between the SPL coming out of the nozzle and your eardrum.
Perhaps an over simplification but where is that wrong... assuming I want to plot the actual SPL and not what some 'speaker in a room' and my head/pinna which aren't in the picture at all make of it ?
-
Did you ever actually compare plots/measurements from Tyll (be it headroom/Innerfidelity) plots with those from Golden ears and Rin's ?
And I don't mean the raw data but the actual 'perception' plots ?
-> Yes. Those sites uses their own compensation Target.
Innerfidelity - ID Target of Head Acoustics HMS
Golden Ears - B&K DF + room gain + 6dB effect compensation
Rin - ISO 11904-2 & olive welti target
Did you ever find any quite resembling plots ?
-> Yes these measurements resemble in characters.
Did these plots match within 1 or 2 dB ?
-> sure it is not. The compensation Targets are different.
How well did any of those plots completely correlate with 'general' or your impressions ?
Do you think this may be caused by different test methods/compensation curves ?
-> Yes.
Which one is more correct in an absolute sense ?
-> Olive-Welti Target in my opinion.
So... I guess the sites you mentioned don't only compensate for the ear canal alone 'cause if they did their plots might have been more similar (the raw plots seem to be).
Just as some of your plots are all derived from the same measurement but with different compensations ?
Which one would you say is closest to how you perceived them ?
-> that would be differ to person. But I prefer OW target.
Do others perceive the same items the same way ?
-> of course not.
Do they use the same tips ? the same insertion depth ? Do they get the same seal ?
-> of couse not
--------------------------------------
The coupler can't simulate the real human ear, but resembles.
We can't find the perfect compensation target for their own ears. Because human HRTF is different with each other.
I want to measure the IEM in Internationally Verified way. Not my way.
That's why I reference the IEC standard, ISO standard and Target which verified in AES.
-
The coupler can't simulate the real human ear, but resembles.
Yes it does, but the coupler without a pinna is nothing more than a waveguide.
When you compensate for the wave guide you end up with the real SPL at the entrance of the waveguide.
The Pinna and the rest is not involved so why compensate it ?
We can't find the perfect compensation target for their own ears. Because human HRTF is different with each other.
Then why bother trying to compensate the measurement anyway with a method you prefer as 'best fitting' ?
I want to measure the IEM in Internationally Verified way. Not my way.
That's why I reference the IEC standard, ISO standard and Target which verified in AES.
Then you need to create at least 4 plots and neither of those plots will actually show what came out of the IEM.
You should post the raw plot, the IEC standard, the ISO standard, any AES verified standards (OW) and that's it.
That's what you've done so far and that's a good thing.
I merely placed some question marks regarding the 'standards' and hoped/thought you would be willing to think out of the familiar boxes and be able to come up with something refreshing like O & W research.
IF I were ever to measure IEM's (they don't interest me so I won't) than I would try very hard to come up with a more accurate compensation than the ones others use for their own reasons to try and get better correlation to reality.
See what pirates think of those graphs and if they correlate to their findings.
That's what I am trying to convey.
THINK about what happens in the coupler, WHAT you want to know, WHY there are different standards that come up with different results and investigate WHY they differ and see the logic and possible flaws behind it. Why would a speaker sound be preferable and why one should compensate for things that aren't there ?
The OW curve is designed to compensate for measurement 'errors' that are not applicable in this case and even though it may be closest to the way you perceive them, it cannot possibly be the correct compensation regardless if others think so.
Why not ask the OW guy, he is active in several forums and can tell you if you can apply the OW target directly to an IEM inserted to a coupler.
I understand you don't feel the need to research this and wonder why, as you seem to be someone that likes to measure and likes to find the correlation between perceived sound and raw data and thought thats why you joined.
If you just want to do what others have done before and come up with identical results (which don't seem to be fully identical) than that's an option and IMO a missed chance to actually improve on creating more accurate plots.
Willing to bet there are a lot of pirates here willing to think and debate it with you.
Lots of peeps here with IEM experience and a few guys that measure them (not acc to exact standards)
-
The IEC standard is the coupler specification itself, it defaults to ISO 11904-2 for the DF equalization target of headphones and IEMs.
Olive/Welti noted that people prefer different amounts of bass and treble quantity in speakers and in headphone after the headphones were compensated to to an in room response on a HATS (which appears very close to DF anyway). If we can trust Tyll to paraphrase Olive correctly, the listener preference in IEMs is also different.
-
That's my point.
ISO 11904-2 describes a manekin based measurement method and is just like OW for measurements of drivers CLOSE TO the ears, not partially shoved inside the ear canal.
Other corrections apply in this case as the Pinna, Concha and other aspects do not apply anymore therefore don't need that compensation.
To obtain the SPL from the IEM port you only need to correct for the part of the artificial ear canal between the sound source and eardrum which differs from the ear canal in the coupler itself.
Rin also has touched this subject if I remember correctly and wrote about insertion depths and its influence.
-
I know what you guys talking about, but my measurement data can't be the actually what you guys exactly hear. Measurements is far from subjective hearings according to IEC 60268-7(It really says).
Because coupler can't be the human eardrum.
But,
However, there is no difference with equalized headphone(OW experiment)&IEM for the coupler(not for human) until ear canal & headphones are Minimum Phase. Blocked ear canal doesn't matters. Your opinion that IEM can't use OW is absolutely wrong.
IEM can compensated by OW.... Yes it is.
-
And I'm kind of tired to do this. I want to rest...
My summary:
1. IEM = headphones on HATS and can compensated by OW.
2. Measurement data is different with what you really hear.
-
1: You are absolutely right ! O&W CAN be used for IEM... they are a perfect match. I just wish everyone applied it just because we can !
2: Indeed, why should measurement data and sound have to correlate indeed, I don't see any point/value in that.
-
sarcasm.. I wish you can understand my situation.
Best way I can do now is follow the Internationally Verified Ways. Because this data is not just for me.
Thanks.
-
SOlderdude, is the O+W canal compensation not suitable for IEM measurements? You said that IEM measurements here are raw, why wouldn't there be any canal compensation for that?
-
Probably because nobody bothered to create a reasonable compensation (IMO).
To do this you would have to have a 'standard insertion depth', tube diameter and length and calculate what the frequency response of that small piece of pipe (which is sealed on both ends).
There is too many variables to match it to how it sounds (as Speakerphone also pointed out) because of all the things involved.
As there are already standards it feels logical to follow them.
However, if the couplers are are closely the same it makes sense to publish the raw plots as these can easily be compared.
They can also be compared using 'standard' compensations but one should pick just one.
On the other hand I am a strong proponent of creating a 'standard' using a transparant 'tube' with depth markings so all inserted IEM's actually measure and can be compensated the same way.
Then the next step would be to try and correllate with sound.
This is almost impossible and one of the many reasons I don't use nor care about IEM's.
-
Probably because nobody bothered to create a reasonable compensation (IMO).
To do this you would have to have a 'standard insertion depth', tube diameter and length and calculate what the frequency response of that small piece of pipe (which is sealed on both ends).
There is too many variables to match it to how it sounds (as Speakerphone also pointed out) because of all the things involved.
As there are already standards it feels logical to follow them.
However, if the couplers are are closely the same it makes sense to publish the raw plots as these can easily be compared.
They can also be compared using 'standard' compensations but one should pick just one.
On the other hand I am a strong proponent of creating a 'standard' using a transparant 'tube' with depth markings so all inserted IEM's actually measure and can be compensated the same way.
Then the next step would be to try and correllate with sound.
This is almost impossible and one of the many reasons I don't use nor care about IEM's.
Why is compensation even used? You are still measuring at the eardrum, whether it's an IEM or a headphone, so the raw data should be close to what is being heard. And compensation is the same for each test, so the value at each frequency is being compensated the same at each point, it's just shifting the FR a set amount each time, why do it at all? I could understand why you would use compensation if you are trying to compare headphone results to speakers, but if just comparing headphones to headphones isn't compensation not needed?
-
the 'signal' at the eardum differs substantially from the signal that enters the ear canal.
We actually hear 'better' at certain frequencies which only becomes obvious at lower levels, check the phon curves.
To put it another way...
When we listen to a piano its sounds reach the entrance of our ear canal (more or less bounced of the Pinna/concha depending on the sources position)
When we would place a microphone in the ear canal entrance the FR differs from the FR meaured at a few cm from our heads.
When we would place a microphone inside our heads replacing the eardrum we would measure a signal that differs substantially from that on the entrance of the ear canal.
Try listening through a thin tube and you'll hear it immediatly.
The sound on the outside of that tube thus has a very different FR from what's measured with an equally flat microphone on the inside at the other end of the tube.
The coupler used for IEM measurements has a mic that is positioned at the end of an open tube (open at one side)
Well not exactly like an ear canal which isn't actually a straight round tube but it bends, gets smaller and then bigger again and the eardrum is under a slight angle.
The coupler is just a small fixed diameter round tube with a mic on the end.
So when you want to find what how the outside 'sounds' like you have to undo what the tube has done to the sound, it has altered the FR (considerably).
That's the compensation needed son SPL on the outside is a copy of the inside mic.
We don't notice that change in the FR because our brain 'calibrates' the sounds by knowing/remembering how a real instrument/voice sounds.
That's also why we don't really notice our hearing getting worse over time untill we are confronted with ear measurements.
Our brains calibrate continuously.
When our hearing suddenly alters we DO hear it as calibration is a slow process.
Of course a tube that is open and 'listens' to the ouside world with a mic inside needs a different compensation than what would be needed if a soundsource is located halfway or partway down that tube and is blocked off with foam or stuff simply because the tube is effectively shorter AND sealed on both sides.
So when one uses the compensation that is required to compensate FR for outside noises (room or headphone sounds) you will need a different compensation than for IEM's IF you want the plots to show a correlation to how we would percieve it.
This means when a plot shows a 'flat' FR the SPL is equal for all frequenties which does NOT mean we experience all those frequencies equally loud.
How loud we perceive SPL with different frequencies depends on the actual SPL and the condition of our hearing.
That's why I mentioned when using OW, that is intended for ISO 11904-2 (on-/over-ear headphones) is used for something other than those type of measurements different compensation is needed than when the same artificial head is used for measuring sounds coming from the front, side or rear and needs to differ yet again.
Fact remains that the FR alterations that are created in a test rig need to be compensated IF you want to know what the transducer puts out.
Then, if someone wants they can also apply a reverse room correction to get a feel of how the same music would sound in a special conditioned listening room.
The raw plots of IEM's are easy to compare as they are all quite similar in 'microphone signal'.
Once 'corrected' all bets are off unless the exact same compensation is used. (that's what standards are for)
Correlating uncorrected IEM measurements with the actual sound is impossible to do and don't look anything like (corrected) FR plots of 'headphones'
You can also measure headphones on 'flatbed' rigs, in fact most measurements on this website (as well as mine and some on personal audio.ru) are made this way and NOT at the 'eardrum' using an (expensive) manekin.
Sites like Rin's, Golden Ears, Personal audio.ru, innerfidelity, headroom and a few other sites DO measure using a manekin and some will show corrected and/or 'raw' plots.
To those not realising what plots they are looking at (compensated manekin, raw manekin, rawflatbed or compensated flatbed) will find obvious correlation between raw manekin and wrongly compensated manekin vs correctly compensated manekin (OW). OW on a manekin and flatbed have a much closer resemblance (for on-/over-ears).
Of course RAW manekin plots and raw IEM plots are comparable in a superficial way as they will both show the typical 'tube' effect but the over-ears/on ears will also have other effects caused by the skin/pinna/seal etc and the peak caused by the 'ear canal' will differ somewhat from that of the IEM measurements so even though the raw plots will look very similar they actually are NOT similar.
In other words IF 2 plots of on-/over-ear and IEM would measure exactly the same they would sound the same.. BUT an over ear and IEM that will put out the very same FR spectrum these will measure and sound different.
So in the end the question remains WHAT we want to see in the plot.
HOW 'average people' (whoever they are) perceive the headphone compared to speakers or not
OR if we want to see what a mic records but in that case we need to know the conditions of the measurement.
OR we want to see what the drivers actual SPL output is.
The latter (IMO) will be closest to how trained listeners perceive the sound when looking at FR plots.
-
Thanks solderdude, that is a clear breakdown.
So would it be safe to say that Rin and Tyll (or whoever) who measure using a an expensive and complicated manekin and then compensate for the coupler, artificial pinna, head, shoulder, etc using a standard do so with one (of possibly several) motivation being to give their measurements a sense of authority and adhering to standards that others are familiar with? (i know, lots of assumptions here)
I frame it that way because I see threads where some do not acknowledge the validity of Changstars/Marv's measuring methods or lack of transparency of measuring methodology. Please correct me on the following: as I understand it the thinking of the people who measure on flatbed do not use the traditional HTRF compensations because they believe a lot of the anatomical acoustic compensation is already handled by the brain, and since they are after a qualitative relationship between FR and perceived sound it is more direct to just find the SPL near the drivers without the influence of pinna, canal, shoulder, etc. And as long as you are comparing full sized to full sized, or IEM to IEM, the relational differences between FR's will give you a clear qualitative assessment of how the transducer will sound at the eardrum.
-
So would it be safe to say that Rin and Tyll (or whoever) who measure using a an expensive and complicated manekin and then compensate for the coupler, artificial pinna, head, shoulder, etc using a standard do so with one (of possibly several) motivation being to give their measurements a sense of authority and adhering to standards that others are familiar with? (i know, lots of assumptions here)
In Tyll's case the compensation he used differs from most others.
That is... IF he was to measure a headphone that has a completely flat FR (SPL wise) than it would NOT show as 'flat' in a plot. It would be sloping downwards from about 2kHz considerably. I thought (but Tyll or someone else correct me) that the decision to use a certain correction type was made together with the manekin supplier and isn't according to an actual standard but deviates.
Interestingly enough he 'converted' some of his raw plots (that are accurate) to OW in one of his articles. Fortunately some well known headphones and when you compare these OW compensated plots you'll find a better correllation with how they are perceived by many (certainly with me)
have a look: http://www.innerfidelity.com/content/first-test-estimated-harman-target-response-curve-various-headphones
having said this I often consult Tyll's more than excellent pdf's and once you know how to 'interpret' the FR plots they are more than useful.
Also the other plots and the neat (and constant) format sets a real example. Headphones can be compared quite well.
His measurement methods are well described and transparant.
Rin uses other, and later on OW, compensation that relate better to SPL IMO.
I frame it that way because I see threads where some do not acknowledge the validity of Changstars/Marv's measuring methods or lack of transparency of measuring methodology.
The validity of Changstar measurements is something that should be acknowledged even though home made gear is used that may not be openly discussed in one single thread but info is spread over various threads.
I think an effort is made to correlate measurements to how they sound and succeed as well, sometimes better and sometimes perhaps worse than Golden ears.
Of course most (if not all) of the guys posting their plots do so knowing their plots are indicative and not to be taken as 'correct' in an absolute sense.
As Marv often mentions... it's a hobby and you have to see the plots as indicative.
That's the absolute truth...
Please correct me on the following: as I understand it the thinking of the people who measure on flatbed do not use the traditional HTRF compensations because they believe a lot of the anatomical acoustic compensation is already handled by the brain,
I can't speak for others but even flatbed measurements need some correction but NOT the typical HTRF. I aim to measure the SPL coming of the drivers and suspect others try to as well. IF a speaker is to sound speaker like it should be tuned that way by the headphone manufacturer (or modder). This will (should) show as downward sloping on plots.
I don't know how others compensate/calibrate their rigs and can only say something about mine. The mic capsules I use (WM61A, now obsolete) are known to have a peak around 16kHz and they measure pretty flat even down low when in free air. A rig is NOT free air and the FR response alters when that same capsule is baffle mounted or infinite baffle mounted.
Knowing what the actual effects are I merely compensated for those 2 anomalies + the DAC/ADC/amplifier chain. The result should be a reasonable approximation of the by the HP generated SPL. I think it correlates pretty well to what I hear (may differ from what others hear).
Also when headphones are EQ'ed completely 'flat' on that rig they sound unbelievably real, not to me alone but those who heard it as well.
This too gives me some confidence it is reasonably accurate.
since they are after a qualitative relationship between FR and perceived sound it is more direct to just find the SPL near the drivers without the influence of pinna, canal, shoulder, etc.
The jury is out on that. I have experimented with (crude fake) Pinna on a flatbed and various other obstructions/fillings/angles and do notice differences in plots taken with or without Pinna which also varies with the HP in question. The differences range between almost no influence to substantial as in several dB.
In general actual SPL is closest to reality.
And as long as you are comparing full sized to full sized, or IEM to IEM, the relational differences between FR's will give you a clear qualitative assessment of how the transducer will sound at the eardrum.
At the eardrum the FR of the SPL differs tremendously from the SPL that is present just outside of the ear canal.
We do not perceive the sound via our eardrum as if the eardrum is a flat microphone.
You could say that our brain compensates for the changes that are made by the ear canal.
-
Thank you! Very informative.
At the eardrum the FR of the SPL differs tremendously from the SPL that is present just outside of the ear canal.
We do not perceive the sound via our eardrum as if the eardrum is a flat microphone.
You could say that our brain compensates for the changes that are made by the ear canal.
I realize the difference between SPL at eardrum and just outside the canal, I meant that when looking at graphs you basically have to group the headphones separately from the IEMS.
-
have a look: http://www.innerfidelity.com/content/first-test-estimated-harman-target-response-curve-various-headphones
This is a quote from Astral Storm in the comments:
"If we were to hold to high fidelity target, the real benchmark would be an accurately recorded (as in measurement microphones) live performance. This means as close to vanilla performance as possible, making any mastering issues moot."
This makes sense, so why isn't the definition of neutral or a flat response directly correlated to this sort of replication of sound?
-
I realize the difference between SPL at eardrum and just outside the canal, I meant that when looking at graphs you basically have to group the headphones separately from the IEMS.
Yes, even the raw measurements of IEM's and full-sized HP's on a Manekin will not be exactly the same.
When the plots of IEM and full sized were identical they will still differ somewhat in FR and sound.
When listening to headphones we don't need any compensation at all for close to 'flat' SPL headphones, unless the FR is far from 'even'.
Some like to add crossfeed though.
The fact that compensation is needed to create plots that have a proper relation to perceived sound already says enough about how much test rigs alter the actual sound and thus should be correctly compensated for their specific errors. Not errors people erroniously think should also be compensated as well.
-
This is a quote from Astral Storm in the comments:
"If we were to hold to high fidelity target, the real benchmark would be an accurately recorded (as in measurement microphones) live performance. This means as close to vanilla performance as possible, making any mastering issues moot."
This makes sense, so why isn't the definition of neutral or a flat response directly correlated to this sort of replication of sound?
To me it is directly related. When a headphone measures a signal 'flat' and is reproduced 'flat' they should sound tonally the same but will still not sound the same in spatial and timing aspects even when the chain is flawless.
To accurately record music it would have to be holographic over a huge area, in other words the direction of wavefronts should be recorded as well.
This can't be done with current techniques.
The same goes for reproduction, that should be holographic as well and have the exact same SPL as the recording had and be flawless to come really close to the original.
Stereo reproduction, even when flawless, is but a very meagre representation of the soundpressure that once was during the recording.
Also NOT every one appreciates 'flat'.
Colouration of many sorts is often preferred over accurate.
Its the reason so many different headphones/speakers exist... to suit ones taste and wallet.
For me 'flat' and accurate is the way to go but don't mind some flavouring now and then depending on the recording.
IMO the recording quality is the real bottleneck closely followed by the transducers.
-
Perhaps a mod could move this of topic discussion (that interest me) to another thread/place to keep this thread for Tio measurements only ?
Its digressing a bit ... ::)