|WikiProject Professional sound production||(Rated Start-class, Mid-importance)|
|WikiProject Physics / Acoustics||(Rated Start-class, Low-importance)|
The curves you can see here are very old curves of 1933 from Fletcher-Munson.
The refined Robinson-Dadson loudness curves are now new standardised by ISO.
A good sentence, but where can you find them?
See http://www.nedo.go.jp/itd/grant-e/report/00pdf/is-01e.pdf for full details with interesting comparisons of new ISO226 with both F-M and R-D curves. - Lindosland
I am not happy with the statement, "headphones change the acoustics of the ear, however, (particularly attenuating low frequencies". My understanding is that the ear is purely pressure sensitive at low frequencies, and since the cavity formed between headphones and ear is too small to resonate in this region, headphones in no way change the acoustics. Provided that the headphones are flat, and a good seal exists to the ear, errors in this region are very small. I have confirmed this by measurements using Sennheiser HD650 headphones, which are flat to 25Hz. Many headphones are not flat, or do not seal, but this is not changing the acoustics of the ear, and usually results in loss of low frequencies. The Fletcher Munsen curves are actually in quite good agreement with the revised ISO 226 at low frequencies, and at 40 phon level (A-weighting level)and it is the Robinson Dadson curve that seems to show increased LF sensitivity, suggesting perhaps that something was badly wrong with the speaker/room response in these experiments.
- I'm not sure I understand how the HD650 is 'flat' to 25Hz when measurement from HeadRoom shows a clear bump of ~5dB from 25Hz up to around 400Hz. Can somebody clearify that for me? Appreciate it! =) -- Dept of Alchemy 04:10, 11 May 2006 (UTC)
Measurements from Lindos
You can find my own measurments at http://www.lindos.co.uk/cgi-bin/download.cgi/Results/184/filename/res.html along with many others on the Lindos test sheet database. I believe this result to be more accurate than the one you quote, as it was made with the microphone in my ear, and the microphone used is known to have a very flat bass response. A bump of only 3dB counts as 'very flat' in the world of speakers and headphones. Measurements made off the head can show different results due to different cavity resonances and damping. Unfortunately I have been banned from putting references to the Lindos Test Sheet Database on Wikipedia for the simple reason that I contributed to it. Links to results for the i-Pod and many other items, which I consider very valuable were taken down. You can legitimately put them up though if you wish! --Lindosland 09:33, 18 June 2006 (UTC)
The key difference with headphones is side presentation, which normally results in greater HF sensitivity than frontal presentation. This is apparent in the comparisons, though not as marked as I might have expected. We are left with a huge (30dB@20Hz) difference between the new ISO results and the Fletcher-Munson below 100Hz (100 phon curve). I suggest that this is due to distortion in the early headphones causing the listener to mistake harmonics for the pure tone. - Lindosland
- I noticed that the Lindos measurements you cited was measured without ear canals, whereas the measurements from HeadRoom Corp. are done with state-of-the-art Head Acoustics Artificial Head Measurement System (mic behind ear canal) in an anechoic chamber. I found the Lindos site to be an amazing site to find measurements for everything from mic to monitors, though. =) Dept of Alchemy 23:44, 19 June 2006 (UTC)
Good, it is getting known better now and I've just added the Sony FX1 HD camcorder! Please do us a favour and put links to that site on any Wikipedia page you feel to be appropriate. I put up forty links thinking I was doing everyone a favour, and then came in for a long battle from a few folk who insist on the rule that you cannot add a link to your own site (despite my protest that my son owns Lindos and runs the site and I only contribute to it). I was forced to give in and all the links were taken down. Some argued that it was primarily a selling site, but that is hardly true so long as you link specifically to the database or an article, as these are in no way pushing products.
Yes, I did measurements first with the ear canal open and the small mic at the entrance but not blocking it. The results did not differ much when I blocked the ear canal mounting the mic on an earplug, and what differences there were appeared only in the HF region. This surprised me a bit at first, but I now conclude that the HF dip is mostly to do with the deliberate correction (for side vs frontal presentation) in the headphones and the resonant space they form over the ear. The effect of the ear canal would no doubt be much greater if I measured at the eardrum, but of course such a measurement has no relevance because any frequency response difference between ear opening and eardrum applies to live sound as well as reproduced sound and is compensated for (expected) by the brain. Routine measurements with an ear plug are less painful! Running a sweep at 90 or 100dB SPL, though not particularly dangerous over 20 seconds, is not something I want to repeat too often with the ear canal open. I often measure headphones at shows, where running a high level avoids errors from outside noise!
My 'state of the art head' is of course superior to any artificial one, and as I said, measuring 'behind the ear canal' is simply wrong when measuring headphones. Strictly speaking we aim to produce the same response at the entrance to the canal as would be produced by a frontal (or strictly perhaps 30 deg off centre stereo, since recordings are made for such presentation) source in free space. The latter has peaks and dips in the HF region. Measurements made at the eardrum must be compared to similar measurements at the eardrum for a frontal source, which would be significantly different from those at the entrance. This is a common mistake made by many writers on the topic. In any case, you can be sure than LF response is not affected at all by such things - the cavity driven is too small to exhibit any resonance below 1kHz. I have tried measuring the response with the earphone pressed against a hard bench top, as well as a rubber mouse-matt, and there are significant, though not large differences, in the bass region. I put this down to the fact that any leakage around hair etc adds damping to the driven cavity, and so I consider results actually on my real head to be the best. When it comes to earbuds, inserted into the ear canal, then things are different, as the ear canal is being pressure driven directly, so that proper comparisons at the eardrum are desirable, but it would be quite wrong to suppose that a flat response is desirable at the eardrum of course, it should be the same as from a free-space frontal source, ie with drastic HF peaks and dips. I suspect that the main difference in the other measurements you cited was caused by an LF rolloff in the mic used. Our MM3 mic is well proven to be flat to 20Hz within about 1dB, as you can see from other pages on the Lindos test sheet database. --Lindosland 13:51, 24 June 2006 (UTC)
Revised Graphs to ISO 226:2003
I have put up a revised set of contours because I feel that the graphic at the start should reflect the tru current position. This is a Lindos1.svg and being scalable-vector can be scaled to any size without loss. I have also ammended the text to explain the big errors that were discovered.
I have also created a new page 'Fletcher-munson curve' and transferred some of this page to there, with the old graphs, for the time being so that it is not lost. I will prepare proper F-M graphs and maybe also the R-D graphs for a Robinson-Dadson page too. Lindosland 22:45, 1 December 2005 (UTC)
I assume that the horizontal axis of the graph represents frequency, but it is not labelled. It would be helpful to label it... or add the units "Hz" to the figures.
I agree. I'll try to add that some time, though a lot of work went into creating the graphs (and others on other pages) and I might now want to do them all over again using Inkscape which I only discovered later. Inkscape is brilliant in allowing easy use of layers and enhanced svg output which is fully compatible with other drawing packages but retains full editability in Inkscape, ie it works in native svg unlike other packages that require saving in a different form for editing. --Lindosland (talk) 17:28, 21 November 2007 (UTC)
Research on psychological/physiological origins?
Does anyone know of any research on the psycho-physiological causes of the equal-loudness contour? (ie. perhaps dampening of highly repeating signals on the auditory nerve, or an increase in those signals, networks in the brain that specifically habituate to "noise" and filter it out before it reaches consciousness, etc) One would think there would be by now, given the prevalence of sound-sensitivity in autism spectrum disorders (..OR..). Jimw338 (talk) 15:07, 1 March 2012 (UTC)
- I feel that there is no mystery to the causes of the equal-loudness contours, but they are many and largely mechanical, not necessarily neurological. First the pinna has a big effect by adding reflections that cause addition at some frequencies and partial cancellation at others, depending on direction of sound source. Then the ear canal clearly acts as a resonant tube, its exact effect hard to predict as it is a function of the impedance mismatch from the concha (inner shell of the pinna) to the tube and also damping dependant on turbulance and the softness and lossiness of the tissues. Then comes the eardrum with further possibilities for resonance, controlled according to loudness by the tensor timpani muscle, and the small bones, which couple differently at different levels of loudness under control of the stapedius muscles. Then the cochlea, with fluid transmission and hair cells, plus active feedback as recently established via more hair cells (see otoacoustic emission). So a mechanical model is extemely complicated, but would be expected to give many high frequency peaks and dips. The LF falloff at low levels does not have an obvious cause and may be more neurological.
Be careful though about assuming hyperacusis in Autism (which greatly interests me). My reading about Williams syndrome, which is compared to Autism but has a distinct genetic cause, has taught me that what is commonly called hypersensitivity in many papers is actually a heightened intolerance of sounds, not an increased sensitivity as would be revealed by tests on tones and equal loudness plots. This fits with ideas of information overload generally and does not necessarily imply any change to actual hearing mechanisms. Lindosland (talk) 14:25, 23 August 2012 (UTC)