Topic: EQ curve to compensate for hearing loss- thoughts...

[andromedanwarmachine]

Hello team,

I am just about to complete a little project I've been turning over for a while now and I thought I'd post up the concept to see what people think.

Every few years I spend a bit of time measuring my hearing response and try to go to reasonable lengths to get it as accurate as I can. This year, in the last week, I have spent about 7 hours measuring both sides using two different methods.

I have been well aware since '97 when I was installing Meyer UPA 1-P cabinets into a major venue here in Scotland that I have a significant response difference between my ears; (it was not until I was listening to these cabinets that I realised!) I am 38 and suffered a perforated eardrum on the side with less response when I was 5.

Now that I have a current "audiogram" of my response it is my intention, in combination with Fletcher-Munson curves, to produce a counter-response curve which I can apply to a variety of monitoring situations to even out some of the inefficiencies of my hearing.
I am aware that this sounds a little nuts as the brain compensates for losses to the extent that it shouldn't matter, but I have become more and more aware that my hearing is not as good as other peoples', even older than me by some distance and am sure this has an effect on my powers of post production.

The curve would hopefully be subtractive and very mild- nothing beyond 6dB (hardly mild, admittedly!) and be used only in critical monitoring situations-  through headphones. (Sennheiser HD-25s)

Does anybody have any constructive criticism of this concept?

I can measure it- but what would people say was a typical monitoring level? (with quiet background)  I know this is a "how long is a piece of string" question!

Thanks

JimP

P.S. does anybody know of a FLAC player that would accomodate a complex EQ setting/program...?

[andromedanwarmachine]

Gosh! No replies!!

this must be quite an unusual/foolhardy concept??

measured a nominal monitoring level last night through headphones and it tops out about 90 dB for me... With program (music) and through my cans that is.

I've got to work with some kind of reference level and I think this will be it...

JimP

[it-goes-to-eleven]

There is an open source project(s) which does real-time EQ to compensate for listening space issues.  You could use something like that to optimize your playback to match both your hearing curves, and your listening space.

I can't find the original project, but if you search for digital room correction, etc, you should find a fair bit of discussion on the topic.

[page]

I started having my hearing tested professionally about 2 or 3 years ago and I learned two things; 1) That my hearing was exceptional for resolution of quiet sounds, and 2) that I have one side that is short about 4-5db at 1k, but the rest of their reference points were effectively the same. I generally don't notice it and try not to let it bother me when I'm critically listening to something.

As far as compensating for it. I think it's worth a shot to try, but I'd take the results you have (vallys in one spot, peaks in another) and reduce the severity of the differences from flat by about half to start with and try that (to see what compensation the brain will do). best of luck.

There is an open source project(s) which does real-time EQ to compensate for listening space issues.  You could use something like that to optimize your playback to match both your hearing curves, and your listening space.

I can't find the original project, but if you search for digital room correction, etc, you should find a fair bit of discussion on the topic.

Thats where I was headed with it.

[Gutbucket]

http://refinedaudiometrics.com/services.shtml

Not an inexpensive solution, but definately worth reading the information on the site to understand the approach. 

Basically, EQ for headphone response / listening preference not for hearing loss, unless using some sophisticated processing to compensate for loudness non-linearities.  Those non-linearities are compounded and multi-dimentional since music is dynamic not only in an overall sense like an averaging VU meter, but also consists of invividual dynamic elements that change volume relative to each other within the larger macro dynamic structure.

It may be possible to poor-man replicate that using chained VSTs in a routing environment like AudioMulch, but setting it up would be tricky.  Off the top of my head I think it would involve running bottom-up, parallel compression using some sort of advanced, many, many multi-band compresser with highly tuned frequency variant threshhold and compression curves that are an inverse of your measured ear response.. then duplicate that chain with the appropriate settings for the other ear along with individual ear EQs.  Something like the seeming unique features of the Voxengo Sonifier might form a basis for doing that. 


[edit to add that Refined Audiometric's free headphone cross-feed plugin VST is the most natural I've found and the only one I ever use on the rare occasion that I do.  Worth the download if you want cross-feed]

[live2496]

I think you have a good idea and isn't that what audiologists are doing when they adjust a person's hearing aid?

Here is a VST plugin eq that someone was using on the har-bal forum. It is a 31-band with a variable Q per filter, and they have a L/R version.
http://www.aay-audio.com/index_products_aQ-31.html

After determining the appropriate curve, you could use that to adjust playback eq for any music that you are working on. But you would have to use DAW software that supports VST plugins. Just put that on the master buss with your personalized settings. I am assuming here that the eq plugin will allow you to save/load settings.

BTW, there is a VST bridge for foobar2000 which would allow you to apply the eq settings to anything you playback in foobar2000 too.

[illconditioned]

OK, I've got some hearing loss myself.

You should *not* be correcting for this unless you know what you're doing.

What do hearing aids do: they act as a expander-compressor in selected bands.  That is, in a band that has hearing loss, the sound is amplified at low volumes.  At higher volumes it is compressed.  Essentially you have to "re-map" the dynamic range of sounds (in level and frequency) from the Flecher-Munson (sp?) curves to those that your (damaged) ear hears.

Old hearing aids merely increased volume.  Next generation was frequency-dependent amplification.  Current generation is multiband expander-compressor.  Next will be adaptive "neural net" tuning to particular listener.  The is just coming on the market, but probably still considered experimental.

  Richard

[page]

Old hearing aids merely increased volume.  Next generation was frequency-dependent amplification.  Current generation is multiband expander-compressor.  Next will be adaptive "neural net" tuning to particular listener.  The is just coming on the market, but probably still considered experimental.

Neato

[andromedanwarmachine]

that's all very interesting- thanks for your feedback...

I'm not trying to gloss over the highly complex issue that this is, nor am I of the belief that this will totally "cure" my response deficits. I'm just trying to massage the glaring differences in top end response between my L & R sides.

I can recall many instances in the past were (older) people have said to me- "oh listen to that bird!" or "wow!-listen to the noise off that battery charger!" and I have been totally oblivious to what they have pointed out. I'm not under any illusions- you can't make the deaf hear but further down the range I think I would certainly benefit in a critical monitoring situation from a sympathetic "shadow" EQ which compliments the differences between sides...

I think that's the key point actually- that I'm trying to even out what I'm hearing between L & R rather than attempt to "hear more"...

JimP

[Gutbucket]

http://refinedaudiometrics.com/services.shtml
Something like the seeming unique features of the Voxengo Sonifier might form a basis for doing that. 

Here is a VST plugin eq that someone was using on the har-bal forum. It is a 31-band with a variable Q per filter, and they have a L/R version.
http://www.aay-audio.com/index_products_aQ-31.html

Probably more accurately described as unique in comparison to typical multi-band compressors, or at least the ones that I've run across..  Har-bal came to mind, but I'm not as familiar with it.  I think the narrowness of the bands needs to correspond to the descrete auditory 'bark' freqency widths of human hearing.

Interesting stuff.

[live2496]

P.S. does anybody know of a FLAC player that would accomodate a complex EQ setting/program...?

JimP,

I guess I should be clearer about this. Foobar2000 will play FLAC files. There is an option available for it that will allow you to use VST plugins.

In looking around I see a new one is available... http://www.hydrogenaudio.org/forums/index.php?showtopic=84947

You can use this to host the aforementioned eq plugin.

Gutbucket,
RE: Har-bal. It is an offline processing tool. There have been requests to have it available as a plugin and I think the developer (Paavo) has some ideas about developing something with respect to room correction. But I don't know if he has any of those ideas in the current beta release that he is working on.

[Church-Audio]

I have my hearing checked yearly I am very lucky my hearing is still very good. I do agree with Richard that doing this can cause more harm then good if its not done correctly. Most hearing tests dont employ a detailed enough spectrum for meaningful interpretation into correction Data. My Audiologist does wide band detailed testing on my ears because I am a sound engineer. I suggest if you are going to do this that you get several tests done on your ears and "average" the response out. Because these tests can be misleading and sometimes mistakes are made. Resulting in problems with the attempt to correct your hearing. The other major issue will be one of equalization points and amplitude required for the correction.. Resulting in music playback that is present for you and you alone.

Typical listing volume for me is around 70db To a max of 102 db for any show I mix.

[andromedanwarmachine]

that's interesting,

I did a full 1/3 octave 31 band threshold test on either side- twice, using different gear.

On average, it takes me about an hour and a half to measure one ear. I was my intention to average them to produce a starting point, but I think I will do another for a better average.

yes- it'll produce a totally individual result, usable with only the cans used for calibration and not through speakers or for anyone else...

thanks for your feedback

and thanks for that live2496- that's tremendous
JimP

[andromedanwarmachine]

Hmm...

well the results are far from in, but I thought I'd share an interesting observation which has come out of the "research" so far...

Having established a "technical" complimentary response curve, which is to say the literal result derived from analysis, I started auditioning well known pieces of music to fine tune the filter by ear, so to speak. This is acting upon only my left side.

The results were quite startling. Initial impressions were of significant over-brightness and stereo image shift. Auditioning a range of material revealed the bottom-end roll-off which I thought was beyond my range of detection (that I had added for completeness) also led to significant shifting of the perceived image placement and so I removed it. Broadly speaking; stereo music is rarely symettrical in pan and image, which I suppose is blindingly obvious, but I hadn't considered how it would affect the tuning of my curve. Depending on the band to be adjusted, significant portions of the image of a peice of footage will move leading to the question;

"what is correct?"

Visual analysis of waveforms shows immediately what's going on. To continue, I need to work with mono reference tones auditioning through both ears and use image placement to reveal how much amplitude adjustment is required.

JimP