Do people with hearing loss (age or noise) hear artefacts in lossy codecs more?

[audio2920]

This might seem counter-intuitive, but my casual observation is that when I work (on film sound) with people who have some degree of hearing loss, they are generally more likely show dissatisfaction when hearing AAC, MP3, EAC3 etc than I or my peers are. Not sure about other codecs.

This sort of makes sense to me, as my understanding is that the content a lossy codec keeps is "chosen" based on models of "normal" hearing. So someone who fits the particular model perfectly will be less able to hear the artefacts than someone who's an outlier in terms of spectral and temporal masking; regardless of how good their hearing is in the real world. (right...?? or wrong?)

My current guess is that maybe because of the spectral masking assumptions used in these codecs, those with hearing loss do not conform to the model. By that I mean for example in very simplistic terms, if it's assumed that a strong signal at 13.0kHz will mask a weaker one at 12.9kHz but the listener can't hear 13.0kHz in the first place because they have a hole in their FR at that spot frequency, it will reveal the codec's hand at 12.9k for that listener. Whereas, for someone who hears the 13.0kHz, the presence (in PCM or lossless) or lack (in lossy) of the lower level content at 12.9kHz is indiscernible. Maybe?

I dunno. Maybe my observations are purely co-incidental! Does anyone have any thoughts, experience or research on hearing damage / aging vs lossy compression? Only out of curiosity...!

 

[staticV3]

How about this: (primitive) lossy codecs tend to throw away high frequency content first. Couple that early HF roll-off with the HF roll-off that people with hearing loss naturally have, and you get an experience twice as muffled->unsatisfying as for those with intact hearing.

 

[Blumlein 88]

How about this: (primitive) lossy codecs tend to throw away high frequency content first. Couple that early HF roll-off with the HF roll-off that people with hearing loss naturally have, and you get an experience twice as muffled->unsatisfying as for those with intact hearing.

I don't think it is that. I do think the OP has observed something others have seen.

If you damage and/or age your hearing the filtering of critical bands (our hear using some 30 or 32 such bands on sliding center frequencies) becomes less sharp. The level of masking changes.

One example if the cocktail effect. You are at a cocktail party with everyone talking creating a moderate level noise. You can mostly tune it out, but if someone mentions your name you pick it out and notice it. With damaged or aged hearing you don't tune out the noise very well, and you hear more of a low level constant din which you have trouble hearing thru even talking to those near you, and your name may not be well heard unless it louder than is the case with normal hearing.

Another example is going to a sports event in a large stadium. When young crowd noise is a constant, but it isn't something too bad subjectively for you and you can hear some detail into it from those near you. Aged people may stop going to such events as they noise seems louder, and they can't make out any detail, and it seems to just tire them out or almost put them in a daze so unpleasant it is.

So marginal cases with MP3 young ears can filter it better and separate out the main part of the music they want to hear. Older people do this less well, and something in a different critical band that wouldn't mask the music for a young person bleeds across to adjacent critical bands and they hear some artifacts because they change the music they are hearing.

 

[2M2B]

My hearing peaks at 15.5KHz on my right ear but my left is 14.1KHz, Yet Musepack at 175kbps is transparent to me. Yet with any MDCT lossy codec I can notice mild issues to ugly as hell artifacts that need 350kbps. I'm a bit lost on why a subband codec based on MP2 is doing what Vorbis/AAC claim at 144 ~ 160kbps?.

The niche nature doesn't hold anymore when any DAP/Phone with a 3rd party app can play .mpc files.


Vorbis at 192kbps = Anything with distortion or crunchy noise I require 320kbps VBR, Cause at 160kbps I get this weird loud puffy noise that sounds like wind.

AAC(any encoder) = I get MP3 like issues like ringing on synth textures and sounds, even percussion can trigger It, It require 256kbps VBR. But there some samples that fail at 320kbps but Vorbis/MP3 V0(48KHz) are transparent.

Opus = only at 80 ~ 118kbps but at 130 ~ 150kbps It 99.9% transparent beyond a very distorted Noise album 2nd track. When Opus does fail It sounds like AAC/Vorbis artifact wise.

LAME/Helix MP3 = Quite unfair since it a 32x subband codec but the final output is fed through a poorly thought out MDCT layer. Oddly doing 256kbps VBR with 48KHz resample helps make MP3 transparent on It own killer samples, So this is what I use If I can't use MPC.

Musepack/MP2 = Tried every killer sample I could find, My own ones I found in my lossless album archive, Artificial samples. I fail to hear --Standard from flac even when the bitrate is 80 ~ 144kbps, While others instead just shoot up to 384 ~ 790kbps for complex sections to the whole song.

 

[abdo123]

I don’t get all the philosophical interpretations.

My right ear has like 50% of the hearing of my left ear and it’s basically like someone dimmed the light of one of my eyes SIGNIFICANTLY to the point where almost only half of each scene i have my eye pointing at is visible. I hope this analogy helps you out.

It’s just not there, it’s not different, it’s not special, it’s just missing.

There is no special stuff, no glam and no shablang. It’s just death, in a way.

 

[Deleted member 23982]

well, you guys speak way beyond the real factor, training, once you clearly heared the difference on mp3 vs flac on a clear high end system, specially with low jitter you will also hear it on lower end systems on some degree, its harder to hear but if you know WHAT you are listening for it becomes way easier

so the real question is, what have your co-workers heared before i think is the right question/answer

i can just say one thing, i never tried and gave up on blind listening once i heared WHAT makes all a difference in enjoying music, different op amps make a huge difference, capacitors of different names make a huge difference, JITTER makes a huge difference and after that and much more (for example audiophile fuses, cables, power cables, connectors) and after all that you also hear a clear difference between mp3 and flac

the best way to test in my opinion is trust your brain unless you go crazy, hear a song across days/weeks on a non-changing system and once you can really "memorize" the sound with critical listening for special things like transients, try changing one thing, you can clearly hear differences then, atleast me, tho it gets much harder to memorize one song if you listening it across multiple setups

my hearing also peaks around 14,5k-15,5k if that helps in any way...

 

[abdo123]

i never tried and gave up on blind listening once i heared WHAT makes all a difference in enjoying music

Just a heads-up, you won't get very far without conflict with this sort of opinion in this forum.

 

[Deleted member 23982]

Just a heads-up, you won't get very far without conflict with this sort of opinion in this forum.

unfortunaly i know, thats why i described my testing method with getting used to the setup for days/weeks before changing something,
the problem with blind testing is in my opinion, deep in your mind you got used to your setup and changing fast between a "second and third setup" blind testing just doesnt make the cut, between your real setup and a change should be no problem tho (if you have a person that actually takes their time with testing with you )
maybe i have a strange mind in this regard but the method i described works for me and always made sense in the end (i always came to the same conlusions, no matter how many times i tried), even if people say stuff like "its not in the audible range" and so on...

 

[Blumlein 88]

unfortunaly i know, thats why i described my testing method with getting used to the setup for days/weeks before changing something,
the problem with blind testing is in my opinion, deep in your mind you got used to your setup and changing fast between a "second and third setup" blind testing just doesnt make the cut, between your real setup and a change should be no problem tho (if you have a person that actually takes their time with testing with you )
maybe i have a strange mind in this regard but the method i described works for me and always made sense in the end (i always came to the same conlusions, no matter how many times i tried), even if people say stuff like "its not in the audible range" and so on...

The only problem is this has been tested in different ways for different issues over the years. EVERY....SINGLE.....TIME the quick switching type of blind testing has proven more discriminating in sound differences vs the long term listening method. Your proposal that long term listening provides additional info or more reliable info or more discerning info is one that feels oh so right, and unfortunately does not seem to be the case.

 

[Deleted member 23982]

The only problem is this has been tested in different ways for different issues over the years. EVERY....SINGLE.....TIME the quick switching type of blind testing has proven more discriminating in sound differences vs the long term listening method. Your proposal that long term listening provides additional info or more reliable info or more discerning info is one that feels oh so right, and unfortunately does not seem to be the case.

well these blind test also always come to the conclusion stuff like connectors/cables/jitter etc doesnt matter, so im not sure either what to believe (maybe just a bad setup alltogether so issues gets blended out and so on...), but just "mind-fooling" after all the testing i have done is not a option - sorry
in the end everyone should do what makes him happy, for me its hunting audio-gremlins in my own way i guess

maybe a more "real world" example would be to listen for weeks to one setups and then changing speakers/headphones, you will clearly hear a difference.... its the same with my testing method just way less obvious but a trained ear/brain is just so accostumed to the sound it heared for weeks that differences gets clear, maybe you cant but your thumb on it each time, but you hear "something" is wrong, maybe im not normal and have a different kind of audio perception - i dont know, but it works (for me)

if you give me a new mp3 320kbit file i never heared before, pretty sure i couldnt tell you if it was mp3 or flac, if i know the recording (and hear it on the setup im accostomed to) its way different story, thats why i think blind-testing has its own flaws in fooling yourself BECAUSE you dont want to fool yourself....

its the same with people saying absolute phase doesnt make a difference, i can just lol on that - sorry x) (tho it gets way more obvious if you exactly know what you are listening for, AND have the right tracks... (and the right setup...))

 

[BDWoody]

well these blind test also always come to the conclusion stuff like connectors/cables/jitter etc doesnt matter, so im not sure either what to believe

Have you done one yourself?

If you haven't, give it a try.

If you have but couldn't determine a difference, why would you not believe it?

It's too easy to kid ourselves. Some prefer to be fooled, and don't *really* want to find out they couldn't hear a difference after all.

 

[audio2920]

Thanks for all the thoughts thus far folks!

I expected (hoped for) differing opinions and I'm quite happy for the answer to my original question to be "No, it makes no difference" as my observation that people with hearing damage notice "sooner" is a very casual one!

But I'd be really interested if someone with better knowledge than me of how these codecs weight/prioritise/discard information could figure out if that might mean that someone with janky FR (holes as well as roll off) might be more susceptible to hearing artefacts in any given scenario. Be that the listening tests described or just a casual listen.

Obviously I am interested (confirmation bias!) in @Blumlein 88 's crowded room analogy and possible explanation.

I take @abdo123 's point too, that it's just lost. Although I'd dispute what we're talking about here is philosophical.

As someone who's enitre professional career for several decades has been as a studio based sound engineer, I'd like to think I'm as aware of how these artefacts present themselves (to me) as one realistically can be without it being their specialist skill area. How to hear them / whether I can hear specifics, although pertinent to some extent, isn't really what this is about for me (although I understand the logic put forward by @ghoostknight that "training" may play a part.)

In the specific cases I've observed, there was no direct fast A/Bing, just (later found to be correct) comment from non-sound department older/deafer people saying "that doesn't sound right" even without necessarily having a PCM/lossless reference point.

 

[audio2920]

I guess a rather crude experiment I could do (next time I'm at a computer) would be to take a lossless and lossy version of the same thing and then mash a bunch of random high Q notches, and maybe some more gentle dips, and a bit of low pass on both of them. Then I could see if I think the perceived quality is more degraded by this processing on the lossy version...

 

[LTig]

This might seem counter-intuitive, but my casual observation is that when I work (on film sound) with people who have some degree of hearing loss, they are generally more likely show dissatisfaction when hearing AAC, MP3, EAC3 etc than I or my peers are. Not sure about other codecs.

This sort of makes sense to me, as my understanding is that the content a lossy codec keeps is "chosen" based on models of "normal" hearing. So someone who fits the particular model perfectly will be less able to hear the artefacts than someone who's an outlier in terms of spectral and temporal masking; regardless of how good their hearing is in the real world. (right...?? or wrong?)

My current guess is that maybe because of the spectral masking assumptions used in these codecs, those with hearing loss do not conform to the model. By that I mean for example in very simplistic terms, if it's assumed that a strong signal at 13.0kHz will mask a weaker one at 12.9kHz but the listener can't hear 13.0kHz in the first place because they have a hole in their FR at that spot frequency, it will reveal the codec's hand at 12.9k for that listener. Whereas, for someone who hears the 13.0kHz, the presence (in PCM or lossless) or lack (in lossy) of the lower level content at 12.9kHz is indiscernible. Maybe?

This is probably correct. In 2000 the German computer magazine ct' performed an extensive listening test between uncompressed audio and MP3 audio in varying bit rates. With bit rates at 256 kBs and higher no one could hear any differences between MP3 and WAV, not even trained mixing and mastering professionals. The only exception was a young student who could also differ between WAV and 320 kBs MP3. He confessed to suffer from a hearing defect, and the conclusion was that he couldn't hear some masking signals so the loss of the signals masked by the masking signal in the MP3 audio became apparent.

 

[Deleted member 23982]

i found that the most obvious things to look for on mp3 vs flac are pronauncement of the letter k,p,b,f,s,z,t (so the harsh ones in general with mics) they just sound smoother with flac and are probably the easiest thing to detect it, they sound more "annoying/in your face" with mp3

the second one is high frequencys in general, they will sound more clear (maybe better transients etc) with flac

in general flac sounds cleaner but if you wanna try to listen for it try the above 2 "methods", specially the first one

maybe that helps in any way to listen for yourself closely, once again, tho im sure you already tried that
i should also say i have studio monitors as "hifi-setup" tho im sure you have too and noticed on less high-end system it gets way way way harder to detect method 1 or in general mp3 vs flac, its always finding the weak link first with this stuff and specially cables for example also have a very nice effect on method 1

 

[bravomail]

I dunno. Maybe my observations are purely co-incidental! Does anyone have any thoughts, experience or research on hearing damage / aging vs lossy compression? Only out of curiosity...!

I don't have hearing loss so I will draw analogy with vision loss. You won't hear differences in MP3. Likely.
I know with aging u will hear ess-es better, lower freqs less.
The thing with low bitrate files I noticed, since switching to HiFi DACs/Amps/Headphones - low bitrate files sound hollow, lacking sub-details, that slight audio noise/details, lifeless, and to the degree - scary, as not from this world.