Understanding Audio Frequency Response & Psychoacoustics (Video)

[paolomo]

@Robbo99999 I deleted my post, because maybe I had misunderstood the intentions of our friend.

 

[Robbo99999]

@Robbo99999 I deleted my post, because maybe I had misunderstood the intentions of our friend.

(I don't think any need to delete, not a big deal)

 

[AdamG]

Thread cleaned up of overly aggressive and basic trolling content. Warning issued where appropriate.

 

[Arc Acoustics]

Nice job!
Here we are "Place of facts" !

 

[AlexHRider]

Thats a great video for sure, also that clarity of presentation makes them so effectives.

 

[Art of sound]

psychoacoustics:
outer(softer) portion of the basilar membrane => more sensitive to higher frequencies. and vice versa.

If someone could study my ear, id get more data to try and relate my low frequency sensitivity and annoyances that go with it in its complete absence.

Id also talk more people into this study, those I've observed that don't enjoy music or go further even to avoid at some cost.

We still don't know how the sensitivity level scales other than the 'equal loudness contour curve' i presume. let me know your thoughts as i learn more as it marks the end of a year as an audiophile. (i refused to accept it all my life)

 

[Suffolkhifinut]

Human hearing isn’t linear it’s logarithmic. This needs to be taken into account when interpreting test data.

 

[Art of sound]

What i understood from amir's video and article: When we read measurements of speakers from the spniorama graph (spl vs frequency), the changes in the lower frequencies are far more audible than the higher ones.

I guess the explanation of ERB itself was a difficult thing to grasp not just for me but for a lot of people based off the comments. i will now refer ERB as 'human resolution filter' going forth. (more here: https://sci-hub.st/10.1016/0378-5955(90)90170-t)

Amir was never wrong.
personally, i don't feel the need to learn all the intricate details, only up the point of understanding it generally so i can reason the beauty of sound.
I don't want to explain though: Feynman's example comes to mind.

 

[Bruce T]

Amir is a wonderful consumer advocate, whom I respect for his knowledge and experience, but I must reluctantly agree with Andrej and jaakkopasanen. Non-technical explanations of this topic are difficult to find, so I'm curious where jaakkopasanen found the plot of ERB in octaves. The best corroboration I could find was the Wikipedia article on "Auditory masking" (see first link). Refer to Section 2.3 "Lower Frequencies." Open Figure B, which shows the masking frequency in the top right corner of each graph, also represented by the red line. The x axis shows the signal frequency. The masking frequency is fixed, while the signal frequency varies. The y axis shows the masking dB level.

The first paragraph isn't relevant to this topic, but it contains useful information. As the masking frequency rises in dB for the 1000 Hz graph, so does the masking of the signal frequencies above it. The masking frequency is much more effective at hiding signals above it than below it. The second paragraph is relevant: "Figure B also shows that as the masker frequency increases, the masking patterns become increasingly compressed. This demonstrates that high frequency maskers are only effective over a narrow range of frequencies, close to the masker frequency. Low frequency maskers on the other hand are effective over a wide frequency range." To compare the graphs, I magnified the browser window to 400%. These are only rough estimates. The 250 Hz graph covers 125-4,000 Hz, which is 5 octaves. The 2,000 Hz graph covers 1,000-8,000 Hz, which is 3 octaves.

Additionally, as Art of sound mentioned, the ERB doesn't completely factor in "Equal-Loudness Contours." Human hearing is most sensitive between 2-5 kHz (see second link), so how can we excel at distinguishing pitches above 5 kHz? We can't. At a certain point, the higher frequencies start to sound like vague whistly tones without a distinct pitch. If you have any doubt, listen to the frequency sweep posted by audiofooled on Page 5.

Auditory masking - Wikipedia

en.wikipedia.org

Hearing range - Wikipedia

en.wikipedia.org

 

[Suffolkhifinut]

Amir is a wonderful consumer advocate, whom I respect for his knowledge and experience, but I must reluctantly agree with Andrej and jaakkopasanen. Non-technical explanations of this topic are difficult to find, so I'm curious where jaakkopasanen found the plot of ERB in octaves. The best corroboration I could find was the Wikipedia article on "Auditory masking" (see first link). Refer to Section 2.3 "Lower Frequencies." Open Figure B, which shows the masking frequency in the top right corner of each graph, also represented by the red line. The x axis shows the signal frequency. The masking frequency is fixed, while the signal frequency varies. The y axis shows the masking dB level.

The first paragraph isn't relevant to this topic, but it contains useful information. As the masking frequency rises in dB for the 1000 Hz graph, so does the masking of the signal frequencies above it. The masking frequency is much more effective at hiding signals above it than below it. The second paragraph is relevant: "Figure B also shows that as the masker frequency increases, the masking patterns become increasingly compressed. This demonstrates that high frequency maskers are only effective over a narrow range of frequencies, close to the masker frequency. Low frequency maskers on the other hand are effective over a wide frequency range." To compare the graphs, I magnified the browser window to 400%. These are only rough estimates. The 250 Hz graph covers 125-4,000 Hz, which is 5 octaves. The 2,000 Hz graph covers 1,000-8,000 Hz, which is 3 octaves.

Additionally, as Art of sound mentioned, the ERB doesn't completely factor in "Equal-Loudness Contours." Human hearing is most sensitive between 2-5 kHz (see second link), so how can we excel at distinguishing pitches above 5 kHz? We can't. At a certain point, the higher frequencies start to sound like vague whistly tones without a distinct pitch. If you have any doubt, listen to the frequency sweep posted by audiofooled on Page 5.

Auditory masking - Wikipedia

en.wikipedia.org

Hearing range - Wikipedia

en.wikipedia.org

Interesting post! Can’t totally agree with your final paragraph which seems to suggest any sound above 5kHz can’t be discerned. Human hearing is most receptive around human voice frequencies, to say it can’t be of any use over 5kHz is just plain wrong. Occasionally play a Nordost test CD where one track starts at a 100Hz tone, every time a different frequency is played you are informed. At 1kHz it’s increased in 1kHZ step, even in my decrepitude tone bursts exceeding 5kHz are clearly audible.

 

[Bruce T]

Sorry, I wasn't saying we can't discern tones above 5 kHz, only that we can discern tones more accurately from 2-5 kHz. Critical Bands, the basis of ERB, claim we can hear best at the highest frequencies, but it contradicts Equal-Loudness Contours. Funny, since both concepts were developed by Harvey Fletcher (collaborating with Wilden Munson for the "Fletcher-Munson curves"). I was trying to reconcile the two concepts. If anyone can distinguish tones more accurately above 5 kHz than from 2-5 kHz, then please speak up.

 

[Suffolkhifinut]

Sorry, I wasn't saying we can't discern tones above 5 kHz, only that we can discern tones more accurately from 2-5 kHz. Critical Bands, the basis of ERB, claim we can hear best at the highest frequencies, but it contradicts Equal-Loudness Contours. Funny, since both concepts were developed by Harvey Fletcher (collaborating with Wilden Munson for the "Fletcher-Munson curves"). I was trying to reconcile the two concepts. If anyone can distinguish tones more accurately above 5 kHz than from 2-5 kHz, then please speak up.

My apologies if I misread your post. It would be very difficult to prove one way or the other, although you are more than likely to be right. Our hearing is optimised for listening for the human voice.

 

[Art of sound]

Sorry, I wasn't saying we can't discern tones above 5 kHz, only that we can discern tones more accurately from 2-5 kHz. Critical Bands, the basis of ERB, claim we can hear best at the highest frequencies, but it contradicts Equal-Loudness Contours. Funny, since both concepts were developed by Harvey Fletcher (collaborating with Wilden Munson for the "Fletcher-Munson curves"). I was trying to reconcile the two concepts. If anyone can distinguish tones more accurately above 5 kHz than from 2-5 kHz, then please speak up.

we hear logarithmically wrt resolution, do not conflate this with loudness, masking like jaak
let me restate amir's https://www.audiosciencereview.com/...ds/perceptual-effects-of-room-reflections.13/

....the changes in the lower frequencies are far more audible than the higher ones......

 

[Bruce T]

Art of sound, I'm trying to connect the dots. My science and math background is limited. Perhaps I shouldn't confuse loudness and masking with resolution, but it could be argued that tones in the most sensitive region of human hearing should provide better resolution. Someone could also make a convincing argument that we hear best in the center of human vocals (Piano C4, "Middle C," at 261.6 Hz). Even lower, they could argue for the fundamental frequencies of male and female speech: 125 and 250 Hz. However, the lowest frequencies are not more audible. Below 100 Hz, pitch definition gets progressively worse. Below 65-70 Hz, it gets murky. Below 60 Hz is traditionally sub-bass.

 

[Art of sound]

try performing an experiment to differentiate 100-120-150 hz tones and 1100-1120-1150 hz. Its not resolution per se, its the changes. this applies in looking at the resonances when measuring speakers/headphones. maybe resolution doesn't even apply to anything below 150Hz so you're likely right on that front.

 

[Bruce T]

Out of curiosity, I tried a variation of the experiment you suggested. Just below Middle C on the piano are A3, A#3, and B3. Tones are 220, 233, and 247 Hz. Near the top of the piano are A7, A#7, and B7. Tones are 3,520, 3,730, and 3,951 Hz (first link). I spread six tiny browser windows across the desktop like playing cards with "Online Tone Generator" (second link), typed in the six tones, and selected square waves--the edgiest tone, similar to an old video game. Then I quickly hit play and stop for each tone. The lower tones were far more distinct than the higher tones. Amir is correct. Lower frequencies are more audible. I still disagree in the midbass to low bass. Feel free to try it yourself.

Piano key frequencies - Wikipedia

en.wikipedia.org

One moment, please...

 

[Art of sound]

Out of curiosity, I tried a variation of the experiment you suggested. Just below Middle C on the piano are A3, A#3, and B3. Tones are 220, 233, and 247 Hz. Near the top of the piano are A7, A#7, and B7. Tones are 3,520, 3,730, and 3,951 Hz (first link). I spread six tiny browser windows across the desktop like playing cards with "Online Tone Generator" (second link), typed in the six tones, and selected square waves--the edgiest tone, similar to an old video game. Then I quickly hit play and stop for each tone. The lower tones were far more distinct than the higher tones. Amir is correct. Lower frequencies are more audible. I still disagree in the midbass to low bass. Feel free to try it yourself.

Piano key frequencies - Wikipedia

en.wikipedia.org

One moment, please...

I guess given most devices clear 16 bits of distortion in SNR and multi tone, measurements do little to identify issues these days and i'm glad that i entered this realm at the right time. if you're looking at 85db SNR in a speaker amp this topic is relevant and the spike is below 500Hz you should be worried it will creep into the sound given how sensitive we are to close frequencies in the low end. Good to have the data to avoid badly engineered products. Having 120db snr vs 105 means squat for sound. In real world this bothers me all the time with refrigerator, HVAC, road, tire noise etc which are almost never masked with other sounds ( >100Hz&>20db i can tell). I usually run something in the background like https://mynoise.net/NoiseMachines/whiteNoiseGenerator.php

 

[Bruce T]

True, but part of the fun of being an audiophile is buying products that are overkill in SNR, bandwidth, speed, dynamics, build quality, etc. Sigh, we fight noise pollution from many sources during our daily lives, but it's almost unavoidable. Ha, I get using a white noise generator. Double pane windows help, too.

 

[Art of sound]

True, but part of the fun of being an audiophile is buying products that are overkill in SNR, bandwidth, speed, dynamics, build quality, etc. Sigh, we fight noise pollution from many sources during our daily lives, but it's almost unavoidable. Ha, I get using a white noise generator. Double pane windows help, too.

spoken like a man who has never driven in northern california. in any case my next car purchase would be more oriented to comfort.
I like clean looking aesthetics but i won't compromise good design and rewarding the right company.
I am not that kind of audiophile. id still recommend people to buy sonos until i see a worthwhile 2.0 or 2.1 that plays out of the box.

 

[Art of sound]

Burson Soloist 3XP Review (Headphone Amp)

This is a review and detailed measurements of the Burson Soloist 3X Performance balanced headphone amplifier. It is on kind loan from a member and costs US $1,200. The overall package feels heavy and solid but I am not a fan of lack alignment of all the connectors, holes, etc. Volume control...

www.audiosciencereview.com

look at the two spikes at 2k and 3k. it could have been designed better and it should be looked into by the manufacturer but one cannot hear these in day to day listening tests in most use cases especially if the said device is above 100db in all aspects forgoing the two spikes. measuring loudspeakers is far more useful though. end devices are the real deal.