• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

high resolution listening test

OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
(maybe I posted this here already) In a similar vein:
"48 dB"

Well, that corresponds to about 7 on this test.

Thank for you for participating!


The rationale for this test was that I was interested in some back-of-the-envelope numbers for estimating actual real world requirements for dynamic range.

All along I've known that the nature of the test signal would somewhat bias the outcome. That is life in the real world in any kind of even semi-serious testing.

So, ideally and somewhat obviously repeating it with different audio selections would shed more light.

I won't amuse you by revealing the provenance of the audio clip that I used this time through. :)

Now, I've got to decide whether trying to pick the most dissimilar test signal would be worth the trouble, or even what that highly dissimilar test signal would be.

4 KHz pure tone? Warble tone? Selection from some utterly compressed heavy metal?

How much to punish us all? :)
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
You really don't need to go further. My similar test was inspired by curiosity about research by Dolby Labs saying you can need 120 db (or was it 125 db?) dynamic range for full fidelity domestic playback. The reason this is so is our hearing is most sensitive in the 2-5 khz region. Domestic noise floors may be 35-40 db SPL over the audible band, but in that most sensitive frequency range they commonly are around 10 db SPL. As we can hear into noise 10-15 db that puts us all over a basic 0 db SPL perceived noise floor. My little on line sample indicated people at their normal listening levels (which I think will be in the 75-80 db SPL range) could hear somewhere around -70 to -80 db into the test tone. Yours isn't indicating too much differently than that.

So due to all that you'll not go far wrong to assume a 0 db SPL effective noise floor for most playback. I think the reason Dolby Labs may have said 125 db is in slightly quieter rooms (which some people have) and with good healthy hearing in the 2-5 khz range people can hear a few db below 0 dbSPL. And in quieter rooms the noise floor in those frequencies will be below 10 dbSPL. So - 5 or -6 db SPL to 120 db SPL is possible.

Now most domestic systems can't reach 120 db SPL. So you'll need whatever they do reach. If it reaches 105 db SPL cleanly then 105 db dynamic range is enough. Of course all you have to do now is go find some musical recordings that exceed 100 db dynamic range that aren't diabolical test signals. ;)
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
http://www.aes.org/e-lib/browse.cfm?elib=7948

This is the Louis Fielder Dolby Labs AES paper I was referring to about needed dynamic range.

Here is the text to that paper without the charts.

http://www.zainea.com/Dynamic range.htm

NOTE: the above is white text on a red background. I copy and pasted the text into word processor software (abiword) so as not to go blind.

Amir has addressed the issue here, and has some of the charts that belong in the Fielder paper in this thread.

https://audiosciencereview.com/forum/index.php?threads/dynamic-range-how-quiet-is-quiet.14/
 
OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
There were comments about the audio selection that I used in my first dynamic range test - which was a brief recording of a male voice. This time I tried to created the most obvious and easy to hear test I could think of - a 3 KHz tone, with interruptions to make it stand out better. Each test is 5 seconds long. The tests are separated by a 1 second of silence.

Usual drill - set the first test for a comfortable but loud listening level and then count how many other tests you can hear as the level decreases.

Link to test file in a neutral place: https://www.dropbox.com/s/4yz96cxig0e3pen/dynamic range test 3 KHz tone.flac?dl=0
 
OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
http://www.aes.org/e-lib/browse.cfm?elib=7948

This is the Louis Fielder Dolby Labs AES paper I was referring to about needed dynamic range.

Here is the text to that paper without the charts.

http://www.zainea.com/Dynamic range.htm

NOTE: the above is white text on a red background. I copy and pasted the text into word processor software (abiword) so as not to go blind.

Amir has addressed the issue here, and has some of the charts that belong in the Fielder paper in this thread.

https://audiosciencereview.com/forum/index.php?threads/dynamic-range-how-quiet-is-quiet.14/


I'm familiar with the Fieldler paper to say the very least. I've read it dozens of times since it first came out. Being an AES member I can download it at will, and also keep a copy of the PDF just in case I want to read it again. I have some comments about it, too. :)
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
There were comments about the audio selection that I used in my first dynamic range test - which was a brief recording of a male voice. This time I tried to created the most obvious and easy to hear test I could think of - a 3 KHz tone, with interruptions to make it stand out better. Each test is 5 seconds long. The tests are separated by a 1 second of silence.

Usual drill - set the first test for a comfortable but loud listening level and then count how many other tests you can hear as the level decreases.

Link to test file in a neutral place: https://www.dropbox.com/s/4yz96cxig0e3pen/dynamic range test 3 KHz tone.flac?dl=0
@Arnold Krueger

CAUTION TOO Loud
At my normal musical listening level the first part is way too loud. You'll ruin your threshold for a few minutes with that. Need to start at a lower level. And fair warning to anyone that downloads.
 
Last edited:

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
By way of explantion of I've calibrated using some of Dolby's ideas. 83 db pink noise in one channel. Gives me about 20 db headroom and a bit more.

I listen to music most often from that level to 6-8 db less than that. I started at 6 db less than reference. It isn't speaker damaging loud, but it is loud enough you get temporary threshold lift from it. This is not a good way to do this dynamic range listening. Starting at -15 or -20 db would have been plenty sufficient. The beginning tone is well into the mid to upper 90 db SPL range at 3 khz.
 
Last edited:

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
Okay, edited out the first overly LOUD tones. Reduced the level of the second group of tones. All the rest unmolested.

Did the test during a thunderstorm. Thunder was distant and rain was moderate so noise level a bit above normal.

Counting the edited out group as one I heard a total of 9 groups of tones. Maybe I could get one more without the patter of rain, maybe not. BTW, this was with volume set so -20 db pink noise would be 86 db with two channels playing. I probably would have scored one less a few db lower on the volume.
 
OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
Okay so comment.

The biggest problem that I have with Louis Fielder's landmark paper or should I say stumbling block paper about dynamic range is that it describes experiments that are very unrealistic. He's not alone in this kind of disconnection with reality.

Let me illustrate this problem with some examples.

One common disconnection of experimentsand their interpetation from reality is the common practice using the threshold of pain and the threshohld of hearing as being definitive of the dynamic range of human hearing. However, exposure of human ears to sounds sufficiently loud to cause pain is well known to cause the human hearing system to relaibly experience a phenomenon that is so common that it has had a given technical name for decades: "Temporary Threshold Shift" (TTS) Google defines it without searching as: "A temporary threshold shift is a temporary shift in the auditory threshold. It may occur suddenly after exposure to a high level of noise, a situation in which most people experience reduced hearing. A temporary threshold shift results in temporary hearing loss." Based on my personal experiences, TTS can commences within a minute or less from the onset of the the exposure to the high level noise, worsens with the length and intensity of the exposure, and can last for 12 hours or more. The Wikipedia article https://en.wikipedia.org/wiki/Occup...shift_(hearing_loss)_after_noise_exposure.jpg provides the attached illustration:
1024px-Temporary_threshold_shift_%28hearing_loss%29_after_noise_exposure.jpg


I find it rather striking that threshold shifts of up to 50 dB are shown in this reference, but one rarely hears of this effect, and it is certainly not mentioned in Fielder's paper's text. If you include the effects of TTS, the dynamic range of human hearing can be justifiably adjusted by up to 50 dB! Depends how you design the experiment. I suggest that if you design the experiment to be similar to a person listening to the loud and soft passages of music, the 50 dB TTS may become operative. If we compare this to the loudest sounds portrayed in the paper, we now have only 78 dB dynamic range, a number that is well within the much criticized 16/44 CD format.

A second problem is that in general this paper vastly overstates how quiet the quiet portions of the real world quiet passages are. For example, a number of 128 dB is given for the peak level of rock music in the paper's table 1. However, should the music cease playing, the room does not go down to absolute silence. Even after the crowd has become relaxed, the room is probably still ringing with sound at something like 80-90 dB SPL or more. The actual real-world dynamic range that you'd witness if you were there might be only 38 dB. If we chase the crowd out, then we are left with the SNR of the musical instrument amps which typically have a rather poor SNR by high fidelity standard. On a good day, the dynamic range at the live performance might be still only be about 50-60 dB, not the 128 dB claimed.

Another example would be live classical performance which of course includes the sound of the audience. Even they are as still as possible they still make noise. They breathe, their digestive systems gurgle, and as they fidget, their clothing rubs and makes perceptible noise Chase out the audience and you are still left with 100 or so musicians.

The paper does not seem to mention "Room Tone" which always exists and is generally way above threshold. Common numbers for background noise in venues and listening rooms seem to be absent people. It is hard most of us to listen without being present! :)
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
Yes, I am aware of TTS which is why I complained about your overly loud test signal.

The Dolby research seemed intent on finding the most extreme limits that might be required. I don't fault it for that as it seems a good approach to begin fleshing out areas where there is no good data. For instance in my domestic use I'm not shooting for 120 or 129 db SPL ever. 110 db seems the most one would want to use at home and I can adjust from their research data accordingly. In at least some of the audible range domestic listening rooms can at times be quieter than public venues. Getting performance down to 10 db SPL likely leaves very little on the table and most of the time will leave nothing on the table that could be heard. Which is why I've had the idea if your gear is clean over 100 db range and properly gain matched you are close to as good as it can get. More wouldn't hurt, but you'll be somewhere between darn nearly no difference to no actual difference can possibly be heard. Then given that by some accounts our hearing has 60 db instantaneous dynamic range with how sound builds and decays in a room even 100 db might be more than is needed. Since it is readily achievable for everything except the recording venue, and your listening venue that tells me where improvements lie and these days it isn't with gear other than transducers and environments.
 
OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
@Arnold Krueger

CAUTION TOO Loud
At my normal musical listening level the first part is way too loud. You'll ruin your threshold for a few minutes with that. Need to start at a lower level. And fair warning to anyone that downloads.

How about this? Shifted the whole thing 10 dB down.

https://www.dropbox.com/s/50ccbna8l498yem/dynamic range test -10 dB 3 KHz tone.flac?dl=0

BTW, for reference, here is the dynamic range test file based on a male voice announcement:

https://www.dropbox.com/s/n50g7b6hs14x8dj/high resolution listening test audacity HD.flac?dl=0

And, just for grins, a paper on listening at loud levels:

https://www.dropbox.com/s/dos4zexnmaga6ur/MIHD_Position_Statement_FINA.pdf?dl=0
 
Last edited:

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,198
Likes
16,981
Location
Riverview FL

"Summary

There is widespread concern regarding risk to hearing health due to the increasingly common worldwide use of personal audio systems (PAS) (e.g., iPods, tablets, smartphones) and other recreational music listening behaviors."

How is some conscientious person supposed to measure their exposure level using IEMs or headphones in order to remain within the limits given the variabilility in volume settings vs driver levels?
 
Last edited:
OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
"Summary

There is widespread concern regarding risk to hearing health due to the increasingly common worldwide use of personal audio systems (PAS) (e.g., iPods, tablets, smartphones) and other recreational music listening behaviors."

How is some conscientious person supposed to measure their exposure level using IEMs or headphones in order to remain within the limits given the variabilility in volume settings vs driver levels?

The only way I know of to monitor your exposure level via IEMs involves capture of the voltage going into the IEM and analyze that in the light of a technical analysis of the SPL versus voltage properties of the IEM itself.
 
Last edited:
OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
Yes, I am aware of TTS which is why I complained about your overly loud test signal.

The Dolby research seemed intent on finding the most extreme limits that might be required. I don't fault it for that as it seems a good approach to begin fleshing out areas where there is no good data. For instance in my domestic use I'm not shooting for 120 or 129 db SPL ever. 110 db seems the most one would want to use at home and I can adjust from their research data accordingly. In at least some of the audible range domestic listening rooms can at times be quieter than public venues. Getting performance down to 10 db SPL likely leaves very little on the table and most of the time will leave nothing on the table that could be heard. Which is why I've had the idea if your gear is clean over 100 db range and properly gain matched you are close to as good as it can get. More wouldn't hurt, but you'll be somewhere between darn nearly no difference to no actual difference can possibly be heard. Then given that by some accounts our hearing has 60 db instantaneous dynamic range with how sound builds and decays in a room even 100 db might be more than is needed. Since it is readily achievable for everything except the recording venue, and your listening venue that tells me where improvements lie and these days it isn't with gear other than transducers and environments.

The problem I have is that encouraged by the name of the article: "DYNAMIC RANGE REQUIREMENT FOR SUBJECTIVE NOISE FREE REPRODUCTION OF MUSIC" people commonly but apparently mistakenly use this article to estimate and recommend dynamic range requirements for home audio systems.

It seems like at the least, the words "Extreme" or "Ne Plus Ultra" or "Perfectionaistic to the point of being dangerously self-destructive", or "Absolute Maximum" should be in there someplace. :)

What the article seems to be to is a set of numbers that are rarely if ever achieved in real world listening environments, even at the loudest rock concerts.
 
Last edited:

garbulky

Major Contributor
Joined
Feb 14, 2018
Messages
1,510
Likes
827
I think dynamic range tests like these are hugely useful. In a day where we judge dacs by their SNR and their dynamic range capabilities, trying out these tests can show how meaningless extremely low SNR's are in real life music listening. When I tried it, I put the 0db at regular volume for the first "resolution listneing test" after 50 db it was pretty much gone. The other guitar tone had accidentally set it much higher and after about 54-60 db it was gone. If I had set it at deafening volumes, maybe 70 db if I was squinting. Meaningfully after about 48 db it was not really useful to me.

For me something like this really changed my way of thinking about "low level" detail. Different DACs I've heard (subjectively) had different ways of sounding. But if all I can hear to is about that level, then where does the sound really matter? The answer is in the ranges above that. Not in some -90 db sine wave.

But forgetting the -90 db sine wave.... where are the tests that can more articulate what DACs are doing in those louder volumes? I imagine if you run most DACs through -50 db, they are all going to be about perfect with the only difference being some inconsequential dip at 19-20 khz of 1 db. That can't be what's changing my perception of the sound.

Another thing I would be interested in seeing is now hearing it with TWO notes please or two tracks. One placed at a db level below it. That way we can identify masking. And masking is important. Because with a DBT test, if the difference is being masked, by a plethora of louder information, perhaps (not saying for sure!) it may be very hard to tell whether there is a meaningful difference or not - if things are level matched.
And also...if it is masked but really audible without it, does it matter too is another good question.

So is there a possibility of making a two tone test playing at the same time. One less louder (-10 db) than the other.
 
OP
Arnold Krueger

Arnold Krueger

Active Member
Joined
Oct 10, 2017
Messages
160
Likes
83
I think dynamic range tests like these are hugely useful. In a day where we judge dacs by their SNR and their dynamic range capabilities, trying out these tests can show how meaningless extremely low SNR's are in real life music listening. When I tried it, I put the 0db at regular volume for the first "resolution listneing test" after 50 db it was pretty much gone. The other guitar tone had accidentally set it much higher and after about 54-60 db it was gone. If I had set it at deafening volumes, maybe 70 db if I was squinting. Meaningfully after about 48 db it was not really useful to me.

For me something like this really changed my way of thinking about "low level" detail. Different DACs I've heard (subjectively) had different ways of sounding. But if all I can hear to is about that level, then where does the sound really matter? The answer is in the ranges above that. Not in some -90 db sine wave.

But forgetting the -90 db sine wave.... where are the tests that can more articulate what DACs are doing in those louder volumes? I imagine if you run most DACs through -50 db, they are all going to be about perfect with the only difference being some inconsequential dip at 19-20 khz of 1 db. That can't be what's changing my perception of the sound.

Another thing I would be interested in seeing is now hearing it with TWO notes please or two tracks. One placed at a db level below it. That way we can identify masking. And masking is important. Because with a DBT test, if the difference is being masked, by a plethora of louder information, perhaps (not saying for sure!) it may be very hard to tell whether there is a meaningful difference or not - if things are level matched.
And also...if it is masked but really audible without it, does it matter too is another good question.

So is there a possibility of making a two tone test playing at the same time. One less louder (-10 db) than the other.

There is no problem doing such a thing. I just want to think a bit about what frequencies to use. I want them to be at frequencies that are highly audible but I don't want the louder one to mask the softer one.

As far as technical tests go, I have been posting this chart in some of my recent tests over in the Digital Music Player subforum:

FIIO X1 I to Microtrack I 2496 04-05-2018_html_m47ba9df7.png


This test shows how the DAC responds to a full-frequency sound as the level goes down in steps of -10 dB. At the bottom of its useful range the DAC's response breaks up and gets lost in the noise. For comparison, this is a test of somewhat better product:

fr+levels 0-110.png
 
Last edited:

dc655321

Major Contributor
Joined
Mar 4, 2018
Messages
1,597
Likes
2,235
As far as technical tests go, I have been posting this chart in some of my recent tests over in the Digital Music Player subforum:

This test shows how the DAC responds to a full-frequency sound as the level goes down in steps of -10 dB. At the bottom of its useful range the DAC's response breaks up and gets lost in the noise. For comparison, this is a test of somewhat better product:

Very interesting stuff. Thanks!
1) Where is the "Digital Music Player" subforum?
2) Are these products identified there? If not, what are they specifically?
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
I think dynamic range tests like these are hugely useful. In a day where we judge dacs by their SNR and their dynamic range capabilities, trying out these tests can show how meaningless extremely low SNR's are in real life music listening. When I tried it, I put the 0db at regular volume for the first "resolution listneing test" after 50 db it was pretty much gone. The other guitar tone had accidentally set it much higher and after about 54-60 db it was gone. If I had set it at deafening volumes, maybe 70 db if I was squinting. Meaningfully after about 48 db it was not really useful to me.

For me something like this really changed my way of thinking about "low level" detail. Different DACs I've heard (subjectively) had different ways of sounding. But if all I can hear to is about that level, then where does the sound really matter? The answer is in the ranges above that. Not in some -90 db sine wave.

But forgetting the -90 db sine wave.... where are the tests that can more articulate what DACs are doing in those louder volumes? I imagine if you run most DACs through -50 db, they are all going to be about perfect with the only difference being some inconsequential dip at 19-20 khz of 1 db. That can't be what's changing my perception of the sound.

Another thing I would be interested in seeing is now hearing it with TWO notes please or two tracks. One placed at a db level below it. That way we can identify masking. And masking is important. Because with a DBT test, if the difference is being masked, by a plethora of louder information, perhaps (not saying for sure!) it may be very hard to tell whether there is a meaningful difference or not - if things are level matched.
And also...if it is masked but really audible without it, does it matter too is another good question.

So is there a possibility of making a two tone test playing at the same time. One less louder (-10 db) than the other.

Can you list two or more DACs that you find to sound different to you?
 
Top Bottom