AVS Forum banner

361 - 380 of 620 Posts

·
Registered
Joined
·
1,805 Posts
Serious evaluations must be at exactly the same level as determined by instrumentation. Not doing this is probably the number one mistake of hobbyists (and some "scientists" too).
This!
Except you should have typed EXACTLY ;)
And checking that with an SPL meter is bull****, that is way too subject to other vagaries. Voltmeter or die.

Working for a $3 billion audio company, argh the @#$%'ed up listening tests I was forced to attend. With volume not even compensated, or in some cases (different head units for example) you simply could not match the level exactly due to volume control steps. Another time we had a Tripath amp chip inside a head unit, and again the gain was just different. But any equipment with a 1 dB volume adjustment will show how sensitive we are to level, especially if the volume is not so high and thus we are into the loudness curve of our hearing.
 

·
Registered
Joined
·
186 Posts
TCP/IP may rule, but Ethernet (where collision detection lives), is layer 2. The LAN and WAN.

https://miro.medium.com/max/1024/1*17Zz6v0HWIzgiOzQYmO6lA.jpeg
OK, I understand. So I have to ask, what is it that collides? Isn't it packets that collide, not frames? If there's an error or collision, isn't it packets that get re-sent? Obviously, I don't know this stuff anywhere near well enough to explain it, I only threw out a side compliment. When one my PCs talks to a 2nd PC, both on a LAN, aren't they still talking with both TCP and IP? Does anything on our typical home networks talk restricted to only ethernet? What reading I've done, it's always TCP/IP that is what's mentioned when it comes to actually moving the data from point A to point B on a network.
 

·
Registered
Joined
·
15,411 Posts
Discussion Starter #364 (Edited)
This!
Except you should have typed EXACTLY ;)
And checking that with an SPL meter is bull****, that is way too subject to other vagaries. Voltmeter or die.

Working for a $3 billion audio company, argh the @#$%'ed up listening tests I was forced to attend. With volume not even compensated, or in some cases (different head units for example) you simply could not match the level exactly due to volume control steps. Another time we had a Tripath amp chip inside a head unit, and again the gain was just different. But any equipment with a 1 dB volume adjustment will show how sensitive we are to level, especially if the volume is not so high and thus we are into the loudness curve of our hearing.
Well, I did use the word "exactly" but not in that size text, no. :)

When I'm betting money I use a volt meter to level match, yes, but in setting up this test I realized something along the way. Level matching is not an exact science when working with dynamic music. What I did is I used a fixed, constant, 1kHz sine wave as a calibration tone to match A to B files but think about this:

- Are we matching the loudest peak in the music?
- Are we matching the average level of all the peaks?
- Are we matching the average level?
- Are we matching at 1kHz?
- Are we matching at 100Hz?
- Are we matching at 10kHz?
- Are we matching with weighting?
- Are we matching with A, B, C, or K-weighting?
- Are we matching the louder of the two channels and using which of the criteria above do so?
- Are we matching the sum of the Left and Right channels added together?

Different methods get different results and I speak from experience:

I notice for example when using my mono, 1kHz calibration tone method one test instrument I use reports "You are spot on" whereas Replay Gain in Foobar reports "You are off by .04 dB".

Differences under .2 dB are usually fairly safe but if you have the means to measure that you might as well go for under .1 dB, just to be sure.

Food for thought.
 

·
Registered
Joined
·
20,839 Posts
OK, I understand. So I have to ask, what is it that collides? Isn't it packets that collide, not frames? If there's an error or collision, isn't it packets that get re-sent? Obviously, I don't know this stuff anywhere near well enough to explain it, I only threw out a side compliment. When one my PCs talks to a 2nd PC, both on a LAN, aren't they still talking with both TCP and IP? Does anything on our typical home networks talk restricted to only ethernet? What reading I've done, it's always TCP/IP that is what's mentioned when it comes to actually moving the data from point A to point B on a network.
This discussion detracts from the OP's topic and requires too much education to clarify/explain. Sorry.
 

·
Registered
Joined
·
186 Posts
I've always loved the McGurk Effect and one I can't remember the name of where changing the lighting or using food coloring can make identical samples of milk taste different. I believe that people actually hear a difference when they change a tonearm or use an $8000 power cable despite science saying it's not possible because they're getting more information than just the sound. The pretense of audiophilia is that "golden ears" are immune to these effects and the way they describe what they're hearing is objective in that someone else should be able to hear these differences if only they were as good at listening as a reviewer. It's an advanced form of the No True Scotsman fallacy.


https://www.youtube.com/watch?v=G-lN8vWm3m0
Yeah, the McGurk effect is a weird one. What everyone should understand is how these illusions are universal, that knowing what's going on doesn't make them go away anymore than optical illusions go away when you know it's an illusions. It's all in the wiring and we have some bugs in the wetware.



I wouldn't say it's the case that there is "more information than the sound", it's really the case that what you 'hear' is created by the brain, often in direct contrast to what sound is reaching the ear. If you get into color, man, that's a whole lot crazier and weirder. Think about timing, if you touch your ankle and your nose at the same time, while looking at your ankle getting touched, the timing of those 3 distinct events is way off from simultaneous, but that ain't what you perceive, you perceive the 3 events as happening at exactly the same time. Think about what kind of massive data processing is going on for that to happen.
 

·
Registered
Joined
·
358 Posts
The Hawthorn Effect - the early stages of Industrial Psychology - seeking to assess worker productivity based on lighting changes in the Western Electric plant in Hawthorne. Gee, more light helps! Gee, less light helps! Gee, ANY CHANGE helps! Simply changing the worker environment demonstrated increased productivity IIRC.

My personal take on this relates to the issue of senation-seeking, that trait that can certainly contribute to various addictions. We get bored with the 'same ole, same ole' which is likely why NOBODY around here can EVER be satisfied with their home theater :p
 

·
Registered
Joined
·
15,411 Posts
Discussion Starter #369 (Edited)
With volume not even compensated, or in some cases (different head units for example) you simply could not match the level exactly due to volume control steps. . . . . But any equipment with a 1 dB volume adjustment will show how sensitive we are to level, especially if the volume is not so high and thus we are into the loudness curve of our hearing.
Maybe in a car it isn't all that important but not since the 90's with gear I've sold and owned have I seen people locked to only 1dB-step volume controls, but I see even some pricey brands like Rotels' upscale series Michi seem to be regressing to this crude design. Not fun for people who need to precisely level match.:(

Even my bedroom AVR I bought for ~$170 $255 has half dB steps. The Rotel Michi preamp I linked to, on the other hand, has no tuner, power amps, Dolby decoding, nor HDMI support and it retails for $4k. The stereo amp add-on is $7K or you can go for 2 monobocks for $14k for a pair, as I understand it. I'd bet money the sound is indistinguishable from my cheap AVR with both amps kept below clipping.
 
  • Like
Reactions: mechtheist

·
Registered
Joined
·
2,884 Posts
I spent hours and hours of my time putting this all together including filming the entire file creation steps to prove that they are what I say, and, including the original video teaching people how to download, install, and use Foobar ABX. I have done all this in the interest of science, the pursuit of truth.

You and others could show your appreciation simply by taking the test and posting your score sheets, just like me and few others already have. Is that such a big request considering what I've put into creating this test?
Honestly, I already know the outcome...my ears are not the best -- first I'm "old" (enough not to hear HF, lol) and they were beat up while I was younger...wish I could go back in time and treat them better. More, I don't have the patience to spend my time looking for differences I doubt audibly exist, especially not through my ear-brain interface. I may as well flip a coin to take the test. But, you're preaching to the choir in me, so....I just like that you did it for others to see.

If, amidst my two and a half jobs, a wife, and two kids, I find time to sit down and give the test the patience it deserves for me to have chance at success then I will. In the meantime, a thanks will have to suffice, sorry :(.

I'm locked to SHA-1, even though I know SHA-256 is stronger, because that's what the Foobar ABX score sheet shows. The SHA-1 were inserted into the Foobar ABX test to ensure test takers don't cheat by simply inserting their own rigged files which merely have the same name but are actually completely different song files. I realized I could similarly use it to prove the files I create for the test are what I say by showing the SHA-1 of each file right as I created it and that was the purpose of the authentication video I linked to earlier.
Fair enough!

Had I re-encoded into FLAC, regardless of most people understanding it is lossless, instead of the original .wav files ripped directly from the CD, then there would have been a contingent of complainers that I had "manipulated the master and mangled its purity", and that's supposedly why listeners can not hear a difference. Using .wav stops that argument in its tracks.
Likewise, fair enough!

*snip*
It is good enough for all formats in any sane living room music reproduction scenario because the limiting factors are not really so much "what is the dynamic range of the music signal and the format in use at the time", it is a combination of (off the top of my head):

- the dynamic range and SNR of the actual recording including the mics, mic preamps, and mixing consoles/processors used to make them
- human hearing limits
- living rooms' ambient noise
- headphones/speakers used as well as the SNR of the system's preamp and power amp

Deceptive marketers would have us believe 120dB is the goal and anything less is discernibly flawed but that's a lie to sell their high end gear. They back their lie by cherry-picking data from Feidler's study and other related works by omitting critical facts. For instance, it is true that under laboratory conditions people with good hearing can detect 0dBSPL or even a tad lower at specific frequencies, in ideal laboratory circumstances. What the deceivers fail to explain is that's only after a 20 minute exposure to dead silence. Hearing is just like vision: in order to detect the very lowest levels of a signal (such as being able to read text from the light of a nearby firefly only) you had better have been in total darkness for 20 minutes prior to the test to get the best possible night vision ("dark adaptation" I think it is technically called).

So too with hearing: when that super quiet -120dB musical passage comes along in a Hi-Scam, oops, I mean Hi-Re$ file, you had better hear it only after having heard nearly dead silence in the musical work for the preceding 20 minutes or your low signal hearing sensitivity is shot. There are no such songs I know of: it is a made up fantasy.
Agree, but there exists a need for better DACs than that, right? What about when driving volume ladders or power amplifiers having input sensitivities of 8v. Or, when an AVR pre-adjusts its digital signal chain levels downward to make digital headroom for REQ or other things (e.g., bass management)? Does the DAC's DR matter more then? If I boost my bass by 10db or more, I have changed the DR of the piece, right? And I boost my bass by more than that below 40hz! I LOVE bass, lol!

To learn what actual dynamic range you need for your room you can take this simple test in under a minute:
https://www.audiocheck.net/audiotests_dynamiccheck.php

The deceivers of course poo-poo this test because it exposes their scam quite elegantly.

The Audiocheck site explains the industry's con well:

"Interestingly, much emphasis is put on 24-bit audio recordings nowadays, with a dynamic range exceeding 140dB. Our example is only 16-bit, with a maximum dynamic range of 96dB, yet that should be plenty. Judge for yourself. "
This I was interested in. So, I went and found my HDMI adapter and connected my laptop to the NAD M17 running my main system (a prereq for the ABX, anyway :D). If I creep the volume up until the pink noise is almost "intolerable" (*) then, at 0dB master volume, I can just hear the voice-over at 78dB below full scale, after that nada. For the record, at 0dB master volume, A-weighted on my cell-phone sound meter (for whatever that is worth) I see an 81.7dB reading during the pink noise passage of the test (if I hold peaks, I read an 83dB peak).

This is in my basement, lights on (they are florescent so I have to listen through a touch of noise there), during the day (it's a little louder down here, but not much), and the dehumidifier is (presently) off. I sit about 12' from my speakers.

(*) due to the aforementioned bass curve, right around -10dB on my receiver volume I can hear my bathroom door shaking a touch, during the pink noise passage. Then I push it to 0dB for the final hoorah. I could go higher on the volume, and arguably should if the FS pink-noise reading should be 85dB. However, first, I honestly don't want to piss everyone in the house off any more than I already am, lol. But, more importantly, the pink-noise is already getting so loud that my low-level hearing is, as pointed out, blown out by the preceding pink noise and making it hard to hear the voice over...I don't know if "more volume" will help.

A "better" test would be, IMO, to remove the pink-noise after the first test, subsequently playing just the voice lower and lower. This way I can increase my MV until the system noise overwhelms the voice over. Now I know the low-level capability of my system. Then, if I dare, I can restart the test and hope the pink noise doesn't destroy my gear, lol. In my case, however, I think I'd just be running for cover from the enormous pink noise levels while my gear laughs at the destruction it causes...at least to the paint on my bathroom door, lol. Well, that or I clip the signal chain...also completely possible!
 

·
Registered
Joined
·
780 Posts
I wouldn't say it's the case that there is "more information than the sound", it's really the case that what you 'hear' is created by the brain, often in direct contrast to what sound is reaching the ear.
The "more information" I'm talking about is the knowledge that a component has changed, not anything audible.
 

·
Registered
Joined
·
15,411 Posts
Discussion Starter #372
If one rides the gain of their MV during a "What dynamic range do I really need?" test you will convince yourself CD's 96dB range is inadequate but that's cheating. [The DR is actually better than that but for sake of argument let's just say it is only 96dB.] In real life we aren't allowed to crank our volume so the loudest possible signal on CDs (0dBFS) clips. [And that's why the Audiocheck Test is smartly designed] We are locked to never hitting clipping. Only in the artificial scenario where we no longer are concerned with clipping the peaks do we find ourselves in a position of "I hear hiss from this format so it's not good enough".

As long as you have enough preamp level controls in the chain you'd find 120 dB DR isn't enough, then 140 dB isn't enough, then 160 dB isn't enough, because every time the music plummets 20 dB in level you cheat and bring it back up, including its noise floor, with your endless artificial added gain volume control cheat where you don't care that the loud parts are severely clipping, all you care about is "How do I get to hear the background hiss?".
---

If one has a not so great AVR that injects discernible noise into the music when room correction DSP is engaged that problem doesn't disappear if you convert to 192kHz/24bit Hi-Re$ PCM incoming files, it remains. The problem occurs down the chain from the format limitations.

Agree, but there exists a need for better DACs than that, right?
In production work where we might bounce the signal from one device to another, and another, and another, and another. . . the noise adds up cumulatively so we need better in studio/production use, yes.

Got to break for now.
 

·
Registered
Joined
·
1,187 Posts
This ABX test also determines if the listener can audibly discern if the RCA wire I selected and the USB cable from the ADC back to the computer's hard drive used to record File B audibly compromise the sound. So far according to all posted results they don't. They seem to be perfect to the ear as well:

File B is: D ----->$7.99 D-to-A------------>analog RCA wire------------------------>A-to-D-------->USB cable back to my laptop's hard drive


The 6ft Monoprice 659 RCA wire is currently $1.05 [I bought them when on sale for 87 cents each]. It was brand new hence "not burned in".

The USB cable was free (it came with the ADC).

So if free USB cables and dirt cheap RCA cables sound perfect, why would we buy anything else?
[The RCA wire has a lifetime warranty by the way and with proper use they last for decades.]



I dunno that USB cable looks pretty fancy to me... you sure the Behringer wasn't free with the cable?
 
  • Like
Reactions: m. zillch

·
Registered
Joined
·
15,411 Posts
Discussion Starter #374
I dunno that USB cable looks pretty fancy to me... you sure the Behringer wasn't free with the cable?
:) The orange tip is from orange duct tape I applied to make it easier for me to ID from all my zillions of other black generic USB cables.

It provides deeper, tighter, bass too. I read it on the interwebs.
 

·
Registered
Joined
·
2,884 Posts
OK, I understand. So I have to ask, what is it that collides? Isn't it packets that collide, not frames? If there's an error or collision, isn't it packets that get re-sent? Obviously, I don't know this stuff anywhere near well enough to explain it, I only threw out a side compliment. When one my PCs talks to a 2nd PC, both on a LAN, aren't they still talking with both TCP and IP? Does anything on our typical home networks talk restricted to only ethernet? What reading I've done, it's always TCP/IP that is what's mentioned when it comes to actually moving the data from point A to point B on a network.
Like others, I don't want to take this much more off topic, but I like to explain things. Thus, I will first say: it's not "always TCP/IP". It may be, for example, UDP. However, it is always Ethernet (on an Ethernet network, lol).

When one PC on your Ethernet LAN talks to another, it has to occur (unless they are directly connected NIC-to-NIC) through an intermediary (e.g., a switch in your router). In older networks, there were also wired hubs, in WiFi networks one can consider every WiFi NIC on the same channel (within range of another) connected to the same "hub". A "hub" creates a "collision domain". A "switch" does not because it presents a "direct connection" between the NIC in your PC and a NIC in the switch, essentially the collision domain is "that wire". However, a hub is like "one big wire" and all NICs on it can "collide" while trying to talk. Of course, in WiFi the "wire" is "air", thus I will use the term "transmission medium" from here on.

So, what "collides" is the physical signal on a "transmission medium". Only one NIC can have "the medium" at any time, right? Think of it like multiple people trying to shout down a pipe all at once. Their voices would get all muddled so you could not understand one from the other. Ethernet's framing allows a NIC to detect someone else is "shouting down the pipe" and alert others when it is about to do so itself. That detection cannot be perfect (there's distance between the two "shouters") and, when it isn't, a "collision" occurs (as detected by an unexpected bump in voltage due to the other "shouter" colliding). When this happens, the NICs need to "restart" because the data is now known "garbage".

A "restart" may mean the NIC back-offs, waits, and tries again. Of course, it may not. If "not" occurs, then the NIC will drop the data and it is up to the underlying protocol (e.g., TCP) to decide to retry. UDP, on the other hand, is not a reliable protocol, so if "not" occurs then UDP just lets the packet drop, at that point it is up to the software sending UDP packets to decide to retry. Certainly there are many other reasons a packet can be dropped and Ethernet cares about none of them.

Clear as mud? :D At this point, I think I've cluttered this as much as I'm willing and probably more than others would've liked. Feel free to take it to PM or start another thread if you want to know more.
 

·
Registered
Joined
·
15,411 Posts
Discussion Starter #377 (Edited)
Nice. I'm going to play around, with this challenge.
Have you had a chance to take the test yet? So far no posted results from me nor anyone else show anyone can distinguish any difference, but that doesn't mean additional data points from other people with other systems aren't appreciated, even if the results are the same.
 

·
Registered
Joined
·
11,079 Posts
Have you had a chance to take the test yet? So far no posted results from me nor anyone else show anyone can distinguish any difference, but that doesn't mean additional data points from other people with other systems aren't appreciated, even if the results are the same.
Have not had the chance yet, sorry.
 

·
Registered
Joined
·
487 Posts
Great test, fantastic work. I look forward to trying it myself. I just installed foobar, but found the Mac version can't install components so i'll have to try later on my PC.
 

·
Registered
Joined
·
15,411 Posts
Discussion Starter #380
I dunno that USB cable looks pretty fancy to me... you sure the Behringer wasn't free with the cable?
Ha ha. :D On a serious note, I recently bought a Canon printer on sale last black Friday and according to my calculations it made sense to buy extras just to get the free printer ink cartridges each comes with, even though they are low capacity, because it cost less than the ink cartridges (I'll later need) by themselves.

I wouldn't be surprised if some shaving razors work the same way: even though you only need one handle the least expensive way to buy new blades is to buy more "starter kits", when on sale, which come with at least one set of blades.
 
361 - 380 of 620 Posts
Top