Apodizing filter

[Ghoostknight]

tho i have to say, you guys said it before "you cant see results before the end", luckily this ABX tester allows for showing results earlier, so you actually get "feedback" what was right and not, tho you guys like to "randomize" tests as much as possible, which is a big flaw imo

 

[71 dB]

Yay! We’ve all degenerated in a delightful manner.

Degenerating thread about apodizing filters... ...that's why I have regulated my own activity here.

 

[2leftears]

Degenerating thread about apodizing filters... ...that's why I have regulated my own activity here.

Ironically, this thread needs an apodizing filter to remove all the ripple after the first 50 posts or so...

 

[Ghoostknight]

Ironically, this thread needs an apodizing filter to remove all the ripple after the first 50 posts or so...

you actually have to know how ripple sounds, or you cant differentiate it

 

[71 dB]

probably could have picked a better song for this one but whatever
yep only 67% accuracy, but 33 samples... it was basicly my first test i continued to do

audacity mp3 export setting was "insane/320kbit"

(in my book) you guys are full of crap

You did well in this test!

There is only 4 % chance you were just lucky to have 2/3 of the answers correct. I am ready to believe you can tell 320 kbps mp3 and wav apart, which is not impossible for humans, but very few can do. However, even you with your very good ears had 1/3 of the answers wrong which means the difference is extremely small even for you. The question is does the difference matter to you? I'd like to think a difference I can tell correctly 100 % of the time is a real difference that matters. If I get is only 90 % of the time correct it still matters, but less. If I get it correct only 75 % or less of the time, I don't think it matters anymore. The difference is so small I can pretend it doesn't exist. Only when I can tell the differences correctly close to 100 % or 100 % of time do they "matter" and when it is full 100 % the difference matters so much it requires action if possible.

Now, 320kbit mp3 reduces the bit rate of lossless version it was made from dramatically including frequency ranges were human ears is sensitive. That's why it is possible (but very difficult as you yourself demonstrated) to hear some differences. Do the differences matter is another question. From psychological point of view they probably do matter, because human's don't like the feeling that something is inferior to something else, even if it is only theoretical. You can teach yourself to tolerate this feeling and accept everything doesn't need to be the best thing in the World to be good enough. From practical point of view these differences do not matter. When you are on the go in traffic listening to "Smells Like Teen Spirit" you aren't thinking about what the lossy coding did to the sound. You are thinking your daily life things: Problems at work. The need to buy milk. What to give to your girlfriend as the Birthday present. Kane Pixel's newest Backrooms video. Stuff like that. That's when you are deaf to the defiances to the lossy coding.

Tell me I am full of crap. I know what humans can hear and I understand what matters and when.

 

[Ghoostknight]

You did well in this test. There is only 4 % chance you were just lucky to have 2/3 of the answers correct. I am ready to believe you can tell 320 kbps mp3 and wav apart, which is not impossible for humans, but very few can do. However, even you with your very good ears had 1/3 of the answers wrong which means the difference is extremely small even for you. The question is does the difference matter to you? I'd like to think a difference I can tell correctly 100 % of the time is a real difference that matters. If I get is only 90 % of the time correct it still matters, but less. If I get it correct only 75 % or less of the time, I don't think it matters anymore. The difference is so small I can pretend it doesn't exist. Only when I can tell the differences correctly close to 100 % or 100 % of time do they "matter" and when it is full 100 % the difference matters so much it requires action if possible.

Now, 320kbit mp3 reduces the bit rate of lossless version it was made from dramatically including frequency ranges were human ears is sensitive. That's why it is possible (but very difficult as you yourself demonstrated) to hear some differences. Do the differences matter is another question. From psychological point of view they probably do matter, because human's don't like the feeling that something is inferior to something else. From practical point of view these differences do not matter. When you are on the go in traffic listening to "Smells Like Teen Spirit" you aren't thinking about what the lossy coding did to the sound. You are thinking your daily life things: Problems at work. The need to buy milk. What to give to your girlfriend as the Birthday present. Kane Pixels newest Backroom video. Stuff like that. That's when you are deaf to the defiances to the lossy coding.

Tell me I am full of crap. I know what humans can hear and I understand what matters and when.

the thing.... i hope i somewhat proofed i can do blindtests myself, but each time (vs casual subjective listening) i also notice the culprits of DBT

1. as you say, why i only get 70% right? humans are simple no machines, you guys say it each time on subjectivists but expect machine like results on DBT? (specially if we talk barely audible stuff)
if you do casual listening "you get a feeling" of how the overall sound ist, this doesnt happen after 1 song, but if you like listen for half an hour, IF you switch then only once to B (mp3) and continue to listen (dont ask me why exactly... but i say this to you as someone who did both, blind tests and subjective listening) its way more audible...

2. if i would have gotten no feedback by seeing the results "IN THE TEST", it would have been way harder, either you do "training sessions before hand" without seeing results while the test is running or it makes it pretty much "random"..... tho after all another point that "probably" washes some results out

3. ABX software sound quality
using stuff in windows that utilizes the windows resampling algorithm washes results out, because windows resampling is simply subpar...
its probably not the only potential flaw that could rise from doing "proper A/B tests", just think of a switch box or such

4. rapid switch and fourth

i was doing it naturally, it helped, but i also noticed points where i got 1-2 wrong and simply got really unsure after that (eq see 2. "feedback" which you naturally get while subjective listening because you simply know what is running "you can make sense of how it sounds"), imo its pros and cons with rapid back and fourth switching

i guess someone who really wants to do serious DBT tests can find their own "working" system after a while, but if i imagine random people get stuck in a room "here is a system and a A/B switch, listen to it" i see how potential problems could arise

-----

Now you can say im eccentric and what not but for me these small changes definitely stack up, specially if i look at the digital side of things i just see why people say "it sounds digital" because it actually does..... transients are subpar, high frequencys harsh because ime mostly ringing and what not

also using EQ is always a compromise, funny subjective story:
i had my EQ set on FIR filtering, looked up roommodes in REW and used this variable online tone generator ( https://www.szynalski.com/tone-generator/ ) to sweep the whole frequency range and make it "FLAT" by ear
now i switched to IIR filtering " Huh? the frequency response is crooked again ", this was my biggest hint that phaseshift really matters how you perceive frequencys

in the end i had to adjust the db values for the IIR filters, so i have the same frequency response as with FIR

i think using REW room mode equaliser calculations are simply off because of the phaseshift, i got way better results by doing it per ear

----

From practical point of view these differences do not matter. When you are on the go in traffic listening to "Smells Like Teen Spirit" you aren't thinking about what the lossy coding did to the sound. You are thinking your daily life things: Problems at work. The need to buy milk. What to give to your girlfriend as the Birthday present. Kane Pixels newest Backroom video. Stuff like that. That's when you are deaf to the defiances to the lossy coding.

you are right, "in practise" it doesnt matter more often than not

tho what is in practise? for me in practise is sitting on my PC setup in the sweet spot, i also hear high frequency changing quite alot (just wiggle back and forth half an meter and you get an nice sound effect lol) tho now imagine the exact same sound effect but -0.5db reduced..... its audible if you know what to listen for...

tho if i wanna know what is audible i dont add 10 variables that could "smear" what im actually testing, i test the one thing as good as possible to see whether this could smear other things

 

[bigshot]

No need to even read his posts any more.

 

[Ghoostknight]

No need to even read his posts any more.

yea too much real world expierence to handle

 

[bigshot]

Yeah

 

[gregorio]

hmm will do a mp3 vs flac blindtest next

How typical, you haven’t addressed the points of the last test but you just move on to screw up another in an unrelated area!

(in my book) you guys are full of crap

How many times, this is not your (BS) “book” this is the Sound Science forum. Plus, you’re the one who's making up BS and defending it with fallacies but we’re the ones full of crap? How do you arrive at that impressively nonsensical/backwards conclusion?

tho you guys like to "randomize" tests as much as possible, which is a big flaw imo

Sure, doing a test properly, that eliminates bias and actually tests what you claim you’re testing is a “big flaw in your opinion”, perfect logic lol. I’m presuming that’s because any test that doesn’t give you a result that agrees with your made-up BS is flawed?! Why do you bother posting to a science discussion forum when you clearly don’t know what science is or even it’s basic tenets?

you actually have to know how ripple sounds, or you cant differentiate it

Yes, you have to actually know what inaudible ripple sounds like at 22kHz. Which of course is why neither you nor anyone else can differentiate it! Unless of course you’re superhuman, a bot or a different species.

the thing.... i hope i somewhat proofed i can do blindtests myself

You’re joking? You have not proved anything at all but you have strongly indicated the exact opposite of your claim, that you don’t have any idea how valid blind tests are conducted and therefore cannot do them yourself!

1. as you say, why i only get 70% right?

You mean why did you get as much as 70% right? I would have to check your methodology; it could just be luck or various other things, such as not compensating for the loss of volume when converting, using an old codec or one or two other possibilities. Don’t let any of that bother you though, ignore and fail to eliminate all the real possibilities that could account for your result, just assume you’re superhuman and move on to another test you can screw-up enough to demonstrate you’re superhuman!

2. if i would have gotten no feedback by seeing the results "IN THE TEST", it would have been way harder, either you do "training sessions before hand" without seeing results while the test is running or it makes it pretty much "random"..... tho after all another point that "probably" washes some results out

Firstly, “random” is the whole point of a properly run blind test and Secondly, you clearly don’t know how valid blind tests are done. The test subjects are always either experienced in differentiating the specific artefact/s being tested or are given one or more rounds of training.

3. ABX software sound quality
using stuff in windows that utilizes the windows resampling algorithm washes results out, because windows resampling is simply subpar...
its probably not the only potential flaw that could rise from doing "proper A/B tests", just think of a switch box or such

On the other hand, windows makes your ears bigger and therefore you can hear inaudible sounds. See, I can do that as well, just make-up BS without any evidence!

4. rapid switch and fourth
i was doing it naturally, it helped, but i also noticed points where i got 1-2 wrong and simply got really unsure after that (eq see 2. "feedback" which you naturally get while subjective listening because you simply know what is running "you can make sense of how it sounds"), imo its pros and cons with rapid back and fourth switching

But fortunately your opinion is worthless here, what’s acceptable/valued here is the science/facts which apparently is a huge surprise to you despite this being called the “Sound Science” forum, go figure! The actual pros/cons have been well studied and comprehensively demonstrate that fast switching is far better (easier/more accurate) for differentiating small differences and what’s more, we (science) even know why but again, don’t let the science/facts get in the way of your made-up BS. And, as you apparently did not employ the easier/more accurate method of differentiating the differences, that again indicates either just luck or an invalid test!

i imagine random people get stuck in a room "here is a system and a A/B switch, listen to it" i see how potential problems could arise

My god, you’ve managed to say something that actually agrees with the science/facts for a change, are you feeling OK? On the other hand, your whole statement is still effectively a strawman fallacy (so normal programming has resumed) because science also sees “potential problems” with the methodology you’ve suggested, which is why science does NOT employ that methodology! This again indicates you do not know how valid DBTs are performed, jeez!

G

 

[BS5711]

What we get here is the tip of the iceberg, perhaps we should count ourselves lucky.

 

[Ghoostknight]

such as not compensating for the loss of volume when converting,

sure inaudible frequencys but the volume changes, alright m8

 

[Ghoostknight]

im sure if i follow your methadology i end up at 50/50 again, tho not because its "the only truth", but that just comes from my book

 

[bigshot]

You actually read books? Never would have guessed. I would imagine you think books are “full of crap”. I do know that you don’t own a dictionary.

 

[noahyd]

sure inaudible frequencys but the volume changes, alright m8

Isn't he referring to your MP3 320kbps vs. Flac test?