Apodizing filter

Feb 21, 2024 at 12:52 PM Post #46 of 426
how do you determine which one sounds best if they all sound the same? I think it’s better to just say, don’t use NOS DACs, because they aren’t audibly transparent.
The filters implemented by Meridian in the V50 made me perceive a difference. I preferred the slow filter because I perceived the least distortion from that, so I stuck to that one. Was there an actual difference? I don't know, and I don't care because the other filters worsened my experience.
 
Feb 21, 2024 at 12:55 PM Post #47 of 426
What’s the point of adding options that degrade sound?
 
Feb 21, 2024 at 1:00 PM Post #48 of 426
What’s the point of adding options that degrade sound?
I don't know, I'd speculate that some people prefer more residual sound because it sounds more analog, kind of like how some people still like vinyl or single DD IEMs for whatever reason. I want the highest level of precision I can get, so I went for the filter that played synthetic sounds back as cleanly as I could audibly discern. It was very subtle, so best I can describe it was a feeling that something's wrong vs hearing a distinctive difference.
 
Feb 21, 2024 at 1:02 PM Post #49 of 426
I think they’re just adding options as “features” to give audiophile one more setting to apply expectation bias to.
 
Feb 21, 2024 at 1:25 PM Post #51 of 426
I had an SACD player that had a button labeled “pure sound”. The instruction book just said, “Push this button for pure sound.” When you pushed it, there was absolutely no difference in sound, but a lovely purple LED would light up. It really was a lovely shade of purple.
 
Feb 21, 2024 at 2:28 PM Post #52 of 426
You certainly can if the problems are all down below -50dB.
Not really no, -50dB seems an arbitrary (and high) level to set as a generalized audibility threshold
 
Feb 21, 2024 at 5:31 PM Post #53 of 426
In my sig file is a seminar by Ethan Winer. In it, he takes a horrible buzzing sound, the worst kind of noise you can imagine, and he plays the noise under music, lowering it 10dB at a time. At -40dB it is inaudible. He links to WAV files so you can hear it for yourself. The decibel scale is exponential, so the first 10dB is much smaller of a difference than the last. -50dB is WAY down low. LP records don’t generally have much music down that far, mostly noise.

People in audiophile groups throw around numbers with no clue what those represent in real world sound. A signal to noise of -50dB is good. It can be a little better if you plan to play your music very very loud and want stone silence in the lead in and lead out. But -50dB is plenty for a pleasurable listening experience, and -70dB is well into overkill for normal music listening.

See the Audio Myths seminar below. Test it for yourself.
 
Last edited:
Feb 21, 2024 at 5:34 PM Post #54 of 426
In my sig file is a seminar by Ethan Winer. In it, he takes a horrible buzzing sound, the worst kind of noise you can imagine, and he plays the noise under music, lowering it 10dB at a time. At -40dB it is inaudible. He links to WAV files so you can hear it for yourself. The decibel scale is exponential, so the first 10dB is much smaller of a difference than the last. -50dB is WAY down low. LP records don’t generally have music down that far, only noise.

People in audiophile groups throw around numbers with no clue what those represent in real world sound. A signal to noise of -50dB is good. It can be better if you plan to play your music very very loud and want stone silence in the lead in and lead out. But -50dB is plenty for a pleasurable listening experience.

See the Audio Myths seminar below.
I've been able to ABX differences far lower than that, as have others. But also in regards to the file you mentioned, keep in mind a constant, otherwise uncorrelated signal component/difference is often less audible than a deterministic or signal correlated one.

screencapture-klippel-de-listeningtest-2020-10-28-21_57_32.png
 
Last edited:
Feb 21, 2024 at 5:38 PM Post #55 of 426
If you have done a controlled test under music with no gain riding on the fade outs (Amir), then I salute you. But I don’t believe you’ve done that. I don’t listen to test tones on my couch with a glass of Merlot. I listen to music.
 
Last edited:
Feb 21, 2024 at 5:49 PM Post #56 of 426
If you have done a controlled test under music with no gain riding on the fade outs (Amir), then I salute you. But I don’t believe you’ve done that. I don’t listen to test tones on my couch with a glass of Merlot. I listen to music.
I've done the same with music too.

In fact I'm working on a product which has an option to change sound with some playing about with harmonic distortion and all changes are below -60dB.
Actually if -50dB was the limitation I don't think there would be many devices available at all that WOULDN'T meet that requirement.

The debate about what level is 'sufficient' or enough to enjoy music is separate from that of what is/isn't audible.
And if something is demonstrated to be audible in a 'synthetic' test for example, then whether it's still audible with typical music simply depends on the content of the music, and makes the discussion of audibility threshold essentially impossible to answer.
 
Feb 21, 2024 at 5:52 PM Post #57 of 426
We are using our equipment to listen to music. Audibility in normal use is completely relevant. In fact, it’s all that really matters. Audiophools chase sound they’ll never hear in the real world down rabbit holes. The standard filters used on oversampling DACs are inaudible in normal use. It’s not worth worrying about.
 
Feb 21, 2024 at 5:56 PM Post #58 of 426
We are using our equipment to listen to music. Audibility in normal use is completely relevant. In fact, it’s all that really matters. Audiophools chase sound they’ll never hear in the real world down rabbit holes. The standard filters used on oversampling DACs are inaudible in normal use. It’s not worth worrying about.
I think we're using the term 'audibility' different though.

I'm meaning that something is audibly discernible. Whether that's to a significant enough extent that it impacts the enjoyment of the listening experience is a different matter and up to the listener.

For me the reconstruction filter does make enough of a difference that I prefer to listen with a better one. For some it may not be significant enough to justify the inconvenience of external upsampling, and for others it may be inaudible outright.
 
Last edited:
Feb 21, 2024 at 7:16 PM Post #59 of 426
Discernibility is relative. If I use a microscope, I can see a paramecium walking across the glass slide. If I'm out walking my dog, I can't see a paramecium walking down the street.

Context matters. We're talking about home audio equipment designed to listen to music in the home. You have to jury rig a complex set of test parameters involving waveforms and illegal signals to be able to come even close to discerning this stuff. I'm quite sure that in your living room listening to music, none of this makes a lick of difference to you or any other owner of all too human ears.

But I don't even think we're talking about audibility here. We're defining the threshold between music listening and equipment fetish. Vanishingly small numbers might mean something if your interest is in theoretical science and collecting electronic equipment that pushes the edges of technology. But they're meaningless if you want to sit on your couch and listen to Mozart with the best sound your ears can hear.

I'm in the latter camp, and it amazes me that people are interested in chasing down meaningless bugaboos like jitter and obscure DAC filters. I see much more important things to deal with, like frequency response imbalances, which are much more common and are orders of magnitude greater than the tiny stuff the gear heads focus on. It's even stranger to me when gear heads refuse to use equalizers because of phase shift, when the phase shift is so insignificant and the response imbalances are so massive.

I guess those elephants are safely hiding in the corner unnoticed.
 
Last edited:

Users who are viewing this thread

Back
Top