Slow VS Fast roll off and Minimum VS Linear phase
Feb 16, 2017 at 3:14 PM Thread Starter Post #1 of 61

WindowsX

Member of the Trade: Fidelizer Audio
Joined
Apr 27, 2007
Posts
1,962
Likes
364
Which one is your preference and why? I prefer slow roll off with linear phase because I prefer having better impulse response with good coherent sound.
 
Feb 16, 2017 at 5:50 PM Post #2 of 61
  Which one is your preference and why? I prefer slow roll off with linear phase because I prefer having better impulse response with good coherent sound.

What is this, a troll question?
 
Each has a point depending on the application.  Better impulse response does little good of aliasing results....etc....etc....blah...blah...not to mention the degree of rolloff/phase response...
 
There is no one answer.  Your "preference" is conditional, and you aren't even stating the conditions.
 
Feb 16, 2017 at 10:56 PM Post #3 of 61
where? at the recording? at the mastering? when converting stuff? in my EQ in the audible range? when resampling? in my DAC(starts in the audible range or not)?
and what magnitude for which resolution?
for DACs, I notice something sometimes at low res but most of the time it's the treble roll off. I have no idea if what I and other people feel is part aliasing aliasing, part reduced ringing, or really just good old FR change in the trebles that makes people say silly stuff like "more analog sound".  then if the band limiting starts at high enough frequency, I don't remember that I could tell the difference.
 
for EQ and filters in the audible range, the differences can be subtle depending on the frequency, and at some point if it messes with phase enough, can change the virtual position of instruments a little. in practice I tend to still go with basic IIR because CPU
biggrin.gif
, and because I don't have to change when I start watching a video on utube to deal with the lag. so to me it's yet another victory of laziness over good sound(subjective or not).
cool.gif
 
 
Feb 16, 2017 at 11:54 PM Post #4 of 61
My choice is a proper FIR steep brickwall filter with enough taps not to have aliasing above the noise floor.  Ringing you say, well yeah it rings in the transition frequency if there is energy there.  That being between 20 khz and 22.05 khz for CD.  Don't think I will hear it.  Perhaps if I used metal dome tweeters (which I don't) and any recordings had such ringing going on at any high level (which they don't).  So put me in the fast roll off school.  Otherwise known as a properly designed filter. 
 
Those dirac pulse plots are a signal that can never occur in an AD to DA system.  Oh, and some of the stuff regarding squarewaves is where bandwidth limiting is confused with ringing.  Bandwidth limited squarewaves are not going to be square. 
 
Feb 17, 2017 at 3:54 AM Post #5 of 61
  My choice is a proper FIR steep brickwall filter with enough taps not to have aliasing above the noise floor.  Ringing you say, well yeah it rings in the transition frequency if there is energy there.  That being between 20 khz and 22.05 khz for CD.  Don't think I will hear it.  Perhaps if I used metal dome tweeters (which I don't) and any recordings had such ringing going on at any high level (which they don't).  So put me in the fast roll off school.  Otherwise known as a properly designed filter. 
 
Those dirac pulse plots are a signal that can never occur in an AD to DA system.  Oh, and some of the stuff regarding squarewaves is where bandwidth limiting is confused with ringing.  Bandwidth limited squarewaves are not going to be square. 

Thanks for your opinion.  I still say it was an odd question.  After all, how many of us actually have much of a choice?  Even if we could choose our DAC reconstruction filter, most of the filters in the total audio path are already in the signal chain on the recording/production side.  Whatever they are, and we'll never know, we're stuck with them.  If you have the capability to tweak the final DAC reconstruction filter, great, but frankly, it's one little step in what could be quite a few unknow, cascaded filters. You've lost any sort of linear phase response a few steps ahead of where you are.  
 
The shape of a bandwidth limited square wave and the kind of composite ringing you'd see through the entire chain would look completely different. 
 
Ever seen a square wave come out of a speaker?  
 
Feb 17, 2017 at 4:02 AM Post #6 of 61
Some DAPs have 5-7 digital filters to choose, some DACs have even tens of combinations. Asking something like this is considered a troll question now? It's harder to post in sound science forum nowadays, isn't it?
 
Feb 17, 2017 at 4:32 AM Post #7 of 61
  Some DAPs have 5-7 digital filters to choose, some DACs have even tens of combinations. Asking something like this is considered a troll question now? It's harder to post in sound science forum nowadays, isn't it?

Ok, fine, some may have payed up for those DAPs and DACs, most don't.  The existence of filter choices is simply evidence of good marketing, not good engineering.  Why you'd put that choice in the user's hands is beyond me.   I find it interesting that some of the best DAC designs do not have that feature.  
 
But that doesn't address the core issue: Why bother messing around with different filters at the end of the line when filtering has already been done, possibly several times over, before you get to pick that final filter.   There's no talking about linear phase. You need the one that works  best at its task: reconstruction without aliasing.  And that still doesn't answer the question because we don't know anything about the DAC itself and how the filter is implemented.  
 
So I still don't understand the question.  If you're asking which to choose on one of those DAPs  or DACs with different filter choices, I'd say pick a DAP or DAC with the right filter already chosen. 
 
Feb 17, 2017 at 5:30 AM Post #8 of 61
  Ok, fine, some may have payed up for those DAPs and DACs, most don't.  The existence of filter choices is simply evidence of good marketing, not good engineering.  Why you'd put that choice in the user's hands is beyond me.   I find it interesting that some of the best DAC designs do not have that feature.  
 
But that doesn't address the core issue: Why bother messing around with different filters at the end of the line when filtering has already been done, possibly several times over, before you get to pick that final filter.   There's no talking about linear phase. You need the one that works  best at its task: reconstruction without aliasing.  And that still doesn't answer the question because we don't know anything about the DAC itself and how the filter is implemented.  
 
So I still don't understand the question.  If you're asking which to choose on one of those DAPs  or DACs with different filter choices, I'd say pick a DAP or DAC with the right filter already chosen. 

 
I asked this questions because MOST DAPS in market today has those digital filter options (Fiio/iBasso/Cayin/etc.). Why? I don't know. I'm just curious what people would pick for scientifically. If you may enlighten me, please show good DAPs that sell today without any digital filter option provided.
 
Regards,
Keetakawee
 
Feb 17, 2017 at 12:21 PM Post #9 of 61
   
I asked this questions because MOST DAPS in market today has those digital filter options (Fiio/iBasso/Cayin/etc.).

Most DAPs in the market? Odd perspective. How about looking at the most DAPS sold in the market? And that would be none of those.
Why? I don't know.

Sure you do. Are you not in a similar market as those DAPS you mention? Is that a big market or a tiny splinter of the total DAP market? Marketing 101, you can't compete and be the same, you have to be different. If the core function of your product is also served by other competing products, you have to beat them in features or price or both.  Anybody placing a product in a market must surely realize that or fail completely in the attempt. 
 
To penetrate the market fully and quickly, yours must provide a percieved 10-fold improvement over the existing product with the same core function. None of those DAPs has penetrated the market significantly at all unless you take the entire DAP market and segment it down to the boutique level. And that's why they all have pointless features they can market to elevate the perception of their product and offset the negative effect their price and presence has.
I'm just curious what people would pick for scientifically.

Scientifically? Then don't ask why people choose a filter feature on a DAP or DAC, users aren't scientific at all.   Talk to a real DAC designer. I'd start with Benchmark. (This hits the reason for my confusion with your question, BTW)
If you may enlighten me, please show good DAPs that sell today without any digital filter option provided.  

I think you can figure that one out for yourself (without baiting me). Though we probably don't agree on what "good" is.  
 
Feb 18, 2017 at 3:07 AM Post #10 of 61
You made a claim that one should choose DAPs/DACs with right filter so I asked you which ones because you suggested me yourself to find the right ones.
 
Quoting every line I wrote like this is ridiculous. You're way too paranoid. Maybe you should do something with your audio obsession. Seek help.
 
Feb 18, 2017 at 4:24 AM Post #12 of 61
  You made a claim that one should choose DAPs/DACs with right filter so I asked you which ones because you suggested me yourself to find the right ones.

Read my post again, that's not what I said.
Quoting every line I wrote like this is ridiculous.

It keeps my reply in context.  Other people do it by numbering statements, I've always preferred this format as it's less confusing.  At least, for most people.
You're way too paranoid.

Would that be a personal attack?? Hmmm....
Maybe you should do something with your audio obsession. Seek help.  

Seriously?  
 
Feb 18, 2017 at 5:21 AM Post #13 of 61
Most DAPs in the market? Odd perspective. How about looking at the most DAPS sold in the market? And that would be none of those.

 
That's what you said and I looked into here.
 
http://www.head-fi.org/f/15/portable-source-gear
 
I'm pretty sure at least 50% of DAPs in that page has digital filter options in one way or another. I don't know how you came up with conclusion that most DAPs sold today don't have digital filter options.
 
Feb 18, 2017 at 8:50 AM Post #14 of 61
http://www.head-fi.org/t/786112/official-x5-2nd-gen-beta-firmware-download-and-discussion-thread-version-1-27-library-search-function-replaygain-filter-options/60#post_12311847

I was authorized at the time to give such a reply to this question regarding fast vs slow rolloff filters on FiiO players.

To further elaborate, the cutoff frequency (22.05kHz or what-have-you) should be purely ultrasonic and inaudible in the first place, otherwise the choice of filter is simply putting you between a rock and a hard place: either cut off sharply at higher frequencies (closer to inaudible ultrasonic frequencies) but with ringing, or cut off a whole swathe of high but audible frequencies without ringing but with noticeably duller sound. 22.05kHz should be just out of the audible range of frequencies for even the youngest subjects who know how to play a CD or audio clip for themselves, so I would choose a sharp rolloff filter (to avoid affecting audible frequencies below at no audible cost of ringing, because it happens to not be audible).

Any resemblance between slow rolloff filter and room acoustic absorption mentioned in the link is completely coincidental; a purpose-designed EQ filter would technically be a much better idea.

Choice of minimum vs linear phase filter is only meaningful in the case of an audible cutoff frequency, in which case splitting the difference between minimum and linear phase ("25% phase" or so) would give the best masking of any ringing produced. If the cutoff frequency range is inaudible to you, it is a moot point.

So here you have it, an answer straight from the horse's mouth of one of the brands you mentioned. Happy?
 
HiBy Stay updated on HiBy at their facebook, website or email (icons below). Stay updated on HiBy at their sponsor profile on Head-Fi.
 
https://www.facebook.com/hibycom https://store.hiby.com/ service@hiby.com
Feb 18, 2017 at 2:12 PM Post #15 of 61
Most DAPs in the market? Odd perspective. How about looking at the most DAPS sold in the market? And that would be none of those.

 
That's what you said and I looked into here.
 
http://www.head-fi.org/f/15/portable-source-gear
 
I'm pretty sure at least 50% of DAPs in that page has digital filter options in one way or another. I don't know how you came up with conclusion that most DAPs sold today don't have digital filter options.

STILL not reading my posts carefully!

Sorry, I can replay all I want, it if you're not comprehending my posts there's really no point in continuing.
 

Users who are viewing this thread

Back
Top