• 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!

Minimum Phase vs Linear Phase

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
9,596
Likes
12,036
Friendly suggestion - please use a single font size and avoid unnecessary bold or italic emphasis, it makes my head hurt and basically keeps me from wanting to read your posts.
Well it also makes them instantly recognisable :rolleyes:
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
I use the enriched edition since the end of the nineties, knowing the laziness and intellectual disability of the Spanish readers of that time. Today it is much worse than before and has become widespread in the West.

Maybe the change in size affects your viewing in the phones? I will try to drastically reduce its use, focusing as before on bold, italic and colors.
 
Last edited:

tmtomh

Major Contributor
Forum Donor
Joined
Aug 14, 2018
Messages
2,636
Likes
7,495
Friendly suggestion - please use a single font size and avoid unnecessary bold or italic emphasis, it makes my head hurt and basically keeps me from wanting to read your posts.

With acoustic and minimally processed electronic music, MULTIPLE font sizes and styles clearly sound better. With processed or autotuned music, the font size and style is a matter of personal preference.
 
Last edited:
D

Deleted member 2944

Guest
I use the enriched edition since the end of the nineties, knowing the laziness and intellectual disability of the Spanish readers of that time. Today it is much worse than before and has become widespread in the West.

Maybe the change in size affects your viewing in the phones? I will try to drastically reduce its use, focusing as before on bold, italic and colors.
Why don't you just try not focusing on bold, italic, and colors and just attempt a coherent message?
I think that might get you off a lot of members "ignore list."

Dave.
 

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
What is the rich edition allowed on the forums for?

I understand that resizing may affect reading on small screens but not the rest.
 

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
Perhaps a couple of new threads could keep this one on track:

- Fun with fonts
- To compress or not to compress, that is the question

- Rich
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
What is the rich edition allowed on the forums for?

I understand that resizing may affect reading on small screens but not the rest.

1567524398454.png
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Abuot raedibaltiy: I awlyas touhgt tihs was fnuny:

It deosn't mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.
 
Last edited:

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
...
I haven't done any blind testing, but there appears to be a certain harshness in loud and high notes of some female vocalists which goes away when the DAC is set for minimum phase. ...
I would guess that the problem is in the recording, not in the DAC. It's common for recordings to add artificial presence in the upper mids and treble that is most audible with female vocals and sibilants. Large diaphragm condenser mics common used to record vocals can have a rise somewhere in the 8k to 16k range.
A linear phase DAC filter usually has ruler flat response that plays this as it was recorded and it sounds too bright or harsh.
A minimum phase DAC filter usually does not have ruler flat response but starts rolling off in the audible spectrum which can attenuate this presence, making it sound softer or more natural.

If this is the case, blame the recording engineers, not the DAC. I may be stoned by audio purists for admitting this, but when I encounter recordings having this excess treble & sibilance (which is all too often), I raise my eyes to the heavens and silently curse the recording engineers, then use DSP tone controls to make it more listenable.

...
My understanding is linear phase filters have pre-ringing. Minimum filters put all the ringing after the pulse. ...
Yes, but a properly implemented linear phase filter rings at the Nyquist frequency, so it should be inaudible, especially since most DACs oversample. Minimum phase filters have all the ringing after the transient, but they ring louder and longer, and have phase distortion in the audible spectrum (in the passband). They also tend to have a more gradual slope, so frequencies in the top audible octave may be attenuated, which can be audible to some people, with some recordings.
 
Last edited:

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
BTW
The pre-ring, brought by FIR (finite impulse response) digital filters used in traditional digital to analog conversion, can be tested and heard very easily.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
That quoted statement is demonstrably false in at least one aspect. The mathematically ideal filter is defined by the Whittaker-Shannon reconstruction formula; the classic sinc(t) function. This filter's pre-ring is at the Nyquist frequency, which is supersonic (thus inaudible) especially when over-sampled which all well-engineered modern DACs do. Even though most modern DACs don't use this reconstruction formula, they use alternatives like Delta-Sigma which produce outputs virtually identical to this ideal response. In short, the pre-ring or ripple that is inherent to digital audio is inaudible.

However, there are many different digital filters and DA reconstruction algorithms. Some DACs provide several different implementations on a single chip, some of which fall far from the sinc(t) ideal and some of which can sound different from each other. So that quoted statement may true from this perspective.

PS: you can see in the above graphs, the minimum phase filter rings louder and longer than linear phase. That's the price paid to push all the ringing after the impulse. But the graphs are not to scale, which is misleading. The first impulse peaks about 4.5 units above the noise, the second about 6 units above the noise. This obscures the fact that the max ripple amplitude in the first is about 2 units, which is 2/4.5 = 44% of the impulse amplitude. The max ripple amplitude in the second is about 1 unit, which is 1/6 = 17% of the impulse amplitude.
 
Last edited:

maty

Major Contributor
Joined
Dec 12, 2017
Messages
4,596
Likes
3,160
Location
Tarragona (Spain)
Regardless of the two filter used, its differentiation must be very easy as I said. Those who do not appreciate the difference should think they have a bottleneck in their audio chain.

At the end, hard or soft, the important thing is the correct implementation and / or programming.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
Regardless of the two filter used, its differentiation must be very easy as I said. Those who do not appreciate the difference should think they have a bottleneck in their audio chain. ...
I don't doubt that some digital filters may be easy to differentiate. My point is that others are not easy.

If anyone finds 2 digital filters to be easy to differentiate, I suspect that either one of both of them are far from ideal performance. You're probably hearing frequency aliasing, phase distortion or other artifacts that less than ideal filters have.

Conversely, if you pick 2 filters each having close to ideal sinc(t) performance, they will be difficult if not impossible to differentiate.

Also: in cases where one differentiates minimum from linear phase filter in a blind test, it doesn't necessarily mean the difference he's hearing is pre-ring (or lack thereof). He's almost certainly hearing something else, like phase distortion, frequency aliasing, or passband frequency attenuation. Why? Simple: the pre-ring of a standard linear phase filter (as in any good delta-sigma DAC chip) ripples at the Nyquist frequency which is above human hearing. So sure, some filters can sound different -- but not for the reasons most people think.

... At the end, hard or soft, the important thing is the correct implementation and / or programming.
Agreed.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
Also: in cases where one differentiates minimum from linear phase filter in a blind test, it doesn't necessarily mean the difference he's hearing is pre-ring (or lack thereof). He's almost certainly hearing something else, like phase distortion, frequency aliasing, or passband frequency attenuation. Why? Simple: the pre-ring of a standard linear phase filter (as in any good delta-sigma DAC chip) ripples at the Nyquist frequency which is above human hearing. So sure, some filters can sound different -- but not for the reasons most people think..

We also know the MQA uses a proprietary slow roll-off filter. I hypothesize that difference heard (if not imagined) are either sourced from a different master or something like "like phase distortion, frequency aliasing, or passband frequency attenuation" precisely because the added frequencies are above human hearing. Occam's Razor at work again. :p

- Rich
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
Speaking of Occam's Razor, Keith Howard wrote a good one for Stereophile a few years ago: https://www.stereophile.com/reference/106ringing
I love their experimental attitude: test and discover! And their candor discussing their results. But when they talk about how hard it was to tell the filters apart, it is kinda funny thinking about a bunch of middle-age guys wondering why they can’t hear a supersonic ripple octaves above the range of their hearing. Especially when most of them understand math & engineering well enough to know why.
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,760
Likes
9,442
Location
Europe
It is actually not at all clear playing a WAV file needs less CPU resources. With modern CPU architectures, the work of moving the extra, redundant data in and out can cause as much load as the actual processing of something like FLAC.

I like to disagree. The amount of data moved out is the same for both formats (PCM), just the amount to move in is different - with just half of it for FLAC compared to WAV. I don't think that a data rate reduction of 50% compensates the heavy processing of FLAC against almost no processing in case of WAV which contains the PCM data more or less ready for transfer to the DAC. It may well be possible that the CPU sleeps most of the time when handling WAV files.

Just curious - in what way is WAV " the reference for perfect SQ"?
Because the WAV file has been used to create the FLAC file. At least this is how I create FLAC files.

Anyway in my understanding there is no difference in SQ between WAV and any FLAC regarding the PCM data stored within. However if the FLAC algorithm has a bug then the WAV file is the reference and may sound better. This can easily tested by recreating a WAV from FLAC and a bitwise comparison of this with the original WAV.
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,760
Likes
9,442
Location
Europe
Regardless of the two filter used, its differentiation must be very easy as I said. Those who do not appreciate the difference should think they have a bottleneck in their audio chain.
Or they listen blind ...
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,760
Likes
9,442
Location
Europe
Abuot raedibaltiy: I awlyas touhgt tihs was fnuny:

It deosn't mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.
Yeah, although I'm not a native english speaker I could read this almost as fast as if it had not been scrambled.:cool:

It may be related how I learned to read. When I came into school there was a new method to teach reading: instead of introducing one letter after the other the pupils learned to read whole words. After a few years the method was sacked because the result was that 10% of the pupils learned to read very fast:) while the other 90% had difficulties to read proper at all:confused:. Luckily I belonged to the 10%...:D
 
Top Bottom