Minimum Phase vs Linear Phase

[Mnyb]

Is not low latency filters useful in studio speakers ? Have not some DSP based systems some low latency modes ?

 

[Julf]

Is not low latency filters useful in studio speakers?

No. Stage monitors yes, but they don't have to be high quality.

 

[Keith_W]

Is not low latency filters useful in studio speakers ? Have not some DSP based systems some low latency modes ?

Firstly there is always some kind of latency involved with all digital devices because of buffers. My RME interface, which does zero DSP, has a measurable latency.

If it's DSP latency, it depends on the type of DSP. Minimum-phase FIR or IIR has the lowest latency of all. Linear phase FIR latency depends on the number of taps and the sample rate. But there's nothing to stop you from configuring your FIR to do minphase filtering instead of linphase. So ... it depends

 

[Jazigo]

I decided to do some measurements on my Topping E50 DAC. It has 3 filter, F1 (Linear), F2 (NOS(?)) and F3 (Minimum).

I use a tool (jack_iodelay) for Linux to measure the round-trip delay (Computer --[usb]--> E50 --[RCA/jack]--> Computer).

Preamp	(in millisecond)
	F1: 84.4
	F2: 83.75
	F3: 84.0
DAC
	F1: 84.6
	F2: 83.85
	F3: 84.0

These are average values and repeatable with +/- 0.1ms.

In other words: there is only a 0.4 - 0.6ms delay between Linear- and minimum phase in the context of a reconstruction filter for a DAC.

 

[MC_RME]

The filter timings are precisely (in samples) documented in the DACs chip data sheet. The timings are therefore the same for anyone using these chips.

 

[Hindsider85]

Minimum Phase is the ONLY correct choice for a natural and analog sound.

 

[tmtomh]

Minimum Phase is the ONLY correct choice for a natural and analog sound.

What is your evidence and support for this assertion?

 

[Julf]

Minimum Phase is the ONLY correct choice for a natural and analog sound.

If you say so...

 

[Hindsider85]

What is your evidence and support for this assertion?

If you say so...

I do say so and do HEAR so.
And there is a long standing and established consensus about that.
Even measurements, about audible pre-ringing.

 

[Julf]

I do say so and do HEAR so.
And there is a long standing and established consensus about that.
Even measurements, about audible pre-ringing.

If you say so...

 

[voodooless]

Even measurements, about audible pre-ringing.

Audiophile Myth #260: The Detestable Digital Filter Ringing and Real Music...

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

 

[MRC01]

I do say so and do HEAR so.

OK, everyone is entitled to his opinion.

And there is a long standing and established consensus about that.

This means other people agree with that opinion, and that is true. Yet there is also a long standing and established consensus behind the fact that linear phase filters sound closer to the original waveform. This should be no surprise, since different people have different opinions and it's been debated for years.

Even measurements, about audible pre-ringing.

Now you're stepping into the realm of objective measurements. The mathematically ideal reconstruction comes from the Shannon-Whittaker reconstruction formula. If the original signal was properly bandwidth limited before encoding (as it should be), it provides perfect reconstruction. It happens to be linear phase, an infinite sum going both forward and backward in time for every sampling point. This makes it computationally infeasible to implement in near-real time. Yet since it represents ideal/perfect reconstruction, it's the waveform that all well engineered DACs get very close to through other more efficient methods like Delta-Sigma.

It's worth mention that pre-ringing is not an exclusively digital artifact. It's called the Gibbs Phenomena and also happens in the analog domain as a natural result of bandwidth limiting. Here's an example intro, and you can skip ahead to Gibbs.

 

[Hindsider85]

I use a NOS R2R DAC with Audirvana and r8Brain and DSD 512 oversampling. And linear vs minimum does make a very audible difference to me.
These are my settings and I needed a couple days of tuning, to finally get the fidelity and richness I always sought after.

 

[oleg87]

I use a NOS R2R DAC with Audirvana and r8Brain and DSD 512 oversampling. And linear vs minimum does make a very audible difference to me.

I suppose I need not ask whether any controls were used to determine audibility

 

[Hindsider85]

I suppose I need not ask whether any controls were used to determine audibility

Critical listening is all I need.

 

[MRC01]

Even measurements, about audible pre-ringing.

If the above wasn't clear, here's an intuitive way to think about those ripples. They are part of the original waveform, they are always there whenever the wave contains high frequencies. And they will be in the reconstructed wave too. Linear or minimum phase filter doesn't matter - they are there either way.

What minimum phase filters do is add a progressive time delay to high frequencies - they shift them later in time. Because they've been moved, not eliminated, there is no pre-ripple and the post-ripple is bigger and longer. This is a distortion of the original wave that was encoded, because that is not when these frequencies actually occurred.

 

[oleg87]

Critical listening is all I need.

All you need to post nonsense on ASR perhaps, but not enough to determine audibility.

 

[Mort]

Critical listening is all I need.

Golden ear is not a compliment.

 

[LTig]

Critical listening is all I need.

Uh huh.

 

[Julf]

I use a NOS R2R DAC with Audirvana and r8Brain and DSD 512 oversampling. And linear vs minimum does make a very audible difference to me.
These are my settings and I needed a couple days of tuning, to finally get the fidelity and richness I always sought after.

Critical listening is all I need.

Maybe try some critical thinking at some point, instead of believing everything you read on audiophile forums?