Does Phase Distortion/Shift Matter in Audio? (no*)

[j_j]

Well, on this one I'm sticking with my constant-delay crossovers and continuous phase response.

 

[gnarly]

There is no interest from the industry (and thus AES etc) to show audibility of phase distortion.

But we are in the lucky position to create dependable evidence ourselves with quite little effort, relatively speaking:

• Take an existing decent setup (good speakers, good room with proper balance of absorption and diffusion, of course acoustical symmetry)
• Apply professional DRC (like Acourate etc) to it at the LP, to create flat (to target) mag response and min-phase'ish phase response
• Make long-term AB comparisons (and then ABX tests once the learning curve has settled), with introduced global phase distortion like what regular speaker XO would introduce (same on both channels, of course), listening for a number of effects like timbre change in the bass, speed of bass transients, soundstage aspects like perceived size and sharpness of phantom sources), etc etc
• Draw your conclusions

Yep. I basically do the process you outline above. Differences being I tune multi-ways from the ground up, on a driver by driver basis, to be either linear-phase or minimum phase. So there's no phase correcting overlay needed (like from Acourate etc).
My preferred short term listening test bed is mono outdoors to eliminate reflections as best as possible. Long tern indoors for stereo. It takes quite a while to form an opinion sometimes...and comes down to how often am I flat out delighted, and how often does something sound a bit off.
The processor I use (qsys) allows instantaneous preset switching between lin or min phase. I compare the same order lin vs min when doing this. Can also set presets to change orders on the fly too.
I test so many filter combos, and since the multis are 5-ways, it's easy to screw up occasionally. It's kind of remarkable how when the sense that somethings not quite right pops up, measurements usually pinpoint the problem immediately. Gives hope to making valid conclusions

 

[gnarly]

If you do care about phase more, then conduct your own formal experiments and come back with that data. Otherwise disagreeing for the sake of what one feels in their belly is not going to get us anywhere.

Sure thing, don't mean to hog the floor.

 

[totti1965]

Nothing to complain at audibly level by my needs and capacities, but numbers are what they are.

123 SINAD of Topping D90 III is better than 113 WiiM Pro Plus, what is audibly or not I don't know.

I found diverse answers in ASR, some of them arguing imperceptible over 120 SINAD and other enough over 110.

Yes, I am more in the latter team. From my personal experience with the WiiM an the SMSL D6s (a Friend kindly lent it to me), I can say that there is (via Roon) at least an objective tiny tiny difference between direct use of the WiiM versus WiiM digital out to SMSL D6s
When the next song plays, there’s a tiny tiny barely audible „klick“ bevor the Song Starts. Nothing to make a big thing about. But with using only the AKM Chip inside the WiiM, the tiny tiny „klick“ was not there. It was flawless.
This was the objective difference!

The digital signal has so much shorter Signal Path to go, if it stays inside the WiiM, so my brain tells me on the „subjective“ side also that the sound was more musical and more natural and less harsh by just enjoying the WiiM Pro Plus.

In my case the 113 dB topped the 120 dB. But who knows, perhaps it was a synchronization Problem with the SMSL D6s and that causes the tiny sound degradation also…….

 

[totti1965]

It's the sole SINAD number that does not tell the story I want to point out.
For example this popular amp has a 70dB-80dB SINAD at 5kHz but a good SNR. It won't be a problem at all in its good operational range and no one is reporting such.

(I admit I always tend to think speakers, so... )

Wrong link? The 3e has 100 dB SINAD.

 

[Miguelón]

Yes, I am more in the latter team. From my personal experience with the WiiM an the SMSL D6s (a Friend kindly lent it to me), I can say that there is (via Roon) at least an objective tiny tiny difference between direct use of the WiiM versus WiiM digital out to SMSL D6s
When the next song plays, there’s a tiny tiny barely audible „klick“ bevor the Song Starts. Nothing to make a big thing about. But with using only the AKM Chip inside the WiiM, the tiny tiny „klick“ was not there. It was flawless.
This was the objective difference!

The digital signal has so much shorter Signal Path to go, if it stays inside the WiiM, so my brain tells me on the „subjective“ side also that the sound was more musical and more natural and less harsh by just enjoying the WiiM Pro Plus.

In my case the 113 dB topped the 120 dB. But who knows, perhaps it was a synchronization Problem with the SMSL D6s and that causes the tiny sound degradation also…….

When I wrote the last paragraph I didn't calculate how much SNR I needed in my configuration, now I realized that 80 dB SNR should be enough even in the most exigent conditions.

To listen any noise at 110 SNR you should listen at 140 dB which is very unlikely (and dangerous).

Personally I'm very happy since I abandoned my DACs obsession and forget those attentive listenings to detect fails and sound quality: if one acts like that is very easy to really find imaginary issues.

Even at room interaction level, when things get audible, one should admit minor flaws or if not we never enjoy music enough.

 

[amirm]

To listen any noise at 110 SNR you should listen at 140 dB which is very unlikely (and dangerous).

??? If there is no music playing, then -110 dB between 2 and 5 kHz would be audible by itself.

 

[Sokel]

Wrong link? The 3e has 100 dB SINAD.

You missed the frequency in the post
Look at the Frequency vs distortion chart.

 

[Sokel]

??? If there is no music playing, then -110 dB between 2 and 5 kHz would be audible by itself.

-110dB on it's own?
Or through the chain gain+recording?
Cause this way it would be not -110dB on the end.

 

[amirm]

However it gets there.

 

[Newman]

??? If there is no music playing, then -110 dB between 2 and 5 kHz would be audible by itself.

When played at what level in SPL at the ear, Amir?

 

[bmc0]

I think not so much, at least my setup peaks at 110 dB @ 1m and my room is not one of the most silent, let’s be optimist and put 35 dB ambient.

I’m listening at 2 meters, so can expect reasonable 105 dB peaks, that means I would be well deserved with the mentioned 70 dB SNR.

Am I right?

You have to take into account the spectral content of typical room noise—It's mostly low frequency rumble. I've mentioned before that I can hear a 2kHz-5kHz sine chirp down to approx. -4dBSPL in my listening room. That was with my head turned a bit to maximize ear gain in that range.

 

[Miguelón]

??? If there is no music playing, then -110 dB between 2 and 5 kHz would be audible by itself.

I supposed a 30 dB ambient noise and 140 dB listening levels to justify a situation in which 110 dB SNR will be necessary, in the case if the rest of the chain doesn’t impose their own noise.

In well isolated headphones maybe I can listen -110 dB, but I’m pretty sure that for my setup (max short term 105 dB at my listening distance, pair of speakers) is enough to have an 80 dB SNR DAC because my room is quite noisy.

 

[amirm]

I supposed a 30 dB ambient noise and 140 dB listening levels to justify a situation in which 110 dB SNR will be necessary, in the case if the rest of the chain doesn’t impose their own noise.

Neither is appropriate. Your noise floor varies with frequency and there is no home gear that can produce 140 dB of dynamic range.

 

[j_j]

Before you state an SPL noise level, please measure it by 1/3 octaves. That's what really matters, give or take.

 

[Mike123]

In well isolated headphones maybe I can listen -110 dB, but I’m pretty sure that for my setup (max short term 105 dB at my listening distance) is enough to have an 80 dB SNR DAC because my room is quite noisy.

Of course not. The DAC noise is amplified by the amplifier. That's why many people use resistor-based volume controls (variable resistors or relays and fixed resistors), which reduce the DAC noise. -80 dB noise is easily heard by many people from a speaker system at a distance of two meters. Even I, an elderly person with tinnitus, can hear -75 dB. That's why I wouldn't even consider buying a DAC with noise worse than -100 dB. And preferably no worse than -110 dB. For modern DACs, this is not a problem at all, including ultra-budget ones.

 

[Miguelón]

Neither is appropriate. Your noise floor varies with frequency and there is no home gear that can produce 140 dB of dynamic range.

Was a hypothesis.

I’m still confused: we were discussing how much SINAD will be enough on a DAC to be transparent for average healthy population (suppose them non trained).

I received some answers saying “70 dB” although a random search on the ASR browser gave 110 to 120 dB range.

Was just a curiosity originated by a discussion with other member, my WiiM Ultra is around 120 dB and my audio interface one of the worse I had, around 109 but I cannot tell the difference between them.

Are 120 dB a good point? Why those answers stating 70 dB?

 

[Miguelón]

Of course not. The DAC noise is amplified by the amplifier. That's why many people use resistor-based volume controls (variable resistors or relays and fixed resistors), which reduce the DAC noise. -80 dB noise is easily heard by many people from a speaker system at a distance of two meters. Even I, an elderly person with tinnitus, can hear -75 dB. That's why I wouldn't even consider buying a DAC with noise worse than -100 dB. And preferably no worse than -110 dB. For modern DACs, this is not a problem at all, including ultra-budget ones.

Hmmm, that gives me another possibly setup alternative to current (WiiM Ultra to Genelec G Threes) that is WiiM Ultra => Ifi Zen DAC signature (which has a physical attenuator knob) => Genlecs G Three

Was discussed in another thread, but most members consider the difference inaudible because WiiM implements extra bits for avoiding fidelity loose and extra noise floor when playing at low volumes. I don’t know so much about if is that possible (oversampling to 32/192 by the system according to WiiM).

 

[totti1965]

You missed the frequency in the post
Look at the Frequency vs distortion chart.

I read 5 Watts instead of 5 kHz, so I thought you were referring to @amirm‘s first graph (on top of every test of any device). You are absolutely right!

 

[Sokel]

Hmmm, that gives me another possibly setup alternative to current (WiiM Ultra to Genelec G Threes) that is WiiM Ultra => Ifi Zen DAC signature (which has a physical attenuator knob) => Genlecs G Three

Was discussed in another thread, but most members consider the difference inaudible because WiiM implements extra bits for avoiding fidelity loose and extra noise floor when playing at low volumes. I don’t know so much about if is that possible (oversampling to 32/192 by the system according to WiiM).

My answer of 70 was more about the fact that SINAD is not enough.
If this 70 is defined by the distortion element of SINAD while noise is at check it will be more than ok with music, in general.

There can be conditions though, tailored to make such distinguishable, not any that I have come across though, and believe me, I have tried, at least with normal known distortion patterns.

There, around must be the reason for the everlasting 0.1% pro standard.
Of course it's all for talking now, one has to really try to find such a DAC for the last 20 years unless is deliberate or failed attempts with the knock-offs going around.