Minimum Phase vs Linear Phase

[markanini]

In theory minimum-phase would be sharper, not smoother, given reduced pre-echo, and the ears being insensitive to post-echo.
In practice any style of temporal smearing is above the audible passband. And, as @Mike123 said, legal audio signals are already low-passed and negate the chance of audibility.

There may still be a reason for minimum-phase sounding sharper, not smoother. The phase distortion may increase the amplitude level of peaks, whereas linear-phase keeps peaks more intact.

 

[welwynnick]

In theory minimum-phase would be sharper, not smoother, given reduced pre-echo, and the ears being insensitive to post-echo.
In practice any style of temporal smearing is above the audible passband.

I don't think that's true because many filters cause the phase response to become non-linear WITHIN the audio passband, especially with 44.1 / 48kHz sample rate.

 

[markanini]

I don't think that's true because many filters cause the phase response to become non-linear WITHIN the audio passband, especially with 44.1 / 48kHz sample rate.

Phase distortion is not generally audible within or beyond the audible passband.

 

[Keith_W]

Phase distortion is not generally audible within or beyond the audible passband.

Sorry, this is incorrect. Please see JJ and Thomas Lund's posts in Bass and Subwoofers and Does phase distortion/shift matter in audio.

 

[markanini]

Sorry, this is incorrect. Please see JJ and Thomas Lund's posts in Bass and Subwoofers and Does phase distortion/shift matter in audio.

Do you have direct evidence? I don't think these links can satisfy. One of them affirms my position in the very tile. The other deals with speaker engineering where filter choices have the goal of mitigating unwanted relative phase distortion in the audible passband. My position concerns absolute phase distortion audibility. I'm primarily requesting ABX-able audio samples.

 

[Keith_W]

Do you have direct evidence? I don't think these links can satisfy. One of them affirms my position in the very tile. The other deals with speaker engineering where filter choices have the goal of mitigating unwanted relative phase distortion in the audible passband. My position concerns absolute phase distortion audibility. I'm primarily requesting ABX-able audio samples.

Unfortunately, I don't. I asked JJ in that thread whether he knows of any published studies that prove his assertions that phase distortion is audible, and he responded by giving me a file to play on my system. It was quite easy to hear. If you are interested, that file is somewhere in the thread. Unfortunately, that thread is rather long so it might be difficult to find.

The discussion in that thread "Is phase distortion audible" starts with classical thinking supported by many audio authorities including Linkwitz and Toole and even has a name - Ohm's Acoustic Law - that phase distortion is inaudible. So you are in good company there. From what those guys are saying, it seems as if the view has changed in the last few years.

FWIW, I use linear phase DSP and it is possible to make the overall phase response of my system completely flat from 20Hz to 20kHz. Realistically I can't obtain reliable measurements below a certain limit, so I linearise my phase between about 200Hz up to 20kHz. I also have a set of minimum-phase filters and I can switch between them instantly with zero latency using Hang Loose Convolver. I can hear a difference. I realise this is an uncontrolled subjective anecdote with little persuasive power, but if you have the ability to implement linear phase DSP you should give it a try.

Look, I don't expect to convince you and I understand your scepticism. I was VERRRRRRRRRRRY sceptical myself and I could not bring myself to abandon what was very clearly stated by Toole in his book. Like you, I want studies and published evidence. But in the end, what I heard was pretty convincing, and if JJ is prepared to stake his reputation on it ... it's good enough for me.

 

[welwynnick]

Phase distortion is not generally audible within or beyond the audible passband.

I don't think that's true, either.

PMA was able successfully ABX DBT a rough square wave with and with out phase distortion:

 

[tmtomh]

I don't think that's true, either.

PMA was able successfully ABX DBT a rough square wave with and with out phase distortion:

View attachment 465412 View attachment 465413

I don't think a square wave falls within what @markanini was referring to when he wrote that "phase distortion is not generally audible within or beyond the audible passband."

Music signals aren't generally isolated square waves.

 

[Mike123]

I don't think that's true, either.

As soon as we want to detect the influence of filters, we immediately use meander signals . When creating these meanders, the influence of filters is noticeable. Because the filtering is applied very low - from 1 kHz. In real music there are no meanders, and the music is cut by sound engineers to 22 kHz with high-quality filters. Perhaps you will even hear the difference. But their influence is already in the recording and does not change in any way when choosing filters in the DAC. Filters in the DAC have nothing to filter, everything is already filtered before them. In the picture on the top graph, I applied a 20 kHz filter to a 100 Hz meander. Feel the difference

 

[welwynnick]

I don't think a square wave falls within what @markanini was referring to when he wrote that "phase distortion is not generally audible within or beyond the audible passband."
Music signals aren't generally isolated square waves.

To the best of my knowledge, we're not generally sensitive to phase distortion with continuous wave signals, but we are sensitive to phase distortion with transients, so a square wave is more likely to sound the same, while music is more likely to sound different, as it does have transients. PMA's finding seemed to go against the flow of common understanding, but it was a convincing DBT and it stood out to me. If it's possible to hear phase distortion in a CW signal, it suggests we can hear it in music as well. I'm coming at this after years of listening to electrostatic speakers, which have more linear phase response than cone speakers, especially three-way speakers with complex multi-pole crossovers, which can have fantastic amplitude-frequency response, but are a dog's dinner with phase / time domain response. Companies like Genelec and Neumann and many others are heeding research that shows our sensitivity to group delay, and are now producing DSP active speakers with Extended Phase Linearity, Automatic Monitor Alignment, Compensated Phase Response and Phase Correction. The result of all these schemes is to linearise the phase response and give coherent time domain behaviour. The benefits are usually described not as better fidelity or resolution, but as better soundstage and focus, which is what all my electrostatics were so good at. That's what Trinnov users hear, too, and Trinnov is effective at improving phase linearity, like Acourate, Audiolense or rePhase etc.

 

[markanini]

I don't think that's true, either.

PMA was able successfully ABX DBT a rough square wave with and with out phase distortion:

View attachment 465412 View attachment 465413

Thanks, but I'm trying to maintain a robust criteria to satisfy my challenge. I'm even ignoring my personal criteria for using linear phase filters.
You can't essentially prove audibility of phase distortion this way. It proves a limitation of integer bitrates. The distortion can be mitigated by adding negative preamp, or using a float-format.

Here is an example of the same logic reaching a different flawed conclusion:
"I applied a +6dB peak filter at 1kHz to a 1kHz sine wave at -1dBFS amplitude. Next I applied a -6dB peak filter at 1kHz to the signal. The signal was distorted. EQ is not reversible."

 

[welwynnick]

You can't essentially prove audibility of phase distortion this way. It proves a limitation of integer bitrates. The distortion can be mitigated by adding negative preamp, or using a float-format.

Could you elaborate on why you think it doesn't identify the audibility of phase distortion?
The only thing that changed was the phase of the third harmonic.
Are you talking only about the phase response of the reconstruction filter?

 

[markanini]

Could you elaborate on why you think it doesn't identify the audibility of phase distortion?

The second waveform is clipping. It's not reliable.

 

[KSTR]

To the best of my knowledge, we're not generally sensitive to phase distortion with continuous wave signals, but we are sensitive to phase distortion with transients, so a square wave is more likely to sound the same, while music is more likely to sound different, as it does have transients.

It is well-established knowledge that at low frequencies we are sensitive to the actual waveform. A steady-state mix of sinewaves sounds different depending on the actual phase relationships.

The ultra-simple and most robust test signal for everyone to check for themselves goes like this:
Take a sinewave around 100Hz or so, say 80Hz.
Then add the second harmonic of it (equal amplitude) but ever so slightly detuned, like 160.5Hz, 0.5Hz off. The spectrum of this is rock steady.
But when we listen to it, it has some periodic 2-second cycling pf perceived timbre, alternating between fatter and leaner sounding. That's the effect of the actual waveform which slowly cycles through the phase rotation of the 2nd harmonic..
Make sure you don't have clipping nor significant distortion, best use headphones at a rather low volume.

 

[markanini]

The ultra-simple and most robust test signal for everyone to check for themselves goes like this:
Take a sinewave around 100Hz or so, say 80Hz.
Then add the second harmonic of it (equal amplitude) but ever so slightly detuned, like 160.5Hz, 0.5Hz off. The spectrum of this is rock steady.
But when we listen to it, it has some periodic 2-second cycling pf perceived timbre, alternating between fatter and leaner sounding. That's the effect of the actual waveform which slowly cycles through the phase rotation of the 2nd harmonic..

Do you believe this represents the height of phase distortion audibility?

 

[welwynnick]

The second waveform is clipping. It's not reliable.

That's not clipping at all - if it was, the FFT would be different, but they're identical.

Take a sinewave around 100Hz or so, say 80Hz. Then add the second harmonic of it (equal amplitude) but ever so slightly detuned, like 160.5Hz, 0.5Hz off. The spectrum of this is rock steady.
But when we listen to it, it has some periodic 2-second cycling pf perceived timbre, alternating between fatter and leaner sounding. That's the effect of the actual waveform which slowly cycles through the phase rotation of the 2nd harmonic..

That's different, that's not phase distortion. Again, the FFT would be different.

It sounds like all we're agreed that phase distortion is audible.
Nothing wrong with that, it's entirely consistent with audio science principles - we can hear it and we can measure it, and it's really not that difficult or complicated.

Personally, I've been struggling for DECADES to reconcile conflicting subjective and objective measures, but now it's all coming together.
High fidelity simply requires a simultaneous conjugation of envelope and linearity:

• Wide bandwidth
• Wide dynamic range
• Wide load tolerance
• Linear amplitude response
• Linear frequency response
• Linear phase response

Everything altogether - not just best case 1V, 1W, 1%, 1ms, 1dB, 1kΩ, 1kHz, 1µV etc

 

[Kiko1974]

I own an SMSL D400 PRO DAC that uses the top of the line AKM chipset, that is AK4191+AK4499EX.
I use for all my music listening the Short Sharp digital filter. It may be because It matches well with my humble, yet good sounding,Marantz PM 6007 integrated amp plus KEF's Q550 speakers, room acoustics, or just my personal taste.
Other filters on this DAC sound "typical digital", like Sharp Roll-Off, or dull like Slow filters.

 

[Julf]

Other filters on this DAC sound "typical digital", like Sharp Roll-Off, or dull like Slow filters.

Uh uh...

 

[Kiko1974]

Uh uh...

That may be, as I said earlier, because of my personal taste, or a matter of sinergy between my integrated amp, speakers and room interaction.
I think the choose of digital filters available are there for a reason, for example to do a better match between the DAC and amp and speakers, room acoustics or to tame overly bright recordings.

 

[Julf]

That may be, as I said earlier, because of my personal taste

Quite possibly.

or a matter of sinergy between my integrated amp, speakers and room interaction.

Hopefully your amp is not adding coloration and/or distortion.

I think the choose of digital filters available are there for a reason, for example to do a better match between the DAC and amp and speakers, room acoustics or to tame overly bright recordings.

If you want that, proper EQ is a better solution. The reason vendors provide multiple filters is to make sure they cater for all customer preferences and beliefs.