Mornin' Srajan, I'm attaching an email from David Chesky that addresses an interesting topic: linear- vs minimum-phase filters in music playback. David says this: "This became viscerally apparent to me recently in the studio while comparing linear-phase equalizers to minimum-phase EQs. Set to identical filter shapes, the sonic difference was striking. Linear-phase filters preserve phase relationships across the spectrum but introduce pre-ringing artifacts — a kind of temporal smear that occurs before the transient. Minimum-phase filters, by contrast, do all their damage after the transient, creating post-ringing that, while technically less 'accurate,' can feel more musically natural to the ear. The ear can hear this ringing — not as an overt tone, but as a kind of blur, a clouding of the leading edge of a note, an inability to localize or feel immediacy. And this is just from a software filter. Now imagine the cumulative effect of every physical object in the playback chain doing its own version of ringing, from capacitors to cables, from enclosures to air gaps." Interestingly, as a multiple Lumin device owner (X1 and L2), a Lumin representative says that the company employs a linear-phase filter in its playback design. As a side note, the QRONO d2a filter from MQA Labs, now offered as an option by Lumin, employs a minimum-phase filter. So, my question: do you have an opinion on the use of said filters, or, as with many things in audio, do you just trust your ears as to the choice? All best, Michael
In nature there is no pre-ringing. A sound doesn’t announce itself beforehand. It happens when it does, instantaneous. So pre-ringing is unnatural. Post ringing is not only perfectly natural, it's a redundant term. Ringing always implies a trigger of which ringing is the result like the strike of the undamped bell creates the ringing. The trigger comes first, ringing happens after. So in my book, post ringing is preferable because it's natural. Cause then effect, not effect first. Srajan
Comme d'habitude, your point makes perfect sense. I am going to double back with Lumin to understand why the company prefers a linear phase filter, which according to Chesky, introduces pre-ringing. Maybe it's time I more critically compare the Lumin-favored linear phase filter with the minimum phase filter employed by QRONO d2a. Michael
Theory and practice. Particularly in digital they can measure things which we may or may not be able to hear. Since digital sits at the very beginning, there’s plenty of room downstream to obscure, shift, filter or otherwise modify the signal. For example, if our speakers suffer phase shift (most of them do) which is a time-domain error, won’t that be a coarser error than digital pre-ringing? Certain theoretical ideals might win out if our system resolution is very high and time correct but could very well not matter in most real-world conditions. In the end, we listen for enjoyment, not to conform with ideals. If we enjoy our sound, we’re already home-free. It’s when we don’t that we get curious as to why not and what we might fix to get more satisfaction. Srajan
Srajan, your follow-on response reminded me of a putative key attribute of my Vivid Audio B1 Decade speakers: timing. Thus, could any negative effects of pre-ringing sort themselves out in the downstream presentation, as you suggest? Here is a report on the speaker's timing elegance from HIFI+:"Its key attribute is the one that I hold in highest esteem: timing. It sounds so right in this respect that I was bowled over and had to listen to all of Kurt Vile’s ‘Pretty Pimpin’ [b’lieve I’m goin down… Matador] – in fact, just writing this makes me want to repeat the experience, it was that compelling. You can make a speaker that ‘times’ well with a single driver in a simple wooden box, but it won’t be as effortlessly revealing, as low in perceived distortion, and not nearly as evenly balanced in tonal terms as this. Several more tracks later, I was reminded that you can normally hear what a speaker is made of; cone and dome materials in particular have distinct characteristics as do many cabinet materials. Here… nada! Absolutely no hint of metal, no emphasis on cymbals or electric guitars, and no sign of cabinet resonance. Instead you hear huge differences in recordings because the tonal character of each voice and instrument is presented without the ‘little and often’ harmonious colorations that most speakers add. And, as mentioned, timing is spot on, and this applies right down to the lowest notes."
Well, as you rightly point out, in the end it's the enjoyment of sound that matters the most. But I have to admit, trying little tweaks such as employment of various filters does contribute to that enjoyment factor! Thanks for the conversation. All best, Michael
You can’t correct errors in time in the analog domain. That would require time travel. You can correct for time delays in the digital domain, for example where speaker alignment and placement are concerned. So if a tweeter was slightly forward/early because its baffle placement put its acoustic center closer to the seat than the woofer, a digital crossover could delay the tweeter by just the right amount so the time arrival at the ear is coincident. But you can’t correct digital pre-ringing downstream in the analog domain. It's now embedded in the signal. My argument was the opposite. I argued that grosser time domain errors in loudspeakers might make digital pre/post-ringing differences inaudible because the speaker errors far outweigh the digital errors. By that logic, a time-correct speaker driven by electronics without phase shift should introduce fewer time errors, ergo what happens in the digital domain in the time domain might actually be audible. Srajan
I had to think a bit more about what you wrote; thanks for correcting me about correcting errors in time! I did some more digging about pre-ringing caused by Lumin's use of a linear phase filter and discovered: 1/ With modern very-long filters running at high sample rates, pre-ringing amplitude is extremely low. 2/ Above 96 kHz or in DSD mode the time offset is so small that it’s typically below audibility. 3/ Many listeners actually prefer the greater phase accuracy of linear phase filters to the warmer but phase-shifted minimum-phase filters. In addition to the above, if I use Roon, which I do occasionally vs. Qobuz Connect/Lumin app, I upsample everything to DSD 128. Apparently by doing so, I further minimize any issues. Whew! Thanks, Michael
As David Chesky pointed out, there are many opportunities in the playback chain and before it, in the recording chain, for time errors to slip in. Passive crossover components ring. Coupling capacitors incur time delay. If we focus on eliminating just one culprit, it may be insufficient to make audible improvements down a chain which is still inherently time confused. So my thinking is that if we're more time than amplitude-triggered listeners, we can't just stop at one time-error contributor like digital pre-ringing but we might want to add wide-bandwidth direct-coupled amplifiers, widebanders without 1-3kHz crossover, active not passive crossovers, cables that don't add their own fuzziness, directional bass to minimize room-mode ringing, active bass traps etc. In short, we might need to apply a holistic not partial view on the topic of time and address as many of its contributors as possible to make enough of a dent? It's certainly how I've approached building my most recent systems guided by Raal 1995's Immanis headphone which because of its openly vented ribbons suffers no phase shift, energy storage or cavity resonance. It's a good example for a transducer not bedevilled by common timing issues. It's why I think that exposure to summit HeadFi is key even for speaker listeners to know what's possible and then apply that insight to their speaker selection, setup etc. One needn't end up becoming a headphone freak to benefit from such exposure. Srajan
Srajan, I now see your end game: the holistic audio system. I am part way there with my Kinki wide-bandwidth direct-coupled amp and complementary Earth loom. I guess I could include my cable risers to avoid minute vibrations from infecting my speaker cables and various IsoAcoustics vibration isolators on my audio devices. At any rate, thanks for the illuminating discussion. Michael
The word 'system' and 'holistic' are virtually interchangeable when we think about it more. In a system, everything interacts with everything else. System means interconnectedness after all. That's what makes it into a system not an arbitrary crowd. Whilst we might think we're just changing one thing, we in fact always nip and tuck on what's already there. So the only difference is to apply intention and awareness. Do we do this thing haphazardly like, let's see what happens, or do we have some sort of vision that charts a path we can follow? Regardless, the arbiter whether we're making progress or not is completely personal. Are we getting the 'aha' experience more often or not? There clearly is no one-size-pleases-all formula. Just because I'm a bit more focused on timing doesn't at all imply someone else shouldn't instead concentrate on scale, dynamics, imaging, timbre, textures – whatever triggers them most. Srajan
