Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion Meridian - Systems Apodizing filters already obsolete?

Here is a short article from the Stereophile blog. It seems Ayre has been using apodizing filters for some time and has recently replaced them with a better Minimum Phase filter. It may be that apodizing filters are more widespread than we think in the audiophile community.


Stereophile link

But no pre-echo dictates an apodizing process, does it not?

It sounds like the still use apodizing but have gotten the phase error/distorsion lower than their previous generation.

That would be my understanding as well.

From the description of the talk given by Bob at Christmas.....

"In recent years, in collaboration with Peter Craven, he has investigated the effects of digital anti-aliasing and reconstruction filters, one intriguing result being that Meridian's latest flagship CD player - the 808.2 Signature Reference - uses minimum-phase rather than linear-phase output filtering."

Also worth noting was Bob's comment that they Tested many different forms of Apodizing and in the end ("to Rhonda's "displeasure"") they decided that the one that needed the most DSP MIPS was the one that sounded the best.

In other words, there are likely many different ways to "Apodize" and Meridian have decided on their way

Just to clarify - mimimum phase implies no pre-echo. A linear phase filter has the same pre-echo as 'post echo'. Pre-echoes don't happen in nature, but ringing (in this sense post-echo) does. Pre-ringing = pre-echo, ringing is normally equal to what I just called post-echo, although technically it applies to both.

Apodizing means that the filter taps at the edge of the filter (normally the ones with the least contribution to the frequency response) have had their values reduced.

My understanding is that they're two independent things, but achieving similar results. Mimimum phase gets rid of pre-ringing at the expense of different delays for different frequencies. Apodizing reduces the time that ringing occurs. You could do one without the other.

From the talk, it certainly looked like Meridian were doing both.

As to how Ayre have achieved one cycle of post ringing, that sounds remarkable. Even if it means one sample at the original sample rate after it's been upsampled x4. A time domain response like that implies that the filter is only four taps long, which would be very compromised in the frequency domain. Perhaps they do traditional analogue filtering too, and the digital filter is very light?

Fascinating discussion. I love to learn more about this topic. In the end I am glad that in the end Meridian are making the judgements to get the best sound. I think that is what many people miss- they do not see that every new innovation likely has tradeoffs, and only the true masters know how to make the correct tradeoffs.

This is why I appreciate Meridian -- they actually publish their work in peer-reviewed settings.

According to the Ayre documents, they describe their 2nd generation filters as being based on Craven's "apodizing" filter which, they point out, had "multiple cycles of post-ringing." They go on to say that their 3rd generation filters "combine Craven's minimum-phase digital filter with a slower roll-off that reduced the overall amount of ringing. The resulting filter has no pre-echoes, and only about 1 cycle of post-ringing."

You can't believe everything you read in Stereophile.

Kal

Sometimes it means that competitors can get hold of the latest tech too, though.
I guess Peter Craven is operating more in consultant mode?

I hope that we get the filters soon, it would make sense that Meridian's existing (not just new) customers get them before a competitor's customers. I would hope that it would be proof of their superior support and architecture.

I'm guessing that Ayre won't be offering free upgrades for existing customers?

Meridian isn't using the filter published in Craven's paper. At the AES talk, Bob showed a diagram of the filter which clearly had a slow roll-off. It also had a deliberately-crafted null at twice the Nyquist frequency (IIRC). In other words, I believe that Meridian is using a state-of-the-art filter. I can't recall whether or not Bob showed the impulse response though.

Edit: Clearly though, apodizing/minimum-phase filters are "in vogue" and we have Meridian to thank for pursuing (and funding) this direction of research. Knowing Meridian's penchant for secrecy, I have no doubt that they have not revealed their "special sauce" which is applied on top of the basic, published, research.

Ayre has posted a white paper on the apodizing ("Minimum Phase") filter. The white paper on the USB DAC also looks interesing.

Ayre white paper