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

Hmm. I might need to get hold of a copy of that...

VirusKiller Have a look on Duncan's site for the review

Regards

Thanks!

Hi Ian,

I missed your welcome the first time I read your post. The vision's not as sharp as it used to be....

Thank you for the kind words and also the warm welcome. I had no idea that this forum existed until a few days ago. Nice place you have here!

Cheers,
Charles

Your eye sight is fine - your response even sharper - as you had managed to respond in the few seconds it took me to add the welcome to the original post.

http://www.Meridian-audio.info/public/meridian_808_2_hfnrr%5B586%5D.pdf

Hi -

Thanks for the tip. I have been waiting for Hi-Fi News to put that reprint up for sale on their website, but it wasn't there when I last checked about a month ago.

If you look at the impulse response in the Lab Report sidebar, and then compare it with the impulse response on page 3 of our white paper at:

http://www.ayre.com/pdf/Ayre_MP_White_Paper.pdf

you will see that the two impulse responses are quite similar. The images in the white paper are of the actual filters used in our previous generation of products (pages 1 and 2) and the new "MP" series products (pages 3 and 4).

While the two digital filters are very similar in overall performance, the final sonic performance of any product is influenced by a host of other factors. For example, we found that performing the oversampling in one pass instead of the usual concatenation of 2x stages (done to save computational requirements) made an audible difference also. (This has been done in various top-of-the-line Japanese CD players over the last ten years or so, most of them only sold in the Japanese home market.) Then there are the variations in DAC chips, power supplies, analog circuits, et cetera, et cetera.

But the "minimum phase" bug is definitely catching on. The latest top-of-the-line DAC chip from Wolfson has some minimum phase options and is being used in the new PS Audio DAC. I'm sure we'll be seeing more products with minimum phase filters in the future.

I had been thinking about this for a long time but never got around to trying it until we designed our first new digital product in many years. (I was also curious about filterless "non-oversampling" designs, but these didn't work well in our application.) Also, this type of computational power only used to exist in DSP chips. Now it can be done with FPGA's instead, which offers some advantages for us.

Charles Hansen
Ayre Acoustics, Inc.

Hello Charles,

Just like to add another welcome to you on this forum. It's really very interesting to read your posts from the practicing digital designer p.o.v.
We often speculate wildy about such things.

I noticed that the PS Audio equipment is getting some very good reviews and guess that this may be much to do with the listeners' preferences for this type of filtering.

Interesting to note that whilst you may have all started to converge on a similar filter approach, you indicate that there are still plenty of ways to differentiate your products. We often tend to mistakenly believe that ultimately all digital transports will offer the same performance.

Your thoughts on this ?

Cheers

Thanks for the kind words and warm welcome.

Regarding digital transports, my experience is that *everything* makes a difference. Sometimes these differences are large and can be heard by just about anybody in any listening situation, sometimes these differences are small. But when you are listening to the same system (your own) for hundreds of hours, listening to your own favorite music, even the small differences (cables, racks) can be heard.

That's why it's always best to listen to *any* component you are considering in your own home. A good dealer should be willing to arrange for an in-home evaluation, at least overnight. (And you should expect to pay full retail for such service.)

Transports fall into the category of things for which we have nothing more than "hand-waving" explanations of why they should sound different. Because I've heard differences with things like disc-damping mats in equipment that I've designed and can think of no plausible mechanism for such differences to exist.

The bottom line is that setup is probably 50% of a system's overall sound quality. If you haven't already done so, I'd recommend Jim Smith's book "Get Better Sound". It's well worth the $50. Most of the tips are free or will cost less than $10 to implement.

Welcome Charles. Nice to see you hear.

I hope you don't mind two OT questions.

1) Is there an easy way for an amateur to measure the word length of the part of the signal that actually contains information? Some of the older Meridian products only accepted, e.g., 18 or 20 bits. I was curious if one could determine the actual word length a source is outputting. I know it is an odd question.

2) If one upsamples 48/24 to 96/24 and then back to 48/24 with no operations inbetween, will the original 48/24 be identical to the up and downsampled 48/24? If some DSP is done inbetween at say 48bit, in general terms, how close will the up and downsampled 48/24 (now decimated from 48bit to 24bit) be? I know, another odd question.

Thanks.
George

If you want to know what actally reaches your ears I would think you could do this with a series of test patterns with successively wider words.

Thanks for the nice welcome!

1) I don't know of any way to do that easily. I know that JA has mentioned doing this several times in some of his equipment reviews, but I think he was using some special software that came with an RME sound card. You can do an Advanced Search on Google for "RME" on the "www.stereophile.com" website and find some more information on this.

2) The first part is going to depend completely on the algorithms used by the upsampler/downsampler. Ideally it would work this way, but there are lots of ways for things to go wrong. However, the odds of success are *much* greater when doing a simple 2:1 conversion. Going from 44.1 to 96 and back is guaranteed not to be bit-perfect.

When DSP is being done, the 48 bit result is almost always going to be dithered back down to the 24 bit level. Lipshitz and Vanderkooy "proved" that +/-1 LSB of TPDF dither is mathematically correct about 20 years ago, so that is what most people use. Therefore you can have up to 2 LSB's of difference just due to the dither applied at the bit-depth reduction stage.

Hope this helps.

Very helpful.
Thank you.