Not I disagree with some of your sentiment, but Naim have many years in designing CD players… and with many key aspects the considerations between streamers and CD players are the same… indeed they are both use micro controllers or generic computers.
It’s funny in an irritating way, I have a very recent MScalar, and the supplied BNC connectors are better than anything else I have for the job in hand … including my Wave Storm reference cables… which are irritatingly now redundant.
Perhaps Chord Electronics sorted the quality / design issue out on those cables.
My view it’s simply an alternate way of achieving reconstruction…
As you no doubt have heard me say digital to analogue reconstruction is an imperfect process… it can only ever be an approximation… you choose which compromises to make and focus on other area…
For most on average replay equipment this all happens below the performance/resolution level of your replay chain. On higher perfoantvequipment and replay chain, these compromises can become noticeable … and so you choose which is the most comfortable replay chain for you.
I am wary however with companies who focus on ‘convolution filter’ as a plus… this is marketing hype.
Convolution is the mathematical term used to describe multiplying two number series together … in other words one is simply using Finite Impulse Response filters… which is where filter kernel and taps come in.
An equally valid alternate are Infinite Impulse Response filters … which is what Naim use… there is no convolution here, or taps… these are simply design choices perhaps to suit your hardware footprint and your replay compromises as mentioned above.
However if the use of FIR filters in your implementation means you need to modify all your media in non real time… something to my mind is sub optimal and inefficient … and how are you going to reconstruct streamed media… feels like tail wagging dog…
Therefore you might be better off with a solution using IIR filtering instead or as most do undertake FIR filtering in real-time.
I will hopefully get around to converting a couple of albums with that app but, I feel it’s the least efficient system available at this point for upscaling etc and, as I mainly use Roon to stream new music rather than buying / downloading material, it would only be of use with my existing library… which can be upscaled via HQPlayer.
I’m still interested in trying it, just can’t envisage how it will be of benefit at this point.
Sure if you are interested, there are many tools you can play with to adjust media… including DAW mastering modules… if you have a DAW of course (digital audio workstation)… but you may find many modern masters are pretty much fully optimised already… but older recordings it can be fun.
Interesting about your use of Roon, I have let Roon take control of my media access, either cloud or locally stored… I have simply merged it together… and of course no there is no ‘SQ’ difference between cloud or on prem with Roon… in current IT parlance my music media is licensed to me on a hybrid licence arrangement
Music in many ways is just like software… you never actually own the media content (unless you own the copyright ) … you just own the right to replay it… either perpetually or via subscription
I have about 1tb of media on the Roon NUC and of course Qobuz… it all just blends into one source with Roon. I just let it do its thing… now and again I spot that it has used local material other times Qobuz. If I really like an album, I will buy it just in case it disappears from the streaming service.
Ah ok, you use it like me… albeit I have Roon core mount my NAS.
That’s interesting as I dismissed the supplied BNC leads in favour of some Mark Grant HDX1, thought the supplied BNC plugs were too wobbly I may pop the supplied leads back on for a further listen.
I think as Simons says, his MScaler was a recent purchase so they may have improved them a little. I know one of mine was beyond wobbly and you could hear crackling over the speakers and cut out when touched.
My MScaler is only a few months old and the supplied leads do wobble whether this is an issue I’m not sure but the Mark Grant HDX1 are a nice tight fit.
The HDX1 is a good cable. Out of all the BNCs I had, it is the one I kept just in case I need one at any point.
Yes I have seen reports about leads being wobbly etc, but not recently, and mine are not wobbly which is why I think they might have changed the leads.
I have been on the PGGB beta test team since the end of last year playing it to my DC4 Dave via USB. Part of the reason for getting my K50 was so I could play 705 and 768 files. It also has 12TB of local storage so the large files are not a particular issue. I process the PGGB files on my iMac in the background whilst doing other work on it.
A recent upgrade in sound for me has been getting the AudioWise SRC.DX usb to dual bnc converter. With usb input to Dave I was processing the PGGB files at 32bit but when using the SRC.DX the files need to be processed at 24bit. The logic in using the dual bnc input to Dave is that one is bypassing the Dave Amanero usb controller which is said to have a high latency. This all started with the guys using the Taiko Extreme but many of us have found there are benefits with Antipodes and I am also now using an Innuos streamer with the 2.0 beta app at 705/768 and that also works well with the SRC.DX.
That is the good news, the bad news is that the SRC.DX is equally sensitive to the bnc cables used and this was identified by austinpop on AS when he discussed the use of the SRC.DX with the Dave and PGGB files. I have experimented myself and just to be doubly sure before posting here I popped in a pair of ‘normal’ bnc cables. They lasted 2 minutes before I stopped the track and swopped back to my favoured cables.
I do like PGGB a lot but I am not selling my MScaler and will continue to use it for streamed music and for playing locally stored audio files that I haven’t bothered to convert. Also, I am sure at some point Chord will drop a new Choral range MScaler with many more taps and that will be an interesting prospect.
I wonder - there is a relationship between number taps (filter kernel size) and the sample word length, so it start to become a case of very much diminished returns unless the sample word increases. There is also a relationship between filter kernel size and bandwidth - so for a given number of ‘taps’ the effect is more beneficial at 44.1kHz than 192kHz - so that could be a future advancement…
However I wonder if the focus might move into more the WTA filter optimisation for the same number of ‘taps’ as currently used in MScalar. More taps creates more noise and starts to counter the benefit of increased filter kernel size - and there is some point on the graph given the diminishing returns that the effect or more taps is actually a negative.
It is rumoured that RW is working on a 6M tap MScaler so I guess he considers that the number of taps pays dividends.
He may but - but it is a rumour - however I have discussed the above points personally with him(other than optimising the WTA filter) and he did agree with a smile…
So to have the same benefit of 1m sample filter kernel (taps) at 44.1 sample rate would required 4m taps at 176.4 sample rate - so this might be an angle he is looking at.
But yes if you are using FIR filters the larger the kernel size the better (mathematically) - up until the point the advantage is cancelled due to the side effects from the engineering of realising it… hence why windowing methods can be a beneficial addition.
I can’t exactly remember the source of the rumour but it might have been RW in one of his Head-Fi posts . . . . .
Thanks for the reply Nick, so can you tell me exactly how big an album becomes when you run it through the PGGB process? I did not look at it too closely yet, but most of my stored music is CD quality (16/44.1).
Hopefully Toonartist has some time to convert a few albums this week, to give his opinion of its worth.
I’m hoping to copy some stuff tomorrow. Stuff around the house stopped play but I’ll use my work Mac tomorrow to copy a couple of albums. I’ll try a cd quality album and maybe some high res stuff to see what the difference is… in size and sound.
If you process to 705/768 32bit then I work on 20GB per album.
32bit is best if connected to Dave by USB.
However if using the SRC.DX converter so that you can connect to Dave by dual bnc then it is best to process the files at 705/768 24bit. The PGGB test team are unanimous that a 24bit dual coax to Dave is better sound quality than 32bit usb to Dave no matter whether using Roon, HQP on pass through or MPD etc (or indeed if using the Innuos 2.0 app which I now have to beta test).
24bit also saves about 25% on file size. If your playback system recognises wavpack wv files then PGGB can be set to save in that format to save even more space.