Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Physis Piano

Thank you for your reply Gary.

Do you know the architecture/vendor of the CPU? For example, is it x86 or Risc based? Intel, AMD, ARM, etc.?

Cheers,
James
x

Thanks for contributing Gary.

One or two questions/comments:

Is the stand seen with the Physis Piano supplied with it? It looks very stable and it appears to be branded Physis Piano.

Have you any experience of the V-Piano? If so, please compare tone/touch and dynamics etc between Physis Piano and V-Piano.

Just how far/much is Viscount going to properly back the Physis Piano? In the UK, the supply chain and product support for most Italian brands has traditionally been very flaky and the last thing anyone wants is to spend lots of money on something they can never get fixed or get queries answered. The key action being Fatar is not good news - not because some of their actions aren't good but rather that Fatar are known for failing to offer good after sales support for their products.

If the price is even close to V-Piano then I think Viscount is wasting its time. The V-Piano comes from a very respected brand with an acknowledged track record. But even so, purely in terms of sales it has been a relative market failure, perhaps mainly due to the high price. Viscount will need to VERY CONSIDERABLY undercut Roland - in the UK I would say (regardless of how good it is) £3000 as an absolute maximum otherwise it is doomed to fail in my opinion - even at that price the going will be tough because Viscount/Physis is almost unknown.

All that said, it sounds ok to me and I'll look forward to trying one out one day.

Cheers,

Steve

From the description this would have to be an 6-core Intel Xeon type CPU. HP and Apple sell computers with this processor. Street prices are around $3,000 for either a PC or Mac with this CPU. So there is no way the piano is going to be cheap.

The lowest price RISC machine that can do 30B instructions/sec on 6 cores cost WAY MORE than $3K, like 10X more and the ARM does not come close to the described performance. So Linux running on a Xeon sounds about right.

So in short this runs on "PC like" hardware but more like the hardware found in a data center than on a typical PC desktop. Or to "Apple peole", it is a mid-range Mac Pro running Linux.

So I guess it must be better cause of its advanced technology. I guess I just don't hear it and it's not just that I'm only hearing web recordings. I think the piano is too organic to be made by robots. Sampling isn't perfect but it at least captures the organic essence of wood, steel and felt. I like the fact that sampling keeps improving with technology but it's still originating from a design that's taken years to perfect.

That's still an argument for continuing with modelling though. When they finally get it right, it will have significant advantages over sampling. First, the size of the program itself will be much smaller and have an unlimited amount of velocity layers (layers being a sampling term). Second, as super fast processors get more and more common-place modelling will not tax the average processor much at all, so the hardware cost will drop. The only real problem is that modelling is still missing a few pieces of the puzzle - but it will get there eventually.

As big as they are, with as deep pockets as they have, and their experience and resources as they have in the real McCoy, it's surprising to me that Yamaha isn't light years ahead of everyone else in terms of modeling. Yet they barely seem like they're even in the game. Especially if what Gary said earlier about them breaking ground early, then pulling out due to a lack of hardware horsepower.

I'm pretty sure Yamaha are working on physical modelling too.

I'm sure they are, and probably don't want to put out something half-arsed. Just surprised they haven't yet.

Yamaha uses a hybrid sampling+modeling system (SCM = Spectral Component Modeling) in their CP1/CP5/CP50

Yeah, I think you're spot-on.

Regardless of the feel, which is unknown, and based only on how "real" the sound is, V-Piano is safe.

Yet, I welcome any new digital piano that's not a recompilation of the standard Yamaha sample set. We need more of these things on the market. I wish the developers all the best.

I would love to wrap my ears around modeling but you might say that they have made the clone but they can't make it think for itself yet. I'm always ready to dump my current DP for something really better but not only does the DP have to have the sound, it also have to have the perfect feel.

pianoxcape,

Granted you are trying to promote this new instrument so any comparisons you may have with the V-piano or any digital piano will probably be biased in favor of the instrument you are helping to promote. However it is good to see competition for modeled piano market as I do agree it is the future. Your opinion is welcome in this forum and any comparisons you may have will also be welcome. However I too will reserve judgement until I get to play one in person with my own headphones. It would have been nice to have a representative like you that was as knowledgeable about promoting the V-piano way back then. It may have generated more interest in the product early on. However the price of the V is still too high to bring it to the mainstream. Furthermore Yamaha hasn't even budged on trying to go back to the technology anytime soon. I do hope this does well and holds up as a great keyboard on its own. It will be interesting to see it go head to head with the now older V-piano.

I do have two questions for you though. While you are playing, can you take two distinct piano sounds and play them together or is it only modelling one at a time. Also, can you be playing a song and without interruption can you switch over to another modeled sound. Both of those features the V-piano cannot do, so it would be a selling point to mention that if the new Physis piano does.

I believe that depends on your definition of the word 'safe'.

My definition of the word 'safe' is something big, heavy, and difficult to move.



So I guess the V-Piano qualifies.

Perfect!

That's it.

Though in recent years storage and sampling technology have progressed more quickly than processor and programming (modeling) technology in my view. No need to assume sampled pianos and the associated technology will not continue to improve as well. It appears that the best pianos will have some elements of both for some time. In my opinion the day when the best sounding pianos have no elements that have ever come through a mic is very distant. Not only will they need to overcome the hurdle of sounding like a piano, but then they will have to overcome the hurdle of sounding better than the acoustic pianos used in sampling, which are the product of hundreds of years of innovation.

Though I guess whatever tech the big three use to store their onboard sounds has not improved at a similar rate as the computer storage we use for software pianos--otherwise they would not continue to loop and compress--so perhaps you are right.

With respect to the price argument, in the spirit of nit picking I should point out that the processor hardware comprises a very tiny portion of the cost associated with, for example, the V piano. So processing speed gets cheaper that doesn't really affect the price of a V. The cost is in the programming and design, which is inherently (I believe) more expensive than the cost of sampling and then adding a few extras (like resonance, layer blending, etc.). So I don't see costs dropping for modeled pianos any faster than they do for sampled pianos, unless it be because of supply/demand considerations.

I was pretty much going on the figures ChrisA posted on the first page as an argument for why the Physis will be an expensive beast. It may not explain the V-piano's price but it might well explain the price of the Physis. 6 core processors are pricey. If Roland uses a proprietary chip, they have to recoup that cost too.

In any case, I suspect what is partially hindering the modelled piano concept is that it's being done by engineers with fatigued ears. I really think you can lose perspective of tone when you are hearing the same kids of sounds over and over again. (a bit like when you spend days mixing/mastering multi-tracked recordings). You start hearing things in a very different way. It seems to me that it's critical that the development team not be too overworked when developing a modelled piano. From what I can gather these modelled instruments are put together by very small R&D teams under a certain amount of pressure to get the product released. I'm not convinced they are getting, nor giving themselves, the necessary objective feedback about the tone. I mean, even basic spectral analysis should be able to point them in the direction of what might be missing from the sound. A model needs to resemble something in result terms, not just blindly go off the physical model calculations. If the result is off, something is amiss in the model. Hard to believe they can't work out what it is. Despite the hyperbole of the Physis demo, 100 parameters in a complex model is not really that many.

Of course, this is all speculation on my part. Just thinking out loud.

I still think Sampling will always be somewhat cumbersome because as you get closer to perfection, you also keep increasing the number of sample layers. That's not an ideal situation IMO.