Good level piano simulator

http://www.synthogy.com/

Very interesting! Thank you!

Evaluate libraries as well... There are a myriad of them. In my opinion, they are still superior to physical models.

I chose this one and I am still very satisfied with it: http://www.soundsonline-europe.com/Pianos

P.S. Simone, what do you mean by "physical models"? Thanks.

From Wikipedia...

"Physical modeling (modelli fisici)

As mentioned above, physical model generation analyzes the original sound in its complexity and not just in its elementary components, allowing for a reproduction (it is more correct to speak of a re-generation) of the sound with the characteristics of the design. In other words, whereas for a sampled piano one can choose whether to include the reverb of the room in which the samples were recorded, or to enrich the sound with a separate digital reverb module, a physical modeling piano will already possess the sonic structure of the struck strings, the vibration of other instrument components, the acoustic reflections of the piano case, and so on, and therefore will not require signal post-processing.

Having these models available, the physical modeling piano can allow for a certain margin of modification of the produced sound, in which the musician can assign, for example, an arbitrary depth to the soundboard, damping given by a particular type of wood treatment, and other details that are extremely complex, if not impossible, to achieve if simulated traditionally."

For me, it will always remain inferior to samples. Reality is too complex to be explained with mathematical formulas. Physics is not an exact science; it always works within a certain range. If the precision of the formula is sufficient for the purposes of the application in which it is used, there is no problem and it is a powerful predictive tool. If the required precision far exceeds that of the theoretical data, then we cannot rely on this solution. Unfortunately, or perhaps fortunately, the human ear is an extremely precise instrument—far too much! It notices everything and, at least in my opinion, physical modeling pianos are and will always be inferior to excellent samplings. One strength of these virtual instruments is the dynamics, which beats that of samples. For the rest, in terms of sound, I don't think there is a comparison. But that is just my thought... With this, I don't want to take anything away from physical models, which remain beautiful instruments and are also, in a certain sense, fascinating from an engineering point of view. It is only a matter of taste on my part...

TheSimon wrote:

Consider also libraries... There are a myriad of them. In my opinion, they are still superior to physical models.

Very interesting.. do you have any advice on how to find them or any preferences?

@ CromaDiBrera: the link you posted is also interesting.

I believe the limit of samples lies in their size and, above all, in their CPU thirst; to use the full mic setup of the EW Bösendorfer, you need to increase the RAM to 16 GB, even if the sampling is "only" at 44,100 MHz. Then there is another issue to consider, in my opinion: realism is necessary if one dedicates oneself to more or less classical use, but in other genres (such as soundtracks, for example) timbres that deviate from the acoustic instrument are preferred.

@Ombraluce

Search on Google for Pianoteq or Physis Piano

@Croma

Sampling at 44.1 kHz or 44100 Hz; for 44 MHz we are already in the middle of the carrier band for broadcasting transmissions, but it was surely a typo, I only say this so as not to generate confusion for those reading.

Regarding the Bosendorfer samples from EWQL Pianos, I believe the Bosendorfer is exactly the set you are referring to... 8 GB is more than enough if you only load that one. Libraries stutter if the hard disk is slow. That's why the use of an SSD hard disk is always recommended. In reality, the samples are not loaded entirely into RAM. Also because the Bosendorfer, if I'm not mistaken, is around 80 GB, and obviously almost no one has 80 GB of RAM on their PC...

Completely agree with you on the matter of style...

Right, pardon, it was an oversight

As for the RAM, with 8 GB the EW plan wasn't running well for me, but maybe it was also due to the old Play platform (which wasn't 64-bit for Mac). In any case, some time ago you advised me to upgrade everything to 16 (also to manage the other sounds) and, in fact, I had to listen to you and now everything runs great. Besides, it was a modest expense.

Well, having a 32-bit engine implies that even if you have 32 GB of RAM, you can't use more than 4, which is 2^32. I advised you 16 GB because you compose... You understand well that if you have to load many tracks and therefore many instruments, 16 GB becomes necessary, but in the case where you only need to load a single VST, 8 GB is even too much.

But what hard disk do you have?

Hi Ombraluce, I am also of the opinion that sampled libraries are better. Personally, I love the Ivory American Concert D, very affordable price and great quality.

As far as the PC is concerned, rather than lots of RAM you will need an SSD; hard drives freeze constantly and you will be forced to use the lite version of the library.

Happy New Year to everyone..

TheSimon wrote:

But what hard disk do you have?

As for speed, do you mean? It runs at 7200 (1 Tb capacity) and my Mac is 64-bit. I forgot to say, however, that I have never used the piano as a stand-alone, but rather through a notation program or a sequencer; I don't know if this might change anything (Sibelius and Garageband, for example, do not support 64-bit; Reaper 4, which I recently started using, does). Furthermore, there would also be the Play setting (the piano platform in question) in terms of "maximum voices" (currently I have it at 1024) and memory usage; by limiting the voices to 256, for example, it would certainly have worked better (but sounded worse too). It is also true that we are talking about a VST that uses 1268 MB of RAM...

Let's clear things up a bit...

The number of bits in a DAW's architecture limits the amount of addressable data in RAM... At 32 bits, there are 2^32 addresses, which is just over 4 GB; at 64 bits, there are 2^64, so we no longer have problems.

RAM, when associated with a 64-bit DAW, limits the number of loadable VSTs. For example, if I had to load an entire orchestral ensemble, perhaps with 8 GB I would be tight on space, so it's better to opt for 16 GB considering that an entire string section, if of good quality, sucks up at least 4 GB; then we add the piano, winds, percussion, etc., and we easily approach the saturation of 8 GB. If you get close to saturation, the system becomes unstable because if the DAW or other processes need to load more data into RAM, the PC slows down.

The hard disk speed implies the ability to play back what we have written in MIDI without having to use Freeze functions (in Logic) or consolidate functions (in Pro Tools). Basically, if I make everything play in MIDI, the players must continuously refer to the RAM to play the samples of the corresponding MIDI notes at the expense of the hard disk; therefore, if the hard disk is too slow, one or more MIDI tracks are rendered—keeping the MIDI track but muting it—so that the workload for the hard disk is lower, as it no longer has to play MIDI data to be addressed to the VSTs, but rather a rendered wave track... Keeping the MIDI track on mute ensures we can still make changes, i.e., being able to go back by deleting the audio wave track and re-recording it once the changes are made, then putting the MIDI track back on mute.

If a solid-state hard drive (SSD) is used, the data transfer speed is much higher than any other standard magnetic disk hard drive; therefore, it is possible to almost always keep everything in MIDI without having to consolidate tracks.

Regarding voices...

32, 64, 128, 256, 512, etc., are the number of samples that can be played back together. The sound quality is not affected, so you don't hear better or worse by varying the number of voices; rather, polyphony is affected. A piano has 88 keys, so 128 would be fine, but we must consider that in some cases we use tonal pedal + "una corda" (una corda) pedal; therefore, to be sure that all voices are playable, we should work with 256 voices. But then why in a player like Play do we go up to 1024? Simply because it is possible to mix different positions (3 perspectives, to be exact), so the number of voices to guarantee polyphony goes up to 1024. The same argument could be made for other players where we insert various libraries; for example, in Kontakt, we can insert many libraries by assigning a different MIDI channel to each. In this case, we must guarantee polyphony. What happens if the number of voices that should be played is less than the polyphony number set on the player?

Voices are loaded into an array of addresses that performs the routing of a MIDI datum to a position in physical RAM, which in turn redirects the data to the corresponding sample. If the number of data items in the queue exceeds the available positions in the array (the number of positions of which depends on the set number of voices), the first voice called is deleted to make room for the last one dictated by the last incoming MIDI datum. Practically, at an acoustic level, the first sound played relative to the time axis is cut off to allow the last one to be heard.

TheSimon wrote:

Regarding the voices...

32, 64, 128, 256, 512 etc. are the number of samples that can be played back together. The sound quality is not affected, so it doesn't sound better or worse by varying the number of voices; what is affected is the polyphony.

Yes, of course; I wrote crudely that "it sounds worse," but I meant the fact that the cutoff of the polyphony can be heard; without knowing how to measure and explain the problem as you did, I set the voices to 1024 because I noticed by ear that this was the minimum value to make everything sound. Thank you for the explanation.

Great !! As they say, in the end it doesn't matter how you reach a result, the important thing is to reach that result !

I highly recommend Piano in Blue 2 by Cinesamples.

I use Pianoteq with satisfaction. The thing I need and like very much, besides the countless permitted variations, is the customization of the velocity curve which allows me to adapt it to the key response of my weighted controller.

I have tried many of them, and so far there are 4 that perform quite well, although using different technologies, in order (in my opinion) from the best to the "worst" piano: Galaxy Steinway; The big one Sampletekk (which after requesting samples with another range besides Gigastudio 3 I can finally use), Truepianos and Pianoteq (paradoxically 2.0 compared to version 4.0). Regarding impact (which means not having delved into the impact resulting from changing parameters to modify the sound), these are

my impressions, which do not take into account The Pianos by East West, which I do not have but consider excellent, and the library recommended by geppino12575, as I am not familiar with it. But let's not forget the products from 8Dio.