Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Piano Soft vs DP Internal sound

Hi,

I was just wondering why so many of you are using softwares such as Ivory, Pianoteq instead of the internal piano patches of your DP ?

No experience with these softwares ... Do they sound far better than the main piano sounds of high end DP (CV409, CLP380, HP207, 700SX/GX, ...) ?

If yes, do you know why ? I mean why can't we find better piano sounds with high end DP ? Is this related to fact that high level libraries require a strong CPU ?

Regards.

I think many of us simply need a new toy now and then. We are compelled to try new things.... but that's just a guess because I'm not one of those who've tried the "software piano".....but I would like to experiment with some of them.
Bob

The following two recordings represent the same piece being played by me. It is Clementi Opus 36 Sonatina 1.1, which of course is a practice piece. Also, I am a beginner (can’t you tell from my slow playing?).

Virtual (Software) Piano - Akoustik Piano

Resident Clavinova CLP-230

The computer is a DELL XPS with 3GHz dual core processor that cost me $1900. The external audio interface unit is a $130 EMU 0202 (to record the CLP-230).

My own opinion is that for a beginning player like me it is not worth the expense of a virtual piano. Perhaps when my playing improves the differences will seem greater. But at my current level it is hard for me to even tell the difference between the two. I usually play with the resident sound of the CLP-230 and keep the computer in its recording configuration, rather than switching various connecting cables to use in virtual piano configuration.

I needed a new high end computer for recording purposes anyway, so I don’t regret that. But the $300 Akoustik Piano software was in my opinion a waste.

You be the judge.

I may in future purchase different software, perhaps Ivory, but not for a while due to recent economic downturn.

Software libraries are hands down better than the ones that come built into your piano. Storage, CPU, RAM it all comes into play and a $500 computer far exceeds specs on a $2000 DP.

Consider 40GB Ivory samples to a measly 20MB built into a low end digital. That's 3 orders of magnitude of difference. That's why digital pianos need to use tricks - like looping of sustained sounds, and interpolating between keys to squeeze everything in. The end product is piano audio that sounds artificial.


Orez Eno:
For future computer purchases, depending on your familiarity with computers - you may want to save yourself hundreds of dollars (in your case almost $1000), and build your own machine (you can buy parts really cheap from tigerdirect.com). On top of that you won't have to rely on proprietary dell hardware.

Eternal,
As a retired electrical and software engineer I feel reasonably qualified to select the model of computer that I need, although I refrain from calling myself an expert. The computer industry changes so fast there are no experts. Depending on processor and a wide variety of configurations, the DELL XPS series is considered by many in the industry to be the best for high end applications, and they can cost in excess of $5,000. The system I purchased has phenomenal speed and I am very satisfied with it for the price I paid. Indeed I have never heard anyone who has invested in such as system ever complain about it.

I’d much prefer my XPS to any “Rin-Can, Crash-If-You-Can, Guaranteed Nameless, Third World Reject”. Did you know that most motherboards from places like tigerdirect.com fail to meet their advertised specs for speed? Oh sure, you can put a 3 GHz processor on them, but their motherboards are so poorly designed that the processor must go into multiple wait states before the address and data buss’ settle during the execution of each op code. Rather than take the exact number of clock cycles advertised by the processor for a standard op code, like perhaps a memory fetch, the system takes a few more. So the computer operates slower, but the speed reported in the “System Information” window still says 3 GHz. Of course you would never realize these defects if you didn’t run known benchmark software. I test every system I buy with benchmarks that I wrote myself in C++. Each benchmark is designed to test a different aspect of the system. I am very impressed with the performance of my XPS. Thanks, but no thanks to tigerdirect.com.

Of course prices and system capabilities change almost weekly in the computer market. I paid extra for an E8500 processor. I notice that I can buy that processor from tigerdirect.com today for just under $300, which sounds a bit less than I remember. So, if I bought the same system today it might cost $1800 instead of $1900? Or, perhaps even much less. I haven't returned to the DELL site to check. I bought the computer 6 months ago. At the time it was in my opinion very price competitive considering its performance capability and robust construction.

But the point of my post was to compare two recordings. You didn’t even comment. Of course the Ivory samples are larger than the resident samples on my piano, and Ivory supports dynamic sampling, which means the computer must keep more recorded notes in active memory. My Akoustik Piano software also has large samples and also supports 16 level dynamic sampling. The computer must keep in memory 16 separate recordings for each key. My system memory (RAM) is 4 GBytes in order to handle such software.

But what difference does it make to the sound of an average beginner's playing? Both recordings are very high quality. What you hear on your end (please use studio headphones) is what I hear when I play. Plus I am using 4 three-way speakers, two per stereo channel, and each speaker has a 12 inch woofer. Yes, the actual sound in the room sounds like an acoustic.

Again, I ask you to be the judge.

Orez Eno, which 64bit OS are you using?

Kind regards,
James
x

IMO, after going soft piano, there is no going back to the original sound, though the investment cost is much larger (PC, software and sound system).

Hi JAWAI James,
Thank you for your response. I’m using XP Pro, Version 5.1, SP-3. Please advise if you feel I should be using a different operating system.

I chose that operating system because it was recommended by Native Instruments, the company that publishes the Akoustik Piano software. I had already decided on that software before I purchased the computer, and luckily at the time DELL was still building some machines (just the high end ones) with XP Pro. After installing and using Akoustik Piano, I suspect that Ivory would have been a better choice, but not for the quality of the sampled sounds, which is what everyone raves about. My immediate complaint was that Akoustik Piano did not support partial pedaling (half pedaling). My piano does, and I got used to it. Akoustik Piano doesn’t and I noticed it immediately. But that is a whole other discussion.

I have played various recordings of myself to enough unbiased people to know that the tonal difference between the Akoustik Piano virtual piano system and the native sounds of a CLP-230 are almost imperceptible, at least for a beginner's playing like mine. And, the CLP-230 does not even support dynamic sampling. I have also had several people listen to my playing live with the same results. When I play live the virtual piano software outputs to a $300.00 PreSonus FireBox external audio interface unit, which feeds a HiFi system with four three-way speaker units (two per channel), each unit having a 12 inch woofer. Most people are amazed at the quality of the sound. But they notice little difference between the virtual piano and the resident sound of the CLP-230. The difference is noticeable, but so slight that it hardly seems worth it.

When I first set up the virtual piano system I was posting often here at PianoWorld and I had promised to submit a complete technical review of my experience. But I never did because I knew everyone would disagree with my perception of it. So I didn’t bother. I decided to post less often and spend more time practicing piano.

I intend to submit in the next PianoWorld recital. Perhaps I will submit an Akoustik Piano recording rather than a Yamaha Clavinova CLP-230 native sound recording. I haven’t really decided. But I do admit that in the last month I have not even fired up the virtual piano. It does not even seem worth switching the cables around for.

I keep my computer in a configuration that runs Finale notation software into the piano, as well as into my other instrument, a Yamaha YPG-625, with very good results. In that case the computer is controlling the instrument, rather than the other way around when it is set up as virtual piano. That explains why I have to switch a few cables.

One thing I am disappointed about is that none of the software that I use is compiled to take advantage of the dual core processor. I use the System Task Manager and see that only one core is active. In contrast, if I execute a system operation, like searching all files for a test phrase, both cores are used. My understanding is that that is because the operating system was compiled to take advantage of both cores. But software like Finale and Akoustik Piano are not.

Indeed, I just upgraded my Finale software to the latest 2009 version expecting it to use both cores, but it still uses only one core. This is frustrating because I know from being a programmer that all they need to do is select a different option in the compiler when they produce their executable. My experience is that most companies know very little about compilers like Microsoft’s Visual Studio. They just accept its default configurations. Often, they do not even upgrade it. I wouldn’t be surprised if the company “MakeMusic”, who publish Finale, are still using Visual Studio version 6. It is silly that they are using only half my computer. When I ask Finale to play back a score the software pushes that one core to its limit while the other is idle. The Task Manager reports 50% CPU use. I even have to shut down my internet/network connection and virus protection to give the software the most CPU power as possible, otherwise I hear occasional spikes in the sound. With the network shut off the software still uses one core full bore, but there are no spikes. Well, maybe in the next release they will upgrade their compiler? I hope so.

Orez Eno, am I correct in assuming that you are using the standard 32-bit version of Windows XP? If so, the following information (quoted from Wikipedia) regarding Windows XP x64 Edition may be of interest to you:

As piano sample libraries continue to grow in size, the 3.2 GB memory barrier of 32-bit operating systems will become more of a factor, thus I expect that the major virtual piano software developers will provide native x64 versions of their software in the near future - if they have not done so already.

Finally, I believe it is a little unfair to assume that professional computer programmers do not wish their software to perform as efficiently as possible. Without viewing the program source code, it is very difficult to predict the optimisations that can be achieved through adjust compiler settings, however it is almost certainly true that such alterations may lead to hardware incompatibilities for certain customers.

On a separate point, I would be rather interested to test your benchmarking software - may I ask if it is fully 64-bit aware?

Kind regards,
James
x

Hi Kawai James,

My system has 4 GB of physical memory and it is all visible. I found out about the typical 32 bit memory limitation after I had ordered the system and I was worried that I might have made a configuration mistake. But when the system arrived the full installed memory was visible. I guess that is the advantage of purchasing a high end system directly from DELL. They wouldn’t allow the customer to configure a system in such a manner unless it really worked, whereas a local computer store, like CompUSA or Circuit City would just throw it together and ship it. Those consumer grade computers are typically build by 18 year old kids who work part time and who don’t really understand what they are doing. Not that I would understand either. These configuration issues change so often it’s hard to keep up with them. Also, when I specified the system I could have chosen much more than 4 GB of physical memory if I had wanted to. I stopped at 4 GB because at the time I suspected that higher than that was a foolish expenditure. When I retired from my last job we had servers that did not have half as much memory. But at the rate that the footprint of software is increasing, perhaps it would not have been so foolish after all. Well anyway, the amount I have does seem to be plenty at this time.

I still wonder if it is possible that my operating system is not really 64 bits but can still access more than 3.2 GB. After all, the 64 bits refers to the data buss, not the address buss. Although typical 32 bit systems have limited memory capability, there is no reason theoretically why one cannot be redesigned to have a wider address buss (assuming the processor can support it), in which case it could use more physical memory. Or am I wrong about that? System design has changed so much since my time perhaps I am too much out of touch to know what I am talking about.

Sorry, but my benchmarks were compiled for 32 bits using version 6 of Visual Studio, which is what my company was using at the time I retired. They will run on a 64 bit system if in fact my system truly is 64 bits like I think it is (so I guess they are 64 bit aware if that is all that the term means), but they would be more informative recompiled for specifically 64 bit systems. I did upgrade my compiler, but I never did any serious work with it, like I thought I would. I found out that retirement caused me to lose my competitive edge in programming very quickly. And now, so much time has passed that it would be too difficult for me to install the compiler on my 64 bit system and recompile the source code of my benchmarks. Still, I find the benchmarks very useful for making comparison tests between systems. When I got my system I made comparisons with similar clock speed systems at various computer stores and my DELL XPS beat them all out. I was very happy.

My opinion that many programmers are unfamiliar with their compilers comes from first hand experience, and I'm just as guilty as anyone. In addition, I worked on many systems that would not compile on new releases of Visual Studio. For example, my company had several projects that were stuck in Visual Studio version 5 and could not be upgraded. It was always caused by the way the source code was written. This happened most often with large pieces of software that had teams of software writers who didn’t talk much to each other. It is especially typical of Windows Win32 API or MFC projects that also use a lot of generic standard library underlying code. Programmers typically don’t understand the intricacies of connecting underlying generic C++ code to the hundreds of system functions that these Windows APIs provide, particularly those of the user interface. Some projects used GUI wrapper software that the company paid thousands of dollars for, but these were even more problematic than allowing the programmers to use Win32 directly, and they were especially difficult to upgrade to new versions of Visual Studio. My comments may have sounded a bit disrespectful, but they weren’t intended to be. It’s just the way it is.

My benchmarks are not publishable pieces of software. Like most home made utilities, their interfaces are not clean or professional looking. But they do provide accurate timing for various standard operations, like drawing random lines or filled rectangles to test the video processor, or calculating fractals to test the floating point processor. If you give me your address I’ll mail them to you on a CD, but I’m sure you will find them a bit dated. For example, I do not provide for the user to enter the number of filled rectangles that the program will draw. I believe it is fixed in code to be 100,000 rectangles. With today’s clock speeds I wish I had designed them to draw 1,000,000. You might think it would be simple to modify them, but it’s been so long since I touched the source code, and also since I used the compiler, that I don’t feel motivated enough. I prefer to spend my time practicing piano, which for me is a new hobby that I started after retirement.

The reason I found my benchmarks useful is because they contained such a limited number of operations. I knew exactly what the programs did and how they did it. Running professional benchmarks, like Wheatstone (I hope I spelled that correctly - it’s been a long time) involve such wide varieties of operations that it is difficult to evaluate different aspects of a system, or even to compare different systems against each other.

I hope all of my rambling has been useful to you. It wasn't my intention to talk so much about computers. I really wanted people to listen to my recordings to help them realize the difference in sound between a virtual (software) piano and the resident sound of a standard model, which is what this thread is supposed to be all about.

Once again, send me your address and I’ll dig out a few of my benchmarks and put them on a CD for you.

Orez Eno, thank you for your informative post - you make some very good points there.

With regards to how much memory the computer 'sees', I believe that the amount displayed by the BIOS may differ to that displayed in Windows XP (the 'System' applet, for example), however as my laptop is only equipped with 2 GB of memory, it is impossible for me to confirm this.

Regarding your benchmark tools, while I appreciate the offer of receiving a CD, I do not wish to trouble you with the postage cost etc. However, if you are willing to make the files available online (I can provide means if necessary), I would gladly test the programmes.

Kind regards,
James
x

Thank you for your reply. However, do my recordings reflect your experience?

If they do, you have a better ear than myself and most of my friends.

If not, perhaps my recordings are incorrectly done? What do you think?

Orez - our professional background seems to be similar. I'm also an electrical engineer, doing a lot of Visual C++ programming these days, but plenty of embedded projects sprinkled in as well.

I'm afraid we'll have to agree to disagree as far as Dell/tigerdirect. From my experience you'll pay much more for the same hardware from Dell. You are of course getting something in exchange for the extra money - like a peace of mind (warranty), and no headaches when it comes to setup. In my case, the extra cost is not justified. I enjoy playing with the hardware, and being able to customize everything that goes in, and I'd much rather spend the extra money on upgrades.

But going back to the original question. I'm at work right now, and I doubt my tiny speakers will do your recordings justice. I'll listen to your files tonight on my home piano setup and give my opinion then. Better yet - can you post the midi file so I can sequence it through my Ivory setup and add another file to comparison.

By the way I noticed you said you have to switch cables around - that definitely can be a pain and take away from your enjoyment of your virtual piano. Perhaps you can modify your setup so nothing needs to be switched around? It's definitely worth it, even if it means you have to build a custom box. All I need to do is turn my PC on and within 1 minute Ivory pops up and I'm ready to go, without even turning my monitor on. I also run Finale (2008), so I guess the fact that you have 2 pianos hooked up makes it an issue of sorts?

Orez Eno
I liked both samples very much but I could definitely tell a difference in favor of the software piano, richer and less brittle in sound. I've been playing acoustic piano for over 70 years so I think I have pretty good auditory discrimination. But your digital piano sound is also very good. By the way, I think both samples sounded better than most of the acoustic pianos I've had. I hope my new Yamaha CP300 will sound as good.
Bob Dettloff

By the way Orez Eno, you are a pretty darn good "beginning" pianist.

Bob

I certainly would NOT want to discourage anyone from an activity that they enjoy. Myself I haven’t had the best of luck building systems. And, I guess you can sense that I am more anxious to actually use the system than to build it. Even my programming experience has always been at a level far removed from the hardware or system configuration level. Another problem I have, which you have probably read in one of my previous posts, is that since retirement I have lost a lot of my technical edge. I wrote those C++ benchmarks many years ago when I was a competitive programmer and knew what I was doing. It’s easy for me to run them on new systems that I buy out of curiosity. But I could never write those programs from scratch again, or even modify them for that matter. Perhaps it’s best that I let other, more skilled people build my systems. The truth is, I suspect that some of my beliefs about my system are actually not what I think. But I’m not too concerned. My hobby now is piano, not computer science. And the system does a pretty good job in my opinion. I try not to let computer work interfere too much with my practice time.

Sorry, I don’t have any MIDI files. The links I provided are for mp3 audio recordings. My Yamaha CLP-230 will allow me to copy MIDI files from the computer into piano memory and play them from the instrument, but the recording features of the piano do not allow me to transfer the recording in MIDI format to the computer. I don’t have access to the section of memory where the piano’s recordings are saved. And, I suspect that the recordings would be in a proprietary format anyway. I believe the CLP-240 has that feature of exporting recordings in MIDI format, but the CLP-230 does not. I have noticed that the new CLP-330 also has the feature. But my piano does not. So after working on a performance and getting a recording that I consider acceptable, I must then record the audio output of the piano using Audacity software.

In the case of my Akoustik Piano software, from what I can tell they also save recordings in a proprietary format. The idea is the same as MIDI in the sense that they are small instruction files. But they don’t seem to be in MIDI format. The recorded files have an extension called .APS, and the files cannot be played using Media Player, which can indeed play MIDI files. I have tried loading them into Media Player, but the software complains that it doesn’t recognize the format and cannot play them. So I assume the files are not MIDI. The software does allow recordings to be exported to .WAV format, which I have used. I then use a utility to convert them into mp3. But I haven’t noticed any provision for exporting to MIDI format from the Akoustik Piano software. The software will load and play MIDI files, but it seems it will not save my performances in MIDI format. Perhaps I don’t know what I am doing here, but I’m not too concerned because I am primarily interested in the audio recording anyway, not a MIDI recording.

As far as cable switching is concerned, yes I can set it up so that all cables are connected to run the system in any configuration. But as often happens with audio systems, if I do that I get a lot of noise pick up. For example, the digital MIDI signal between the piano and the computer can be heard in the HiFi set. I also get noise pickup due to ground loops between the various interconnected units. You may have noticed that the noise performance (hum and noise) of my recordings is very good. To achieve that I must disconnect all stray, unused cables from the system.

Sounds like you have your Ivory system working pretty good. I realize that I accidentally chose the wrong piano software. At the time people here at PianoWorld said it was the absolute best. Now it seems popular opinion has switched its allegiance to Ivory. I will probably eventually get Ivory.

My point, and I think you are confirming it, is that both recordings are high enough in tonal quality that either is acceptable for a beginner’s playing. I say “for a beginner” because I suspect that an accomplished pianist would be able to draw greater difference between the two systems.

You can see that my entire experience setting up a virtual piano has been a bit of a disappointment. Yes there are differences, but they tend to be exaggerated by pointing out technical differences, like sampling rate and levels of dynamic sampling, rather than comparing the actual sound. I don't regret purchasing a rather expensive, high end computer. I need that anyway. And I may in the future go back to using the virtual system, perhaps with different softare, like Ivory. But for at least another year I feel I should concentrate more on practicing and building playing skill, rather than building computer systems.

Naturally, one person’s ear might be better than mine. I wouldn’t fault anyone for setting up a system themselves and raving about it, if the raves are really authentic. In my case I openly admit that Akoustik Piano software is not worth it. My CLP-230 is fine enough.

I appreciate very much your kind assessment. I bought my piano three years ago but suffered from severe pains in my hands and arms. This was a big surprise to me. I had to give it up completely for five months, and then reintroduce myself to it in only very small increments. If I played more than three days a week the pains would return. For another year I played only occasionally and did not progress at all. I was beginning to think that I had wasted my time trying to play piano. I even took up guitar for a short while. I also did a little strength training with weights and squeeze exercisers. Then about a year ago I began to realize some significant improvement in strength and stamina. I was both surprised and pleased. These days I can practice three hours at a stretch. I’m so happy I stuck with it. I am also intrigued at how the human body can adapt if you are patient enough to give it enough time.

When people ask how long I have been playing I answer one year because the first two years were not really about learning to play piano. They were all about strength training.

Don't feel bad about Akoustik. I've bought Ivory almost 2 years ago, and the opinions were split down the middle. In fact Akoustik has one advantage over Ivory - with use of Kontakt and some script files you can set it up to reproduce sympathetic resonance. That's something that Ivory can't do, and it can add a lot to realism to the sound.

I got Ivory because of a few reviews that claimed it's more playable than Akoustik - but quite frankly you can find people claiming otherwise. There isn't much difference between the two if you ask me (and in fact Akoustik may have an edge due to the resonance feature I mentioned).

Also - the pedal on Ivory plain sucks. I usually just hold it throughout the whole piece - it acts more like a reverb effect, than a real pedal (I use Pianoteq for pedal practice).

Let me ask you about your setup - what audio card are you using? One of the main advantages of software pianos is that you can end up with studio quality (noise free) recordings, because obviously only the MIDI signal is traveling over the cables, while the actual sound is being generated/recorded right on the sound-card.

You also should consider recording to MIDI and then rendering. There are so many advantages to that method. You can render to different instruments, adjust gain, sampling rates, filters, fix possible mistakes - everything right on the raw MIDI file stage. And if one day you upgrade to a better software piano, you have your raw MIDIs right there ready for upgrade rendering.

I'm pretty sure you can set up Finale to record your MIDI. I use Sonar Home Studio myself. Here's my Braveheart submission from last year's recital - rendered in Ivory:
http://www.box.net/shared/4b9xp4hkfq
If you want I can post MIDI file for that piece, so that you can compare with Akoustik. Also - I'm a beginner myself, so clearly this should not be the benchmark for Ivory/Akoustik quality.

Eternal:
Wow! That is a fantastic performance. I am downloading the file into my MP3 player, which I call my active listening repertoire. It’s what I listen to every day. Your playing is truly inspiring to a person at my level. Your dynamic expression is impressive. If I could play like that I would consider features like dynamic sampling more important.

My playing is just now reaching a level where I am beginning to accomplish some dynamic expression. But since as you can see that I am still working on Clementi skill building pieces, I have a long way to go.

I hear you about the MIDI files, and yes, using Finale to listen to my playing is an acceptable way of doing it. It would be an advantage today to give you the file and then listen to it rendered on your system. I will look into this. Perhaps on my next recording I will make a point to put it in MIDI format.

You asked about the audio of my system. I’m using a PreSonus Firebox, external audio interface unit. I had the experience some time ago using a different computer where the internal sound card was picking up noticeable EMI (electromagnetically induced) noise from a wide variety of signals within the case, most of which might have been from the switching power supply, but some possibly from the motherboard. I switched to a low cost, USB controlled, external interface unit and the difference was phenomenal. Since then I have aspired to the practice of having external units based on the argument that they can be physically placed sufficiently far away from all those switching signals and possible EMI pickup. On my new computer I didn’t even bother to try the internal sound card. I simply connected an external audio interface unit immediately out of principle. At first I had an E-MU 0202, but just recently I purchased the Firebox. I get noticeably better noise performance with the Firebox, I think partly because of its balanced audio inputs for both stereo channels. On the E-MU 0202 only one channel was balanced and the other channel had some barely perceptible noise. With the Firebox both channels are completely clean as far as I can tell.

Note that the recordings that I posted earlier were made using the older E-MU 0202.

I’m glad you told me that the pedal performance of Ivory is not very good. You are the first to admit that to me. That’s a bit of a turn off for me since my CLP-230 has such excellent half pedal performance, which I have grown so used to. I have to admit, I recently played on the new CLP-330 and I am sufficiently impressed that perhaps I will remain with resident instrument sounds rather than virtual, software sounds. I won’t really be upgrading to the CLP-330. Although it is noticeably better than my 230, it is not sufficiently better to warrant an upgrade at this time. But perhaps in a few years, when my playing is sufficiently skilled to take better advantage of dynamic sampling, I could upgrade to something like a CLP-370 or even CLP-470 if it’s available at that time. I could also choose equivalents from other manufacturers. At least by purchasing a complete instrument I can be sure that the pedals will work correctly.

Thanks for the compliment Orez - the Braveheart piece is not that hard but sounds great, even in inexperienced hands (I've been self taught for 1 year when I learned it). I just listened to your two files, and they do sound great. I did a blind test on myself and I could tell which one was done by Akoustik (slightly richer sound) - but I have to admit the difference wasn't what I would expect. It's also possible some of the extra frequency information from Akoustik recording was attenuated by mp3 compression.

But at the same time - you are starting off with a much better instrument. I am playing on a $500 Casio Privia - so listening to Ivory was like night and day - I remember I started spending much more time at the piano after I got it.

It's funny you mention external cards, because I have the exact opposite bias. I hate any piece of external hardware. I view it as another communication layer in-between, where things can bottleneck or go wrong, because good driver programmers are hard to find. Things did change from the days where USB drivers were just plain horrible, so maybe I should start looking into those solutions again.

Your point about the advantage of audio interface separation from the rest of the system is well taken. Luckily I haven't ran into those problems just yet (even with my current Shuttle system where things are really crammed. I just don't like the mess and prefer when things are out of sight. It's bad enough with a dozen audio cables running between the Piano, Speakers, Headphones and PC.

Do record a MIDI version of your piece, and we'll see if there's a significant improvement in Ivory.