ChristopherM

Sorry if this is too vague (and in any event you have sorted your problem). I have had similar issues in the past when disk space on my Windows SSD was getting low and Windows automatically and silently changed certain folders to "compressed". It turned out that Cakewalk could not read the compressed folders and assumed that the contents had vanished. Solved in my case by manually unsetting the "compressed" switch on the folders in Windows. Solved more effectively by upsizing my SSD, eventually.

So that the "unwanted" plugs are excluded from scans (and hence do not delay matters if they have any issues, say) I move their DLL into a folder that is not in the scan list. This means that if I ever want them back permanently or temporarily, it is easy to drag'n'drop them back again. Whether this is better or worse than using the Exclude feature of PiM, I do not know.

Using MOscillator to get a reliable 440Hz signal and NeuralDSPs plugins side-by-side, I have been able to demonstrate to NeuralDSP support that the issue is real and is confined to the Time Henson plug. The tuner reports 404.2 Hz. Note that 44.1/48 x 440 = 404.25. They wanted me to try switching back and forth between 44.1 and 48 sample rates to see how that affected this bad behaviour. There is no documented way to do this in Cakewalk AFAIK, so I switched to Cantabile. Switching from 48 to 44.1 then back to 48 solves the problem. NeuralDSP supports neither Cakewalk nor Cantabile, but they did admit that they have seen this issue in VST3 version only in Cubase, which they do support. So they have informally accepted that this is a bug. Just for kicks, I tried altering Cakewalk's Default [sampling rate] for New Projects setting to 44.1 and back to 48. To my surprise this immediately fixed the issue in the running project. I say surprise, because I had always understood that the sample rate is fixed in Cakewalk at the instant the new project is created, but apparently not. Funny how you find things out!

When using this plugin in Cakewalk, there appears to be a problem if the sample rate is not set to 44.1 kHz. I normally use 48 kHz, in which case the tuner in the plugin works incorrectly. More importantly, the Multivoicer (a harmoniser, only available in the Tim Henson plugin, AFAIK) does not work properly either. Changing the sample rate to 44.1 kHz in Cakewalk corrects both issues. This does not happen with other NeuralDSP plugins that I have (Cory Wong, Plini, Gojira); in those, the tuner works correctly whatever sample rate is set in Cakewalk. In stand-alone Time Henson, the sample rate can be changed in its own settings page, and the tuner continues to work correctly (as does the Multivoicer). I have contacted NeuralDSP support, but I'm anticipating that, as Cakewalk is not one of their supported DAWs, they will blame Cakewalk. (Hope I'm wrong, but this seems to be the way it goes). The fact that the problem is not present in the other similar plugins strongly implicates this plugin, IMHO. Anybody else using this plugin in Cakewalk? Am I missing something, or is this a bug? Compliments of the season!

@Jonathan Sasor - thank you for your engagement with this. I have PM'd (in three parts) the relevant files. Edit - I have also added a link to an (18GB) dmp obtained by procdump64.exe during the memory-hogging phase.

Well, I'm happy to report some positive progress here, at last. I can load the AAS instruments now, quickly and successfully with no sign of the memory-hogging. The key was resetting my entire AUD.INI to default (something I did after AAS's tech guy asked me if I had altered anything from default, as he could not replicate the behaviour). It occurred to me that, as I have used Cakewalk for many years, the init file has presumably evolved along the way. The first use after this was not encouraging. When I attempted to load Chromaphone 3 VST 3 by dragging and dropping within Cakewalk, the entire Cakewalk GUI vanished instantly. Cakewalk.exe continued to appear in Resource Monitor, but with zero CPU and occupying its normal amount of RAM (i.e. the previous RAM-gobbling behaviour was not present). After some time, Cakewalk.exe silently exited, the GUI never having reappeared. To cut a long story short, each time I restarted PC and/or Cakewalk then attempted to load the AAS instruments, I got a little further (although with two different Unhandled Exceptions along the way). I now seem to be able to load these instruments successfully. However, I'm very puzzled as to what the initial issue was and also why these following instabilities happened, but are seemingly not reproducible at will (even though I have made no further changes to AUD.INI). I'd be interested to hear @Noel Borthwick's thoughts on any of this. I have the original aud.ini and the mini-dumps from the UEs for anybody sufficiently curious to want to examine them. All observations gratefully received!

Aah! That's interesting. I've finally received a list of diagnostic questions from AAS, so, with your permission, I'll pass your experience to them, too. But the difference is that you are getting beyond the point where my OS gives up and actually having the instruments become usable.

AAS at an earlier point did say something along the lines of "You're using Cakewalk? That'll be the problem, then!" for which the author subsequently apologised (sort of) when I asked him for some evidence. As for piano tuning, ours had reached a remarkably stable point and it just stayed in tune for what seemed like forever. We eventually decided we needed to let a piano tuner loose on it, but it only stayed in tune for a matter of weeks after that. So we found another guy, who came armed with some impressive technology (I can't remember whether it was on a laptop or a tablet, though) which let me decide the amount of stretch that I wanted in the tuning, f'rinstance. He did a great job, which has lasted more than a year so far. He was an excellent player, too, and gave us an impressive recital before he rendered his invoice! And in the UK, he presumably is not prevented from making house calls at the moment (although I personally would probably not invite him in just now). And I must say that, were I an oboe player, and I got my new instrument out of its case, at which point the it expanded entirely to fill the room. but refused to make even a squeak, I'd be just a tad annoyed.

Thanks for replying - I'm encouraged that you are running successfully, but I don't select the NKI option from AAS.

@Starship Krupa - I just read your brilliant piece about the evolution of VST in another thread. Funny and insightful, like all the best satire!

Thanks for your observations, Starship. Resource Monitor shows cakewalk.exe as the process gobbling up memory and it just keeps going until the entire OS malfunctions (something that doesn't seem to happen much in Windows 10). I have tried these plugs in Cantabile (which does nothing other than handle plugins, of course, so it seemed like a good test bed) and they are slow to load, but not massively, and then they work fine. And the standalones work fine, too. That would seem to implicate CbB, but, as I mentioned, Noel Borthwick seemed certain that the issue was AAS's after he had looked at my memory dump, and I've never known him to be wrong! And all of this happens long before the instrument is called upon to generate audio, so I don't feel it's the complexity of the sound generator. What I noticed in Cantabile is that the instrument is loaded and capable of sound generation long before its GUI appears, which is what made me wonder about graphics issues. I used AAS's combined installer, which selects the revised sound packs along with the instrument, so I don't think sound pack incompatibility can be the cause. It sometimes feels that the technology now gets in the way of creative music making rather than aiding it.

16GB of RAM and Quad core cpu. Standalone, Chromaphone 3 uses less than a GB of RAM. In CbB (for me anyway) it filled up all RAM and then created a 32GB swap file before crashing. So I don't think its resources as such. I have wondered whether it's because I use the CPU graphics rather than a separates graphics card, but that has not been an issue with anything else.

Thanks - I'm pleased they are working for you, but that thickens the plot, of course. I also have previously had good experience with AAS support, but this time Eric keeps telling me he's sorry and that he'll get on to it, but then somehow doesn't. They're a small organisation, I know, but 2 months is a long time to wait. I got notifications about updates, but not to these products - I'll have another look, so thanks for the heads-up.

I have always liked AAS products (going right back to Tassman that was included with some earlier version of Sonar many years ago). At the end of November 2020, I upgraded to the new generation Chromaphone 3, Virtual Analog VA-3 and String Studio VS-3. They work fine as stand-alones - and they sound superb - but they are disastrous in CbB. Typically, when I attempt to load any of them, Cakewalk becomes unresponsive and hogs more-and-more memory, until Windows gets into trouble and starts glitching on graphics and eventually the screen blanks. Even when the screen blanks, I can see that there is constant disk activity (presumably Windows furiously writing to Virtual Memory). I then have to restart the PC. Hoping that the new release of CbB might have altered this behaviour, I just tried again - attempting to load String Studio VS-3 caused Cakewalk's memory usage to climb steadily to over 32GB Commit in Windows Resource Monitor before the screen blanked and I had to reboot. I posted about this phenomenon at the time in the CbB Feedback thread (but that version is history now, so a new thread seemed necessary). Noel was kind enough to look at a memory dump and said he thought it was an AAS issue. I had already attempted to get AAS's Tech Support to take an interest, but I have had no meaningful response in 2 months - makes me wonder whether they are in some kind of difficulty. So 2 questions: Is anybody else successfully using any of these version 3 AAS products with CbB? Does anybody have any clue what is going on and whether therefore this is a configuration problem in CbB or Windows 10? I am using the most recent Windows 10, CbB and AAS versions.

Thanks, Noel - if it's any consolation, it also took a while to compress the file! I assume you are OK with my forwarding your message to Eric at AAS. To your question, I was not playing, because the plug had not completed loading. I don't know whether this has any relevance, but when this plugin is loaded, I think I have seen odd behaviour with the "dirty" marker. I think the asterisk appears again immediately after a file save in Cakewalk, although TBH this is just something that I saw out of the corner of my eye and did not really observe in detail. I'll let you know what AAS makes of this.