Is there an nLite version of Windows XP for this purpose (live playing)? Thanks.

kawaian dont forget that those vmachines are nothing more than laptops with a tiny built in display. they are slower than a good laptop, so they have more dropout problems and they have never solved the licensing issues so they only work properly with public domain software.

it would be nice if it worked for what it is advertised to do but its both cheaper and more reliable to buy a laptop.

if you want reliability over quality of sound, you want to look at things like a yamaha motif rack.

edt, thanks for the advise about not working with commercial VSTs, that was new to me! Then definitely a laptop (either Mac or PC) is the way to go! So disregard the first two links, the only link to check is the last one for live-laptops!

You might want to consider the latest from Muse,

...the MuseBox

It's a smaller version of Receptor. Has all the audio I/O you would need and it will run things like Ivory. I use a Receptor and it's one of the nicer ways to run "big sample" software instuments.

The links are broken. I just sent them an e-mail saying they are. Let's see what happens. Maybe if others also send e-mail they will notice.

Cheers Chris, I'll email them too.

Uh, dude! Re-read.

Wow lots of good comments here. Thanks guys.

Regarding QLP sound and why I picked it - After I did months of listening at pergatorycreek and other OEM demos I firmly believe it is the most deep and beautiful sounding soft tone. But it comes at a cost of 260GB of samples and can sound distant, almost muffled, compared to some others.

For the guy interested in the QLP Yamaha. Unfortunately that instrument is a bit of a joke and the worst of the 4. Yes, Yamaha is supposed to be bright but QLP yamaha is way overboard and sounds like a toy to me. The Ivory Yamaha sounds much better (in fact it sounds great) to me in their demos anyway and would be the main reason I'd want to get Ivory.

I did try the QLP lite versions, player mic only, and even removed some of the sample sets for key off and one other. And I still got the dropouts. All this makes me think it's a bug (either QLP or some driver) more than a hardware performance limitation issue.

I will chime in that I really like Sampletekk from the demos I've heard. Go to their site and listen to the Eleanor Rigby one by Jay Asher- I forget which instrument- it's either Black Grand or Seven Seas. It is Unbelievably Awesome! But unfortunately Sampletekk has no stand alone versions so I won't buy any of them because I'd have to buy another $300+ piece of software just to host it. Otherwise I'd buy their stuff like hot cakes.

Thanks Sullivang for the IOMeter utility tip!!!!!!!!!!!!
I don't want to get my hopes up but that would certainly explain the type of behavior I'm seeing. If I still get dropouts after my re-install I will try that next. I have a 7200 RPM drive so it should be plenty fast but if a weird sleep setting can go on after seconds of inactivity that would certainly bog things down.

I mean a pre-configured nLite package. Something like this

http://forum.cockos.com/showthread.php?t=30162

but stripped down even more. (The one above still has networking enabled, for example).

That's interesting. For the time being I'll have to take your word for it, because you've actually played it. I really LOVE the demo recordings though. I really have to get my act together and install it and the Bechstein.



You're welcome. Mine is also a 7200rpm drive, btw. Note that I assume you have looked at all the other more basic things already. (ensuring unwanted processes, particularly ones that access the disk, have been terminated, for example. I COMPLETELY shut down my antivirus software, for example.)
If you need any help with IOMeter you're welcome to send me a PM. What I do is configure a random, single sector read every couple of seconds or so, to the disk where the samples reside. If this turns out to cure your problem, you/we should inform East West, because it may speed up a fix! It's possible that this should somehow be addressed at the system level - not by East West. Perhaps VST hosts should address this, for example. Maybe the best advice is just to use a performance drive to begin with - that's what most people do.

Greg.

I have Ivory & still find the Yamaha very sterile-sounding, compressed, and bright. The reason why it's so popular is that in a rock band's mix, that piano is perfect! On its own, I find it brittle & frankly useless to me.
Please make sure you have the latest soundcard & ASIO drivers, as well as the latest patch for QLP & the PLAY engine.

Beyond that, most (not all, but most) laptops will have hard drives spinning at 4300 RPM as opposed to the desktop standard of 7200 RPM. Couple that with the fact that you have your operating system & all the applications that run in the background installed on the same hard drive and it quickly becomes a slippery slope. The easiest way to bypass this issue entirely is to have a 7200 RPM external hard drive plugged in a USB 2.0 port. And the good news is that it's a very cheap solution, too.

This is all stuff you can do yourself & frankly, downloading this stuff is illegal... just as it would be illegal to discuss & link to places where one could download pirated versions of QLP or Ivory.

So for the sake of this msg board, I'd suggest we stop this line of discussion right now.

Feel free to grab nLite for yourself & strip down your official version of Windows at will.

PS : As a sign of respect to PW, I invite you to edit your post & remove that link.

I think there are problems in the EWQL memory management. Once the OS X inactive memory fills up completely after playing for a while it begins to page out to disc (which causes drop outs and makes it unplayable) instead of reclaiming the inactive memory. It's easy to watch this happen in the OS X Activity Monitor and hear the results as it happens.

Another possible reason for your "dropouts" could be if your keyboard has a proportional damper pedal. I needed to filter out all damper pedal midi messages except 0 or 127, (and not allow repeated 127's or 0's either from the proportional pedal), otherwise I had problems in the EWQL Play engine with intermediate values.


My only interest in virtual pianos is LIVE play. I agree that the EWQL samples are very nice and have more character than the Garritan I suggested earlier. But I thought your original question was about LIVE playability. The Garritan Basic plays LIVE beautifully. On the other hand, I've spent many hours remapping the EWQL Steinway samples and adjusting individual sample volumes to improve consistency (piano regulation - speaking acoustically), and in my opinion, it still isn't as playable as the Garritan Basic without any "fixes". But that's just my opinion. (Incidentally, I'm waiting for the next version of the full Garritan before purchasing it, because I'm told it does not yet have important improvements that were already made to the Basic version.)

Macy,
How the heck are you doing all those adjustments in EWQLP? Do you have a beta of the pro version, or something? Please explain!!

Greg.

No Pro version yet unfortunately. Just remapping samples manually to improve timbre consistency, which is an enormous amount of work. Plus real-time software conversion per note, per velocity, of midi velocity values. Plus real-time conversion of midi CC messages for damper pedal messages.

Thanks. Have you written this up in detail anywhere?

The single most bothersome thing with the Steinway that I've encountered so far is that some of the bass notes are too subdued.

Greg.

There is a thread on Music Player forum concerning EWQL PLAY engine (not just the Pianos, but a number of different sample sets)
http://forums.musicplayer.com/ubbthreads.php/topics/2169923/1/PLAY_performance

From the thread, it appears that BOTH PC and Mac users are having the OP's problem of dropouts (this includes myself - and my system is way over the minimum requirements to run Gold Pianos).

I don't think it is an OS related issue - it appears to happen on both Macs and PCs. Shame, I really like the Bosendofer and Steinway (although the Yamaha does not favorably impress me. The Bechstein is OK.

BTW - as far as antivirus updates are concerned, if for some reason the system owner considers it necessary to be connected part of the time to the Internet, just disconnect the network cable when using the computer to play EWQL. My business is computer networking - just about all the anti-virus programs that I have seen detect rather quickly when not connected to the outside world, and most of them stop at that point (though they do check again after some time).

The RECOMMENDED SYSTEM is Mac-Pro Quad-Core Xeon 2.5GHz, 4GB RAM. Apple sells it as an "entry level" MP but in the PC world that would be called a server class machine. That's a $3,500 system, more if you add up all the stuff that goes with it.

The guys having trouble are running on notebooks and antique G5 systems. It looks like people are just buying software that is bigger than their computers and hoping that it will just somehow work.

But this seems like a hardware issue, the little Intel dual core CPU is just not enough. Looks to be the best solution is wait until this Tuesday when Apple is rumored to announce new six and twelve core xeon based Mac Pros. Then buy one of those. Or move to a smaller sample set.

The "little" 2.4GHz Core 2 Duo is working just fine for me with EWQLP, notwithstanding the issue I mentioned for which I have a workaround for. In fact, I even get the same polyphony if I underclock my processor to just 790MHz. (verified by running a benchmark and also using the Intel Processor ID utility). My disk is the bottleneck at the moment. My worst case polyphony is about 50 voices. By worst case, I mean performing a gliss with the sustain pedal after a reboot of the machine. Rebooting clears the file cache, so that no samples have been cached. (the system caches the samples as you play, making it difficult do do worst case polyphony testing) 50 may not sound much by today's standards, but that's 50 stereo voices, which is equivalent to 100 in Kontakt. I can use multiple mics, and the convolution ambience processing as well.

Btw, something else to check for on laptops: ensure your processor is LOCKED on maximum frequency. The output from the command window command POWERCFG /Q should show "PROCESSOR THROTTLE (AC) NONE" and "PROCESSOR THROTTLE (DC) NONE". The power management profile I need to select for this is "Always On".

Btw, Pianoteq would be an excellent choice for live playback I think. (I have not used it live though).

Greg.

I would think so too, I don't think Pianoteq has ever crashed on me and since the latest upgrade it will run well on a very low spec machine.

I don't think it's ever crashed for me either. It worked fine on my old 2.4GHz P4, but with limited polyphony. (it was still entirely usable). On my new machine I get the full 256 voices of polyphony.

Back on EWQLP, it's important to set the polyphony to a value that your system can handle. If it's too high, and it runs out of disk bandwidth, many notes will be dropped abruptly. With the polyphony set correctly, the notes that it drops due to note stealing are far less noticable - it is done smoothly.

Greg.

With those extremely demanding software pianos disk speed is always an issue. Therefore I suggested to put the samples on a dedicated USB stick or SSD drive, there is virtually no latency when using them. In case of EWQLP this is of course a problem due to the sheer size. I don't know the exact size, but maybe you can put the specific piano you want to play (lite version is more than appropriate for this, you don't need 24bit samples at all!) on a let's say 32GB USB stick, and there you are. Just try it, USB memory is really cheap nowadays!

I've just looked up the specs of a 32GB FLASH stick, and it's read transfer rate is only 10MB/s, AND that would be for sequential access. (yes, I realise their latency would be low, though) This isn't good enough for EWQLP because it stores it's samples in uncompressed form (as far as I can tell). For 100 stereo voices, we need about 26MB/s. http://www.kingston.com/flash/dt100.asp

Maybe there are better ones though?

At the moment I'm leaning towards the SSD idea, even though it's more expensive than hard disks.

Greg.

Ah: http://www.kingston.com/flash/dt200.asp

20MB/s read. This is more like it. I'm curious to know what the actual throughput would be for EWQLP. I had no idea FLASH sticks were available in such large sizes now.

Greg.

There are better ones: eSATA that have up to 32MB/s, and internal Solid State Discs that have much more speed than standard 3,5" harddrives, but they're expensive.

But anyway I still think that 10MB/s should be sufficient:

100 stereo voices are: 176KBit (for one stereo voice = 44,1KHz * 16bit * 2) * 100, that's ~17MB/s.

There is hardly any situation where you would really need to read all of these 100 voices in parallel: Many of the frequently used voices are in cache memory anyway, so you just need additional voices. But, and that's often the problem, you need them immediately! You press a key -> you need the sound NOW! And here's the big advantage of flash memory: There is virtually no access time to the data, whereas the harddrive has a (for this case) very long access time (typically 8ms). And for worse: On a harddrive you only have one reading head that is responsible to get all the data. If you now want to load e.g. 25 voices at once, you have actually 25 * 8ms for getting access to the data alone (this is a little bit oversimplified because there are hopefully intelligent cache mechanisms of the sampled piano engine, but you get an idea). So actually, if the harddrive would have a specification of let's say 50MB/s read transfer rate, this transfer rate is never achieved, because the data cannot be read sequentially.

These are problems flash memory does not have, so it should really deliver better performance than a harddisc in this specific area. Writing data is a complete different story, but we're talking about read only here...

Yes, agreed/understood - I'm using worst case figures. (which is probably advisable for live playing, though) I'm taking the view that if we're going to go to the trouble of buying storage specifically for samples, we want to get stuff that can easily cope with any situation we may encounter in a live situation.

I'm not aware of any 16-bit samples in EWQLP though. (they're all 24 bits, aren't they?) That's why I used 24-bits for my calculation. I realise we're not just talking about EWQLP though.
The "lite" versions in EWQLP use less samples, and employ sample stretching.

Greg.

Look how easily 200 voices is reached in EWQLP:
http://www.youtube.com/watch?v=uIveoBxnQZY
(go to time 4:40)

We can say that all 200 probably won't be streaming from the disk simultaneously, but it's hard to be sure. ;^)

Btw, the disk i/o meter gives completely wrong readings on my system.

Greg.

I have seen an EWQLP "gold" version, which consists of the identical samples but only in 16bit resolution, and I thought these were also included in the regular release. My mistake. Anyway, I still would think that 10MB/s would be also sufficient for 24bit samples.

Maybe there is someone with EWQLP who has crackling issues and has a 32GB stick at hand and therefore can try this without any risk?

I had no idea Gold was only 16-bit - thanks for the info.

Greg.

Greg, there is no way that a harddisc can handle reading 200 samples simultaneously. These are scattered all over the harddisc, with the before mentioned seektime for each location this is simply not possible to do without any breaks. So there is for sure a mechanism that has the majority of the most frequently used samples in cache.