Helpful ReplySpectrasonics Trilian - freeze track doesn't sound the same

I'm just getting my feet wet with Trilian and haven't bothered to RTFM yet.  I only had 2GB ram so I figured that was my problem - but I've just upped my system to 8gb which is better overall but this particular problem is still the same as before.
 
Windows 7 x64 Sonar 8.5.2 64bit - double precision - ASIO - Core 2 Quad 3Ghz - 8gb ram
 
Here's what I'm doing - if someone can reproduce it or tell me what I'm doing wrong:
I'm at 44.1/24    I've tried this both by creating a simple instrument track - and also by creating a separate audio track with trilian and a midi track that points to it. Same problem either way. I recorded about 32 bars of MIDI notes.  Everything plays back perfectly and sounds right. I'm loading only one instrument - the Chapman Stick - Full range - Trilian setup reports that it's using 149MB of ram - but windows shows 1.6GB of ram used when it finishes loading up the samples (it showed 500mb used prior to loading the Chapman Stick sounds) I should RTFM so I know what the memory stat means here.  I don't see any disk access when playing - at least not now that I've increased the RAM - before, with only 2gb RAM it seemed to access the disk when I was playing.  Anyway, it sounds perfect when I play back the unfrozen track.
 
When I try to freeze the synth - it creates an audio track that doesn't sound right at all.  The notes don't sustain - they cutoff - it sounds like a random place but it isn't (I mean there are some notes that do sustain and some that don't but if I unfreeze and freeze it again it will always be the same notes that cutoff).  I also hear glitches that I don't hear if I unfreeze the thing. 
 
The first hiccup happens in the first few bars so I've saved a really short midi file - and it always hiccups in exactly the same place in exactly the same way every time I freeze it.  If someone wants to test it out - I can send you the midi file.  Don't know why it would respond differently when freezing it than it does when playing from midi.  
 
This is where it would be really handy if I could just create an audio track and point its input to Trilian's output and then record it live - but when I do that, the audio track doesn't have a record option.  Do I really have to physically connect cables from the output to an input in order to record the unfrozen playback?
 
This has me feeling really stoopid right now - hope someone can point me in the right direction. 
 

Right click on the freeze icon on the instrument track and deselect the fast bounce option and see if that helps when you freeze?

Thanks for the quick reply.   I haven't tested your suggestion yet - but something I just tried a minute ago worked.  I turned off the Read Automation and that resolved it.  I don't even know what "read automation" is or how it applies to a midi track - I've never given it any thought before but I just noticed that the button was selected by default within the Trilian plugin.
 
 

And I just tested again without disabling read automation and using your suggestion.  That worked too ... but it sure took a long time to bounce 32 bars of midi to an audio track.

andypanda


And I just tested again without disabling read automation and using your suggestion.  That worked too ... but it sure took a long time to bounce 32 bars of midi to an audio track.

Yeah, Trilian seems to have issues with Sonar when freezing.  I turn off "Fast Bounce" in the freeze options just for that synth (then turn it back on for all others) and that works for me.
 
I hope it's addressed -- but I'm not sure who's fault it is (though I suspect Trilian in this case).
 
Love that synth though!

Yeah ... I love it too.   I barely installed it and haven't come up for air since (other than to run to the store for more RAM).  I have been using the Chapman Stick patch and it's bringing out elements of my playing that I never knew I had.
 
Wait until I have time to try out the upright bass patch - probably won't leave the house for a week.
 

Did you send an e-mail to Spectrasonics technical support about this issue? Might not be a bad idea. Just wondering, man.
I've gotta try out the Stick some more. It sounds like you're having a blast with that thing!!! I know what you mean about the uprights too. I haven't tried them yet as I have just been doing stuff with the electrics since I got Trilian. I cannot believe what this thing has done for some of my recordings thus far.

Hi,

Glad I came across this thread. I've been having the exact same problem with Omnisphere! I'm not at home right now, but when I get home I'm going to try it, my hunch is that if I do what you suggested that freezing the track will work properly. My guess is that there is probably a lot of information there to freeze as omnisphere is a pretty resource intensive synth a typical fast freeze doesn't capture everything. Anyways... if I have more problems I"ll definitely repost

From the Spectrasonics site:
 

Bouncing a streaming instrument to disk is always tricky business. Basically, streaming relies on the fact that there is some spare time available when rendering to read from disk. However, when bouncing, hosts often render as fast as possible, eliminating that spare time which is important for streaming. This can lead to audio artifacts. So, when bouncing or freezing any instrument that uses streaming, it's important to bounce in real time.

Note that it is frequently possible to get a faster-than-real-time bounce of a streaming instrument to succeed, but it is not reliable. When bouncing a streaming instrument, it is important to do your bounces is real time in order to obtain reliable results.

Hosts usually have an option for real time when bouncing or freezing. If your host is lacking that feature, you can usually do a real time bounce by simply bussing the output of the instrument to an audio track, playing the sequence and recording the results to the audio track.

 

I have trilian now...I know I need more ram now too...and i need way more time to play around....got it friday and have only scratched the surface.

BTW....other soft synths have the same issue with fast bounce.  BFD2 truncates cymbals in fast bounce.  Seems any huge sample with a long decay will get cut off.  So is it an issue with Sonar or BFD2 or Spectrasonics???  Only the bakers know !!

Earlier I posted that I tested, on a short - 32 bar midi pattern, by simply disabling read automation and that seemed to solve the problem.   However I just tried it on a complete bass track (3.5 minutes) and although it worked (didn't crash sonar - the sustained notes weren't truncated) - I did hear a couple of glitches.   So I tried it again without the fast bounce - took a really long time (so I opened a beer and had a snack) and it worked perfectly.

So ... Spectrasonics mentions the issue is because it is a "streaming" instrument -- does that mean not all the samples are loaded into memory and some are read off the hard drive as needed?.  Before I started this thread, I tried disabling streaming in Trilian (though this was also before I upped my installed RAM) and the problem was still there.


 

Good question....BFD2 is a disk streaming and RAM application as well. So that may be the deal...only whatever is in RAM gets bounced properly, and additional layers that disk stream are cut off...

However, the export/bounce should not take longer than the song...(ie) if I bounce a 4 minute song it usually takes about 4 minutes to export...r u saying it takes a lot longer to bounce than to play the song normal time?

Any rate, I have not fast bounced in years becuse I always seemed to get truncated cymbals from BFD and weird artifacts from some master buss plugs.

Any rate, I have not fast bounced in years becuse I always seemed to get truncated cymbals from BFD and weird artifacts from some master buss plugs.

 
I used to have that problem with BFD and BFD2 but their updates seemed to solve that for me, at least.
 

It doesn't take longer than the song - it just seemed like a long time since I've gotten spoiled by fast bouce taking seconds.    Now you have me wondering how many tracks I've fast bounced and cut off cymbals without noticing.   I may have to do the slow bounce from now on.  
 
Actually I hardly ever use softsynths in the past because I like my hardware sounds so much.  B4 has really been the only softsynth I've needed to bounce - I've always gotten great sounding bass, rhodes and even guitars and drums from my Motif.  Never heard anything in Dimension Pro, Kontakt, Garritan etc that I preferred over my hardware.  I do like Ivory - but I hardly ever use it in a mix.
 
But now that I'm getting my feet wet with Trilian and Superior Drummer (and starting to wonder about Omnisphere now) - I'm sure I'll be using freeze a lot.

Ba_midi I will check fast bounce out with BFD2 later thanks.

However, I will not check anything out until my new RAM stick arrives.  (Arrives on Friday, so hopefully will be able to use all the full size trilian samples instead of just the lite versions)

Andy -

I am just starting to use more and more softsynths - BFD2, Trilian, Jamstix3, EZ Drummer, Dimension Pro, Rapture etc etc...they are all getting so good sounding, it is hard to resist.

But I still think there is something flakey in Sonar's internal midi routing for soft synths that I cannot quite put my finger on....but I am continuing to narrow it down I think.

JR

It makes sense that streaming samplers might have issues with fast bounces. I'm actually surprised that such issues don't come up more often.

A streaming sampler (which includes most samplers nowadays) only pre-loads a portion of the sample into memory and then sucks the remainder off disk during playback. This scheme works because the file system is faster than the playback, IOW you can acquire the data faster than it's being played back.

But a fast bounce doesn't wait around for anything, and if the disk drive can't keep up and the sample player doesn't tell the host "wait a sec, I'm still reading the disk", you could get things like cut off tails on cymbal hits.

Curiously, I have never encountered this problem myself, despite using software drums and occasionally samplers such as Kontakt. That may be because I don't use a lot of samplers - often only one - or it may be that some samplers are able to send the "wait a sec" message to the host while others cannot.

Some samplers, such as Kontakt, let you tweak the RAM-to-streaming ratio. Maybe Trillian offers something like that, which would be in TFM.