Voice Dropping During Offline Render (NPPE + BBCSO + Superior Drummer 3)

While exporting / offline rendering, I have voices dropping from BBCSO and Superior Drummer 3. They drop during playback as well, but that is sort of expected behavior… I have never experienced this issue before in other DAW’s – usually during offline rendering, the plugins seem to have as much time as they want to index samples, so I’m not even sure how to start with this issue. Recommendations would be appreciated.

-Using audio interface’s ASIO, will try ASIO4ALL as well with its larger max buffer (2048 vs 1024 on current).
-44.1 kHz
-The relevant plugin settings are set a little high (voice limits / memory / etc.) but again, this is not usually an issue during offline renders.
-i7/128GB/7Gb/s SSD – system should not be limiting factor.
~15 instrument score with drumset, maybe 2 voices max per staff at one time.

For the most cases you are right, but it also depends on the implementation of the individual plug-in. The host program (Dorico’s audio engine) tells the plug-ins via a flag whether it is in offline mode, but it is up to the plug-in to react accordingly by e.g. dropping the maximum voice limitation.
What you also have to consider, how the plug-ins make sounds. For example HALion uses - depending on the sound - many layers of voices to create one sound. So even if you just play one note, the plug-in might use up already 5 or even more voices.
Using a different driver and bigger buffer sizes is also a good idea.

Thanks Ulf, I will keep playing around to try and fix it. If I could get your opinion – if I am exporting a group of tracks (parts) from a full score, only running one VST at a time would probably be an easy way to fix this issue if I can’t find a better one. Could I just export from each part layout at a time? What would be the best way to break up the score and export one instrument at a time leaving all other VST’s inactive? And is there a way to make a track’s VST “inactive” like in other DAW’s without removing the VST from the track?

I’m not 100% sure what you are after, but in the Audio Export dialog you can select the option to create separate audio files for each player. Isn’t it that what you are after?

Sorry, I will try to explain:

If I have a project with 10 players / instruments, and I were to delete 9 of them, then export audio, that would mean the VST libraries wasn’t trying to run all 10 original instruments at once. It would only have the input from one that I didn’t delete. This would significantly lower the load on the CPU and probably make things manageable.

Then I could go back to the original project, delete a different 9 instruments to render the next - repeat until all 10 instruments had been rendered individually from separate “subprojects” cut out from the original score. I suppose I could also just remove VST’s from tracks instead of deleting them, but this can get confusing as they are automatically reapplied when the project is opened by the playback template, and because of how NP is set up, I can only deactivate an instance, not an instrument.

Is there a better way to do this in Dorico?

If they are all being exported at once to separate tracks, I seem to get the same voice dropping that I get from real time playback, so I am trying to find a way to render one instrument at a time.

I need to check again, but I’m pretty sure that Dorico (or it’s audio engine) will render each player one after the other.

Hi @Ardemez , I checked again and indeed, if you use during audio export the option Export players as separate files then each player gets rendered one after another. You can even see it yourself, if you browse to the export directory, you will see one file popping up after the other.

Thanks Ulf, I’ll keep working on it and post a solution when I find one.