I am attempting to run my old Spectral Streams set from last year (new gigs ahoy!) and have run into one problem at least
On my DAW PC, I am running Cantabile 4220 and I found that Cherry Audio CR-78, which I use in one song is no longer working well. The beat is all over the place when host sync is selected, but fine when the tempo is free running.
CR-78 is still running fine in host sync mode on my GIGPC which is still running 4187, and it is running fine in host sync mode in Cubase 14
Take a listen to the attached Zipped Wav. It starts off in free running tempo, and I think you will notice when I flick it to host sync mode.
CR-78 has recently been updated to 1.0.11 Build 32 on both machines, but it is playing nicely in Cantabile 4187 and Cubase 14, but not Cantabile 4220.
Is there anything you want me to try or get additional info? Do you want me to try an earlier version of Cantabile on my DAWPC?
I can’t think of anything off hand that’s changed in this area for quite some time (but I could be forgetting something).
The main setting that might affect things is the audio buffer size - is that the same on both machines?
Failing that, the quickest way to narrow this down might be to try 4187 on the machine that’s misbehaving. If that solves it, then some divide and conquer to determine which build introduces the problem would be extremely useful info.
Buffer size on my GIG PC is 304ms (this is a far less capable PC than the DAWPC, which is a bit of a monster for recording), and on my DAWPC it is 224ms. I tried upping it and it made no difference. I also tried turning double buffering off and also resetting the engine. Again no difference.
I went back to 4187 on my DAWPC and it all seemed fine, went back to 4220 and had the same problem, although whereas last night it seemed persistent, this time I had to do things like swapping songs or going into options and back to the song to stimulate the problem. But maybe I just failed to spot that last night.
Strangely, the CR-78 in a test song of its own seems fine even in 4220 when I have the problem in the song I noticed the issue in, which is quite a “busy” song in terms of plugins. So I did try turning plugins off, but can’t say I noticed a change.
I then went for the good old “binary chop” of trying different versions and I would say that I could not reproduce the problem up to 4212, but it appears to be there in all the versions I tried above 4213 (4213, 4216, 4217, 4218, 4219 and 4220), albeit I had to flick between songs or go into the help/about menu and back to stimulate the issue
Is there anything else to try. Would it help for you to have a copy of the song?
On the plus side, whilst I was waiting for Cantabile to install or fire up and load the set, I programmed two new patches into my nice new Moog Muse that Santa left behind! (Mid life crisis number 7, I think, on hitting the big six zero!)
I’ve had a look at the changes around 4212/4213 and there were changes related to how the master transport is executed. Off hand I can’t see anything obviously wrong but will take a deeper dive on it today.
Quick question: what are you using as the master transport - metronome, media file or external MIDI clock?
If you’ve got a moment, could I get you to test one more thing…
In 4220 (or any build where the issue happening), what happens if you set the audio thread count to 1 (Tools → Options → Audio Engine → Number of Audio Threads → 1).
I’ll check tonight when back from work. CR-78 is VST3 as I’m trying to move away from VST2. For the past couple of years I’ve only installed new VSTs as VST3 versions
Re master transport, I think it’s Metronome but I’ll have to check
Number of Audio Threads is by default set to Automatic (14)
And I can easily get the glitching still. If I set Number of Audio Threads to 1 or 2 it seems fine. Anything higher gets it glitching again.
Master Transport is as suspected Metronome.
HTH. Let me know if you need more.
FWIW I am not in a huge rush, as these gigs will not be until later in the year the first in August, but I was running through the old set, as (a) it has been far to long even for a semi improvised set(!) and (b) Imaginando have released a new version of Visual Synthesizer (VS) so I was doing compatibility checks with it against my Cantabile set list (and having to make changes to the VS songs, as Imaginando are great on new features but not so hot on backwards compatibility!)
Before digging deeper on this I want to double check that this is related to a specific change made in that build. For the next build I’ve added a debug flag that will let you toggle between the old and new transport update mechanisms. If you don’t mind, I’ll get you to do some testing to confirm that’s where the problem lies.
Experimental build 4221 should be up by the time you read this. The new diagnostic option is Tools → Options → Diagnostics → “Use old Transport update method”.
NB: Don’t enable this option with a media player as the master transport as it’ll probably hang/crash.
Strange, but it’s also strange that CR-78 on its own in a song on my system is fine. The song in question where I’m hitting this issue is quite a complex one, but as above I’ve narrowed it down that up to a particular Cantabile version, up to which it’s working fine and after that it’s not.
Have you tried flipping between songs and/or options, which definitely makes it go haywire on my system. Also it’s quite a busy song I have so is that something to do with it as well in terms of loading the processing?
“Latent defects” (aka bugs) in software are 100% systematic - hit the trigger conditions and they trigger every single time, and they maybe environment related. So the trick in debugging is to find the trigger conditions. How you’ve tested, like my CR-78 test song for example, with no issues means the trigger conditions have not been met.
I’m currently trying to diagnose a MIDI SYSEX problem a user is having with one of my Java synth librarians that I am not seeing over here on my systems. Such is the fun of remote problem diagnosis! And you sometimes get into test builds like Brad is doing for me.
It is definitely related to that change. As with the debug mode (old transport) set, it seems fine.
See the following video for an example. Obviously sound is just from the iPhone, so not perfect.
I start with the 4221 debug flag off, and you can hear from the off that the CR-78 beat is “interesting” (Jazz drumming?)
I then stop and put the debug mode on, and restart and it is fine (for this test I am starting on Scene 3 where the CR-78 comes in, the song usually builds up though the earlier scenes, controlled from the Montage Scene select buttons).
I then play through a little (still trying to remember what I played a year ago!) just to give it a run and ensure it is stable with other things going on. The MuLabs Looper is providing the sequence data to the VSTs that I then play over.
Audio load is about 50%, and it is quite a busy song with a UVI Falcon Multi, Korg Wavestate Native, CA GX80 and Zebra as VSTs, with Yamaha Montage M, Korg Kronos and ASM Hydrasynth also being played.
Hopefully you can make out it’s meant to be a Vangelis homage - that Lydian mode is the secret sauce he used as well as the sounds!
