External Insert Latency compensation, help

[micv]

Thought I have this figured out before but I still don't understand.

I send a track out through external insert to route to hardware devices (compressor, EQ, etc.).  Using external insert ping to get the latency, checked  use asio reported latency (under preference).  Hit bounce to track (real-time).  The bounced track is behind the original track (delay) by about 38 samples. 

Using positive value offset in the external insert has no effect.  Using negative value for offset delays the signal further.  Manual offset in the Asio has no impact.

How to get the bounced signal to line up with the original signal?

Any guidance is appreciated.

 

[micv]

After a few days really digging and searched on the old Sonar forum into this 'hardware latency compensation' topic I think I got a pretty good understanding of how it's supposed to work.  Let me share:

There are two area to compensate for HW latency, 1) is the Asio driver Manual Offset, and 2) external insert. 

I was confused about the two compensation areas above but now I know that they are not related and should be used for completely different scenario:

1.  Record Latency Adjustment (Asio driver):  This is needed so any new recorded track will line-up with existing vst or audio tracks.

     'Use Asio reported latency' is your hardware reported value.  If you don't check this, your recording will be way off.  Even if you have this checked, your recording will still be a little off depends on the HW.  The issue here is most HW have hidden or unreported value.  You must use do a 'loop back ping' to your HW to get the actual latency.  You can use a tool like CEntrance, or CbB's Ext Insert ping (more later).  Subtract the Asio reported from the actual pinged value.  It should be a positive value (under compensated).  Put this value in the 'Manual Offset'.  This is a static value for any buffer settings so just need to set it once for a HW.

2.  External Insert:  settings in the Record Latency Adjustment will have no impact.  (the CbB's bakers can verify but this is what I observed and t makes sense for overall compensation to follow EXT Insert if you use it).

'Ping' to get the actual HW latency.  However the issue is the pinged value returns by Ext Insert is one buffer less than the actual latency.  Not sure if it's intended or a bug?  The good thing is regardless of the reported value, CbB compensates correctly.

For my scenario, I want to bounce to track.  There is a consistency issue when 'bounce to track'.  If I just select the clips in the track and bounce to track it will be in sync most of the time but not always, (bug?).  What seems consistent to me is select the whole track (click on the track number) and do a bounce to track. 

 

[razor7music]

You've gotten much farther than I ever got. I tried years ago when I still had external hardware and never got it to sync. It sort of forced me to do everything in the box, and now I don't have any external HW except for a mic pre.

Maybe someone else can comment, but I wanted to share that your inconsistent results with the latency are not unique.

Edited March 4, 2019 by razor7music

[jimmy_p]

There was a known bug when I contacted Cakewalk about it when X1 was released, if I remember right it would give different ping values each time it was clicked.

Just tried now pinging through my desk and back in. Gave a consistent 16ms reported so it seems the old bug has gone, at least in what's displayed.

Can't find the bounce to track option to try replicate your scenario!

[micv]

12 hours ago, jimmy_p said:

Can't find the bounce to track option to try replicate your scenario!

It's in the tab 'Tracks' on the menu line with MIDI and Region FX

[jimmy_p]

9 hours ago, micv said:

It's in the tab 'Tracks' on the menu line with MIDI and Region FX

Got it thanks, bit different to Sonar 7 I've switched from!

So I've just had a small section of audio in one track cut into small clips, external insert plugin going to the desk then back into soundcard. Nothing actually patched into the desk insert but its doing DA AD round trip. Did this 10 times and every time its been bang on to the sample. 5 times I selected the track by clicking the track number, 5 times by dragging across the track to select all clips. Was this what you meant by selecting the clips?

Now this is a very basic setup with nothing else going on, but it seems to work for me, definitely leagues better than Sonar 7 and even X1 when I demoed it! How often is it going wrong, and when it does can you work out how many samples it is off?

For my setup I haven't changed the audio settings yet, so using ASIO reported latency of 832 samples, no manual offset. Oddly my ping time through external insert is less than the ASIO reported latency - 708 samples where you say it should be bigger. Could this be to do with the buffer not being reported that you mentioned?

[micv]

14 hours ago, jimmy_p said:

5 times by dragging across the track to select all clips. Was this what you meant by selecting the clips?

yes.  The first time I encountered this it wouldn't sync no matter what.  But on the next day once I got it to sync once, I can only reproduced it in a couple times out of twenty or so.  So I'm just sticking to selecting the track method and call it a day.

 

14 hours ago, jimmy_p said:

For my setup I haven't changed the audio settings yet, so using ASIO reported latency of 832 samples, no manual offset. Oddly my ping time through external insert is less than the ASIO reported latency - 708 samples where you say it should be bigger. Could this be to do with the buffer not being reported that you mentioned?

Exactly.  For example if your asio buffer setting is 128 samples, you would add 128 to the ext insert pinged value of 708 samples (=836 samples).  This is the actual latency of your setup.  And you now notice that the asio reported doesn't matched (lower/under reported) by 4 samples (836-832).  Put 4 in the manual offset, for a perfect sync.  Now anything around a few samples off would be considered as sync, and you wouldn't be able to notice.  However if in this case if you use 256 for buffer, you will be off by 260 sample which will be  quite noticeable.