transient marker sdif export

schwarz · November 20, 2014, 12:51am

Hi, is there a reason why in transient marker sdif export the end marker is always a few milliseconds behind the begin marker (see below)?
This confuses simpler programs (like cataRT) that only look for 1MRK frames.
Of course one could filter out the frames with the end markers, but a fix or an option to export only the begin markers could ease the workflow.
(The “embed markers into soundfile” option works also very nicely, thanks for that!)

1MRK    1       0       0.663973  
  1BEG  0x0104  1       1  
        1  

1MRK    2       0       0.6772  
  XTRD  0x0004  1       2  
        0.974148        0.0103719  
  1END  0x0104  1       1  
        1  

1MRK    1       0       1.67326  
  1BEG  0x0104  1       1  
        3  

1MRK    2       0       1.6869  
  XTRD  0x0004  1       2  
        0.405777        0.00298994  
  1END  0x0104  1       1  
        3

picasso · November 20, 2014, 11:17am

Hi Diemo,
Yes, the duration between the start and the end marker is usually quite short. And those values are correct since they are reported by the SuperVP analysis and read back in AS. So for me the End time is perfectly valid.

What may be confusing is that AudioSculpt merges both ends (and store the start/end time) in a single visible marker. So you might naturally expect to export only this visible part… but i guess the extended informations format in SDIF requires to split the information in a 1BEG and the 1END frame separately.

As an other export solution, in order to filter out the 1END information, you can use a the TEXT export option
Menu > File > Export Analysis As Text > Export Markers As
then select Transient Markers.
It will output an ASCII file that won’t contain the end time info.
exemple:

  
0.000000	0	TrsM	  
0.030950	0	TrsM	  
0.155791	0	TrsM	  
0.435374	0	TrsM	  
0.462178	0	TrsM	  
…etc…

Does Catart need only SDIF format ?
If yes, then i can add an option to export only the start time in the SDIF format.

Tell me what would be the best option for you.

Best,
Charles

schwarz · December 8, 2014, 6:45pm

Hi Charles,

thanks for the explanation! I makes sense now that the segment contains just the transient part.

This hints at distinguishing two types of data to be exported from transient analysis:

the transient *segments* as done now, and
the transient *marker* only as 1MRK/XTRD, in the form of

1MRK    1       0       0.6772  
  XTRD  0x0004  1       2  
        0.974148        0.0103719

Sure, for CataRT the text export will do what is needed.

roebel · December 8, 2014, 7:15pm

Hi Diemo,

Attention, in the segment output the XTRD matrix is part of the end frame! This is the end of the transient and somehow the start of the more stable part of the audio event. I would expect you don’t want the end frame for catart, because you would get all transients cut, you most likely want the begin frame which is located in front of the transient?

To Charles: If I remember right n AS we wanted to display the start time of the transient, to be able to apply treatments to individual notes including the transients?

Best
Axel

picasso · December 9, 2014, 10:43am

Hi Axel, Diemo

Yes of course ! For sure in AudioSculpt, we display the start time of the transient, not the end part
As i suggested to Diemo, the ASCII export of the markers only export the start time. So it is a possible solution (if i understood correctly) for CataRT to get the required information without being confused by the other part.

HTH

Best,
Charles

louis.goldford · August 6, 2019, 2:23pm

Hi all —

Without wanting to rehash what has already been typed elsewhere (I don’t see other threads), I’d just like to confirm a few details about the behavior of transient markers in AS.

Specifically, when I run a transient analysis …

and want to move or correct any of these transient “events”, I notice that one of these markers always moves but the other doesn’t. And
if I want to delete one transient “event” I must do it one marker at a time; that is, 1BEG and 1END, but never both at the same time.
No matter how far I zoom in I cannot see these markers separately in the AS timeline; they always appear exactly superimposed on top of one another, and so it’s not clear from the interface which marker is the the 1BEG and which is the 1END.
That also means when I label one of the markers, I never see my label because another marker seems to be superimposed on top of it, and according to AS documentation the label will disappear when two markers are close enough together.

Does all of this seem like normal behavior in AS to you ?

If you need me to provide an example I can do so, as this is easily reproducible on my side. I’m running AS 3.4.6 on Mojave, and my transient analyses tend to include a threshold of 2.0-4.0 and a minimum interval between markers of 20-100ms. I don’t see a difference in this behavior if I change these params.

I’d much rather run and edit a transient analysis in AS instead of OM because the interface seems a bit more friendly, but I haven’t explored all the possibilities for setting this up in MuBu, for example. Perhaps this would work better if there is no immediate solution in AS.

Thanks in advance for any suggestions and wishing you all the best —
Louis

roebel · August 31, 2019, 2:26pm

Transient markers in AS have two time points. The start and the end of the transient.
You see those two in the inspector window.
In the inspector these are time (1BEG) and endtime (1END) entries.
AS always only displays the begin position.

For moving the marker you select it in the GUI with the mouse and move it around.
If you edit the values the marker should move if oyu edit the time entry, but not the endtime entry because that one is not displayed. I advice against editing these values because you will have a hard time keeping 1BEG and 1End times coherent.
YOu select the marker you want to delete and press delete key.
AS only displays the 1BEG part, so you never see two
Labels are attached to the 1BEG marix.IN AS you do this in the inspector, you just enter the label in the label field and hit return, then you see it appear in the spectrogram display.

Best
Axel

louis.goldford · October 15, 2019, 4:07am

Many thanks Axel, and apologies on this late response.

Let me clarify with one point: When I run one of these analyses I discover that not 1 but 2 markers seem to be present in the waveform editor at every point. When I select and move or delete one of these markers, another one remains at exactly the same point.

For this reason it appeared that perhaps the 1BEG and 1END were both present in the editor.

Can you confirm this behavior on your end? I don’t remember this in previous versions of AS (earlier versions did not produce duplicate markers at the same points like this). This seems to work no matter how I set the analysis parameters.

Thanks in advance —
Louis

cornu · October 16, 2019, 2:48pm

After talking with Axel, I think this behavior is because your file is stereo. There is actually one marker (one 1BEG/1END pair) for each channel. And if I enter a label, it doesn’t get displayed, just as you observed.
Unfortunately, I couldn’t find a way to go through all the markers, it seems the inspector will only show the ones for the first channel. So I guess AS is not well suited for editing markers in multi-channel files.
Depending on how you use the edited analyses, there might be workarounds

cornu · October 16, 2019, 5:27pm

There is a way to select both markers: in the sonogram area (not the waveform), click before the marker, move beyond it then release the button. You can then move or delete them both as a unit.
I haven’t found a way to get the labels displayed though

louis.goldford · October 30, 2019, 4:54pm

Thank you Fred! And thanks Axel!

Now that you mention it, it’s actually the stereo file that’s causing the problem. Now I see that if I’m viewing 1 of the 2 channels (using the “eye” icon at bottom left of the waveform interface), I “think” I’m looking at markers for just one channel but in fact it’s both.

If I do this with a monaural file, there are for sure no overlapping markers, and I can move them one at a time as I’d expect. The problem is definitely the stereo file, but the interface suggests that by isolating a channel, one can also manipulate markers placed only on that channel. Looks like in fact it’s always displaying the overlapping 1BEG markers for each channel in a multichannel file. Unfortunately I don’t see this documented.

Also, now it’s clear, as Axel wrote, that only the 1BEG markers are being displayed.

I should mention that in both cases (mono or stereo), I cannot click & drag to select multiple markers and move them as a unit, as you’ve described. Is there a modifier key you’re using? I’ve tried clicking and dragging, also click + shift + drag, but ultimately I can only move one marker at a time.

Also I should mention I’m late on these recent replies because somehow my notification preferences in the new discussion forum were reset! Sorry about this; I’ve just reset them again and I should for sure be getting emails whenever someone responds on this thread.

Many thanks to you all! —
Louis

cornu · October 30, 2019, 6:06pm

I found three ways to select multiple markers:
One is to click on one then shift+click the others. The advantage is that they don’t have to be next to each other. The drawback is that it only selects one of each in the stereo case.
Another is to select several using the arrow tool: click and drag over the markers, the triangle at the bottom will turn red, it should look like this:

The third is to use the “lasso” tool from the toolbar:

Click and drag over the markers to select them, it should look like this:

Once the marker are selected, click and drag one and all selected markers will move along with it. Both the arrow and lasso methods will move all markers in the stereo case

louis.goldford · October 30, 2019, 11:36pm

Thanks once again Fred! This is great…

Okay, now I see you already mentioned doing this not in the waveform display but in the sonogram display. The arrow and lasso tools behave exactly as you describe there, and I overlooked your specification.

Anyway though, this works much better now. I really appreciate your time and hope you are well! Thank you!

Louis~