Bitwig pitch bend issue (was “How do I perform smooth glides?”)

Official support for: rogerlinndesign.com
RELATED
PRODUCTS

Post

Roger_Linn wrote:Unfortunately I can’t hear any pitch stepping
If you listen really closely comparing the two sweeps, I’m sure you’ll be able to hear the difference.

I agree that it’s almost imperceptible, but there is very minor pitch stepping in the second sweep. I don’t consider this an issue at all as it’s not going to be noticeable in a performance context, in a mix, and especially not with more complex sounds.
Roger_Linn wrote:or irrregularity in the video, nor can I see any physical stepping or irregularity in the video image.
Thanks. Good to know my LinnStrument is working as it should and that I’m not doing it completely wrong. :)
Roger_Linn wrote:So I’m afraid I don’t yet understand the musical problem you are experiencing, and music after all is the goal of a musical instrument.
As I mentioned before, I am not experiencing a problem while performing.

I am however experiencing a problem while using LinnStrument as a controller to record MIDI Note and Pitch Bend messages into Bitwig.

As it turns out, the almost imperceptible pitch stepping from LinnStrument is exaggerated by the way Bitwig processes Pitch Bend Messages. If I play back the recorded MIDI data from the performance you’re hearing in the video at https://vimeo.com/268595517 (the recorded mod-wheel sweep followed by the performed LinnStrument sweep), then I end up with what you can hear in https://vimeo.com/268633691 (where both the mod-wheel sweep and the LinnStrument sweep are played back).

I hope you agree you can clearly hear pitch stepping in the second sweep in https://vimeo.com/268633691, right?
Roger_Linn wrote:One thing you might try is to turn off Quantize Hold or to change its rate. Quantize Hold is in Per-Split Settings under Pitch/X.
The video was recorded with Quantize Hold disabled.
Roger_Linn wrote:You mentioned above “I’m just trying to find out where the constraints are.” If you are specifically searching for LinnStrument’s limitations, I am confident that you will find others more significant than this. For example, you will find some good ones on the LinnStrument Specs and FAQ pages under the heading “What are the limitations of LinnStrument's sensor technology?”
I was aware of the limitations listed there before I placed my order.

Again, I am confident that this is a Bitwig problem. Even though the way they deal with Pitch Bend messages probably works fine for mod wheels, and probably also for MPE controllers such as the Haken Continuum and the flat areas above and below the waves on the Roli Seaboard, Bitwig’s handling of MIDI Pitch Bend messages doesn’t work well for these kind of slides on LinnStrument.

They can probably fix that, either by allowing the option to disable their “thinning” feature as shown in https://vimeo.com/268444348, or by making it smooth out stepping across longer sweeps in a more natural way.

I’ll write a support request for the Bitwig people. Thanks for your help and patience. :)

Post

Given that your earlier pitch-slide video was played live and not played back from a sequence, and that this is a LinnStrument forum and not a Bitwig forum, I have apparently misunderstood that you were pointing out an issue with LinnStrument as a separate issue that was in addition to the one you raised earlier about how Bitwig reduces continuous data. Silly me.

I'm glad there is no issue with LinnStrument and will consider this issue closed.

Post

BobDog wrote:I would try another DAW and see if the problem goes away then we know if it is Bitwig causing the issue.
There’s no problem with for example Reaper because doesn’t do touch the incoming MIDI data at all:
Screen Shot 2018-05-08 at 20.14.26.png
You do not have the required permissions to view the files attached to this post.

Post

Roger_Linn wrote:Given that your earlier pitch-slide video was played live and not played back from a sequence, and that this is a LinnStrument forum and not a Bitwig forum, I have apparently misunderstood that you were pointing out an issue with LinnStrument as a separate issue that was in addition to the one you raised earlier about how Bitwig reduces continuous data. Silly me.
Sorry for all the confusion. I started this thread assuming the issue was primarily with the way I was playing slides.

I am now convinced Bitwig is doing things to the Pitch Bend Messages it gets from LinnStrument that it shouldn’t be doing.

Post

You don't have to be convinced between the video I referenced proves it: Bitwig thins out continuous data, which it should be doing because that's how they designed it.

Post

Oh my, I’m terribly sorry, but it looks like it’s going to get even more complex…

In Bitwig, I was using the “Roger Linn Design / LinnStrument” controller script that was automatically detected. It turns out that the problem is related to this script.

As a test, I added the “Generic / Midi Keyboard” controller script manually for my LinnStrument as well:
Screen Shot 2018-05-08 at 20.31.18.png
After this, I set up two tracks in Bitwig, one for each of these controller inputs. Then I recorded the same slide into both tracks:
Screen Shot 2018-05-08 at 20.41.34.png
Here’s the same thing with the tracks slightly offset in time so that they’re easier to compare:
Screen Shot 2018-05-08 at 20.32.32.png
The blue track was recorded through the “Roger Linn Design / LinnStrument” controller script, the red track through the “Generic / Midi Keyboard” controller script.

As you can clearly see, the stepping is somehow introduced by the “Roger Linn Design / LinnStrument” controller script. Using the “Generic / Midi Keyboard” controller script, the curve much smoother.
You do not have the required permissions to view the files attached to this post.

Post

I'm sure Bitwig will be very interested to learn this when you post it on their forum.

Post

I think it is nothing to do with and inherent data thinning problem in Bitwig because both me and tvdv have the same version of bitwig on the same version of osx and pitchbend here after thinning works fine.

I think the problem is the data arriving in Bitwig, maybe something weird going on with the midi side of things, almost like it is getting two sets of pitchbend messages offset in time. A good video would be to see the pitchbend info as it is being recorded, does that show these jumps?
Bitwig, against the constitution.

Post

whoops, missed those posts!
Bitwig, against the constitution.

Post

Thats weird, I am using the Linnstrument scrip and don't see that issue.

On the blue line you can actually create quite a nice curve using the top or bottom points of the vertical pairs, it reminds me of the awful AU automation recording in Ableton Live, it looks pretty much like that:
screenshot_498.png
You do not have the required permissions to view the files attached to this post.
Bitwig, against the constitution.

Post

It helps a little to disable the pitch quantization on Linnstrument. I can hear pitch stepping when performing slow slides with pitch quantization on. When off, it's smoother. You'll loose those nice and in-tune tones but it lends some 'accoustic instrument' playing style quality and challenges you to learn to land the slides in center of the pad. Disabling pitch quantization is not suitable for all sounds (piano, rhodes etc.) but it helps for some accoustic/real instrument emulations or abstract cinematic sound design.

Post

Roger_Linn wrote:I'm sure Bitwig will be very interested to learn this when you post it on their forum.
For those interested; please see http://www.kvraudio.com/forum/viewtopic ... 7#p7074007

Post

For me both the Linnstrument controller and the generic one give exactly the same smooth curves, so I am guessing Bitwig will not be able to replicate this. Also they don't read the KVR forum much so it is worth contacting them and pointing them to your post.

The Linnstrument controller script is very simple, it does not process midi messages in any way so should not have any impact on pitchbend recording.

Can you go to Settings->Controllers then right click on The Linnstrument controller and choose "Reveal File". You should see an Explorer/Finder window appear.

Zip up the LinnStrument.control.js file and attach it to a post, then I can compare it to what I have here.
Bitwig, against the constitution.

Post

BobDog wrote:For me both the Linnstrument controller and the generic one give exactly the same smooth curves, so I am guessing Bitwig will not be able to replicate this. Also they don't read the KVR forum much so it is worth contacting them and pointing them to your post
I’ve sent a detailled bug report to the their technical support team before I posted on the Bitwig forum.
BobDog wrote:The Linnstrument controller script is very simple, it does not process midi messages in any way so should not have any impact on pitchbend recording.
I looked into the controller script yesterday. The only thing in which their appears to be functional difference with the generic keyboard script is the setUseExpressiveMidi call (https://gist.github.com/tvandervossen/b ... rol-js-L13). When I remove that line, the script does record smooth pitch bend for me.
BobDog wrote:Zip up the LinnStrument.control.js file and attach it to a post, then I can compare it to what I have here.
You can find the script I have at https://gist.github.com/tvandervossen/b ... f346a45d88

It’s just what came with the latest version of Bitwig, so I would be surprised in case yours is different. What kind of Mac are you running Bitwig on? Maybe there’s some weird interaction between the USB 3.0 bus on my MBP and their “MPE mode” for controllers that gets enabled with the setUseExpressiveMidi call.

Maybe we should continue the discussion on the Bitwig forum, by the way.
Last edited by tvdv on Wed May 09, 2018 6:41 pm, edited 1 time in total.

Post

Sorry, duplicate post.
Last edited by tvdv on Wed May 09, 2018 6:38 pm, edited 1 time in total.

Post Reply

Return to “Roger Linn Design”