Ease the hard switching

[RJHollins]

Hi Gang,

I've got a small schematic here that is a MIDI drivin' audio output switcher.

Basically, using a [CC-30] MIDI event to switch between a pair of audio outputs ... this is a utility VST plugin that can be used to bypass another plugin in an FXChain.

The basic idea is working, however ... the switch is rather abrupt [audio wise]. I was thinking of a way to 'ease the transition' to take some of the shock out.

What to use ?

Would something like a 'ramp' or a decay envelope [release] ??

This is a new area for me to work ... hope would you guys best handle something like this.

here's a screen shot

MIDI controlled audio output switcher-1.jpg (65.48 KiB) Viewed 13810 times

and a test schematic:

MIDI controlled audio output switcher-1.fsm

(5.19 KiB) Downloaded 823 times

Thanks for any insights!

[stw]

You can do a crossfade fed by a follower code or if it's not time critical by a e.g. the green slide prim or a ruby code.
Look at the attached fsm.

MIDI controlled audio output switcher-1-fade-stw.fsm

(9.45 KiB) Downloaded 817 times

[RJHollins]

As soon as I saw the name 'stw' ... I knew something cool was coming In fact, just the other nite I was reading across the SM/ FS forums some of your postings. Real good stuff !

I'm just now getting to look at the example you posted ... that X-Fade example is great. I think someone was working up a DJ type app, and this came up.

This is giving me ideas on another project that I'm having to send out a command to do volume changes during a 'patch change' in another VST. Right now I'm using a 'Slide' PRIM to do a type of 'duck' manuver on that ... it works OK ... but I'd like it to be better.

Back to this project ... ok, I have to play around with this X-Fade routine.

I'm working this inside of a 'routing type' plugin called Plogue Biddle. This is a cool app that is helping me to set up some, otherwise, complicated parallel routing. I have several plugs that I'm remotely controlling and I'm using this posted sample as a way to bypass [switch in or out] these various plugins. There is all kinds of crazy latency to deal with in the process, so the audio stream can have some heavy jolts going on.

Thanks stw for looking in and taking the time to share your knowledge ... much appreciated. Now to see how this might play out.

Thanks!

[Nubeat7]

you could also use a de-zipper to slide..

[RJHollins]

you could also use a de-zipper to slide..

Hi Nubeat !

Ahh ... did not know this about the 'de-zip'.

I knew it was used in the VCA module.
In another little utility VST, I use this in a 'remote controlled' Gain adjuster.

So this is great. Several different ways to go about this.

Thank-you Gentlemen ...

[RJHollins]

OK ... just got to test these 'gentler' switching routines ... here's what I find.

1st started with the 'De-Zipper' method [NuBeat] ... when I activated the switch, the audio switched outputs ... however, this left a stream of 'white-ish' noise in the 'abandoned' output. This only occurred when only the 1 pair of OUTPUTS were hooked up [I think it was OK when all 4 outputs were connect ... gee ... now I need to go back and check ! The blast of noise stunned me

went to test the 'follower' method. This too worked, but I noticed an issue. It seems the 1st action CUTS the signal to MAIN outputs as it switches to the AUX outputs. Yet, then switching from the AUX back to the MAINS happens with a nice gentle transition. [I was testing with 250 mSec]. The weird part is that I don't see that behavior in the test module with the numeric display ... also I should note, I had 2 copies of this section in the schematic so as to handle the Left and Right channels.

I've not tried the 'SLIDE' routine as I'm not quite certain how to integrate this in.

Anyway, just wanted to point out some issues ... testing continues!

Thanks

----- EDIT ---

ok ... I see how the SLIDE routine hook-ups ... going to test that.

[RJHollins]

mmm ... just tested the SLIDE method ... seems the same NOISE issue after the switch ... however, the transition was nice.

Went back to the 'follower' routine ... this one stays clean, but [ain't there always] ...

When the switch is made, the MAIN gets hard cut, while the AUX fades in. But when we then go back to the MAINS, the glide in.

IT's tough to test since I'm using the WAV player as source and feeding a stereo pair out. But I can definitely hear that the signal gets a hard cut with the 'other' getting x-faded ... it's the same going in either direction.

Goal ... Fade OUT of Main with a fade in to AUX ... and vis-versa.

also could use some sleep ... that ain't happening

Thanks Guys ... hope you can understand what's happening and help me change this.

[RJHollins]

ahhh ... I found my error Yeah ... surprised me too !

The MIDI switcher is using [CC-30] and is switching ON/OFF with 0 and 127.

The feed into the 'time' module that precedes the 'Follower' wants Boolean.

I switched it ... and, of course, now it X-Fades as designed.

I'm actually very happy ... just disappointed that it took me this long to figure that out.

Off to now test this under actual conditions ... also, I going to check the other routines posted. This is a very nice feature to have, even just in the toolbox.

Thanks again ... off to test.


---- edit---

NuBeat ... the 'De-Zipper' version works as well. As mentioned, there is a slight change in the way it responds [green] ...

[RJHollins]

alrighty then ...

amazing how 'real world' conditions can botch the best intentioned plans

First ... all these X=Fade routines are working just as designed, and under normal conditions would be excellent routines ... however, I seem to find myself more in the 'AB-normal' situation.

As opposed to doing true X-Fading of 2 signals, I'm actually switching a separate VST plugin IN or OUT of the signal chain. Using a hard bypass [normal function] puts the audio chain into temporary chaos as the signal flow is abruptly changed AND things like host 'Delay Compensation' are re-calculating. These are usually good things.

My HOPE ... was to try to audibly minimize the chaos during this switch over. [I should mention that this could involve a 1/2 dozen 'sampler' vst's that have significant latency.

When I did the X-Fades as designed, I was hearing a type of 'Twilight Zone' transition as we moved through the 'Time Continuum'. Could be a neat effect ... would like to save that for maybe some other type of project

So ... need to ponder a different strategy. I still believe it will use some form of X-fading ... but it probably needs to be dealt with specifically transitioning OUT of one signal and then ramping up into the new connection.

Looking at my original 'error' in connecting up the 'Follower' method, I was switching 2 values into the 'time' module. A '0' and then 127. Looking into this 'time' module, I'm trying to understand what the 'switch' value is actually doing to the circuit calculations.

STW ... might you explain this value ... what it does, what kind of range are we talking.

I'm not sure that this would provide a solution ... but it may help point to the right direction.

thanks!

[TrojakEW]

Do you ned "hard switch" for that? What about code/asm multiplexer. This may help your problem when switching fxchain.