Melodyne does not erase deleted clips.

[Glenn Stanton]

28 minutes ago, RexRed said:

There is one lingering problem that I detail in this video.

yeah, that definitely seems like a bug -- the take position data seems ok but it's muted status is ignored so passing both takes... does this also happen if the Melodyne is applied to the main track or only when its in the take segment?

 

[RexRed]

30 minutes ago, Glenn Stanton said:

yeah, that definitely seems like a bug -- the take position data seems ok but it's muted status is ignored so passing both takes... does this also happen if the Melodyne is applied to the main track or only when its in the take segment?

 

It would seem this needs a take over top of a Melodyne instance with Melodyne applied to the take for this bug to appear. The bakers are really smart and they will know exactly how to fix this. ?

I am so excited that this might get fixed soon. I have had to delete many a good take to work around this issue.

I know this probably goes without saying, but please make sure these fixes to Melodyne integration are migrated into the new version of Cakewalk Sonar as well... ?

THANK YOU!

Edited November 4, 2023 by RexRed

[Mark Morgon-Shaw]

1 hour ago, Glenn Stanton said:

yeah, that definitely seems like a bug -- the take position data seems ok but it's muted status is ignored 

Doesn't seem like a bug to me, from my POV this is expected behaviour.  In essence what the OP is requesting is for Cakewalk to instruct Melodyne to "mute" the previous take's blobs which I don't think it can.

 I could be wrong but it seems like all Cake  does is pass audio over for Melodyne to process which it then sends back to the track. Cakewalk doesn't really have any other control over Melodyne such as mute states as far as I can tell.   The previous audio is actually still in there inside Melodyne because the "take lane" is really just a sub-part of the same track. I don't think Melodyne is take lane aware, it just sees all lanes as one track. So as far as it's concerned you are recording over the top of yourself and it's displaying that as expected.  Therefore expected behaviour rather than a bug, just not the behaviour the OP wants.

This is partly a consequence of using Melodyne as a tracking tool in a DAW which now has multiple take lanes in unforseen ways.  Cakewalk never had take lanes for many years after Melodyne came out and it doesn't seem as though this has been considered for a Melodyne based workflow. I'm not even sure it's within the Bakers gift to make it work the way the OP thinks it should.  Whilst we wait to find out  we have a few choices.

1. Do what every other vocal producer I know does and comp your vocal first and apply Melodyne to a consolidated vocal track. It's much easier that way - I don't even get why you would want to use Melodyne on the fly like that anyway ? It's a strange workflow as you're basically working the opposite way round to everyone else by pitch correcting things before deciding if you're keeping them or not.  

2. Use separate tracks instead of take lanes ( like we used to back in the day ) for overdubs

3. Mute the blobs manually in the previous take from within Melodyne before recording the new take then they turn into a ghost note outline. This might alleviate some of the visual issues of overlaying notes. I've not tried it though as I don't work that way. 

4. Delete the old blobs in Melodyne before transferring the new take. You can use the compare switch to go back and forth between track audio and Melodyne audio to check which one you like best.

I may be wrong and maybe the Bakers can make Melodyne put the original take into mute mode automatically ( although you would still have the outline shown above I expect) but  if not I suspect this may not be a quick fix.   

Lastly forget the tech and consider a good vocal coach, I'm no pro level singer either but I went from being pitchy AF to not needing to use Melodyne except for the odd word here and there by having a good teacher and working on my voice. Getting it right at source is far better than trying to fix it afterwards which should always be a last resort to be used sparingly.   

 

Edited November 4, 2023 by Mark Morgon-Shaw

[msmcleod]

@RexRed,  @Noel Borthwick may be able to correct me here, but I'm not sure a complete fix is possible with Melodyne's current ARA interface.

As I see it there are two things we could do here:

1. Automatically bypass the Region FX for that clip when it's muted - you'll still see the blobs, but they'll be shown as muted within Melodyne (they'll have a gray fill); or
2. Remove the Region FX for that clip when it's muted.  The blobs will not be shown, but you'll also lose any edits.

My gut feel tells me option 1 is the correct solution, however I can't find any way in Melodyne to get it to hide bypassed regions within the editor. Without that, you'll still get blobs overlaid on top of each other when you add the new take - although at least you'll kind of  be able to see which ones are muted.

So whilst we could go some way to solving the issue, Celemony would need to implement a "hide bypassed notes/regions" option within Melodyne for it to work how you want.

[RexRed]

2 hours ago, msmcleod said:

@RexRed,  @Noel Borthwick may be able to correct me here, but I'm not sure a complete fix is possible with Melodyne's current ARA interface.

As I see it there are two things we could do here:

1. Automatically bypass the Region FX for that clip when it's muted - you'll still see the blobs, but they'll be shown as muted within Melodyne (they'll have a gray fill); or
2. Remove the Region FX for that clip when it's muted.  The blobs will not be shown, but you'll also lose any edits.

My gut feel tells me option 1 is the correct solution, however I can't find any way in Melodyne to get it to hide bypassed regions within the editor. Without that, you'll still get blobs overlaid on top of each other when you add the new take - although at least you'll kind of  be able to see which ones are muted.

So whilst we could go some way to solving the issue, Celemony would need to implement a "hide bypassed notes/regions" option within Melodyne for it to work how you want.

I do hope there is a more elegant option than the two you have suggested.

Having grayed outlined blobs is kind of ugly (and useless) especially if you have several takes with Melodyne on them.

If grayed out is the only option then that is better than orange blobs all piled on top of each other. 

Edited November 5, 2023 by RexRed

[Glenn Stanton]

since it doesn't happen when you place the region fx on the main track, but does happen on the unmuted and muted take lanes (which is why you see two (or more) unmuted sets of blobs, i doubt this is an ARA issue as much as it's something that wasn't seen in a mono version of Melodyne and rears its head in polyphonic versions. i.e. when the region fx is applied to a take (which physically is a snippet of audio in a long WAV file which the UI displays as a "clip") -- if the information passed to the region fx is not aware of the clip mute/not mute status in that level of operation - so it seems that either Melodyne is choosing to ignore the state, or CW is passing all takes aligned to that clip.

[David Baay]

@msmcleod As mentioned ealrier in the thread, I found that Duplicate Track properly applies the result of automatic slip-editing by Punch Recording with Mute Previous Takes enabled to give the desired result. It seems to me it should be possible to do this transparently within the track. Melodyne already honors manual slip-editing; it's just not getting the message that the clip boundaries have been changed when it's by punch-in.

[Glenn Stanton]

unless i'm mistaken, like a database, new takes for a given clip are appended physically to a given track's wav files, and the "clips"/"takes" are the logical representations - so pointers are needed to create the composite view. 

 

Edited November 5, 2023 by Glenn Stanton

[David Baay]

3 hours ago, Glenn Stanton said:

unless i'm mistaken, like a database, new takes for a given clip are appended physically to a given track's wav files, and the "clips"/"takes" are the logical representations - so pointers are needed to create the composite view. 

When simply recording multiple takes sequentially each clip will get its own dedicated file, except in the case of loop-recording where the clips in multiple lanes will all reference a single file because it's really just one "take".

When you comp, the comp will indeed be based on pointers to the different files.

Not sure what you're getting at, though...?

 

[Glenn Stanton]

just thinking out loud that the region fx state of takes in sequential takes embedded are being handled differently than the composite level. so one short fix would be to apply the region fx to the composite level vs at the take level.

understood separate takes = separate physical file, and comp takes are embedded as appended.

[RexRed]

Please keep talking so this gets fixed. ?

Melodyne retains your edits in a separate channel within itself.

If this was not the case it could not go back to the opened state and completely undo your edits on a  per blob basis.

One would think you could take this edit information and hold it in a buffer and then remove Melodyne from a take and if the take is enabled again reapply the edits to it.  Just a thought. 

[Glenn Stanton]

the Melodyne "separations" are the small blob audio clips it uses to store the audio + the edit metadata. 

[RexRed]

I thought the problem where double blobs would display even when a clip is deleted was fixed but I just found it is still there.

Just had it happen again.

It seems to happen less but each time this happens I have to restart Cakewalk for Melodyne to realize a clip has been deleted.

So in essence we are talking about "two" problems here.

One problem, is in certain instances when a take has been recorded over top of a take and Melodyne applied to the new take and then the old take under it is deleted Melodyne is not told the old clip is gone.

The old clip is deleted from the track but Melodyne is still displaying it on top of the new take with Melodyne applied. A restart is required.

The second problem is when you mute a take under another take, both show up in Melodyne.

This is what has been so confusing, two problems both occurring in the vicinity of each other.   

[RexRed]

Some have taken this opportunity to suggest vocal training and repetitive bouncing and reapplication of Melodyne over and over. Fix a note bounce it and then resample and resample and resample...

Record new takes in another track... 

This is not the way it "should" work. 

One should resample once and only once by adding Melodyne to takes and be able to mute and unmute sections they need. This should all be done in one single track and not multiple tracks with duplicated effects and scrolling all over the place.

We are comping, not putting together a jigsaw puzzle or solving Rubik's cube. Tracking, comping and retracking is already enough of a complex undertaking.

I have tried to bring this point across and some try and explain, "the way it is done". Yes, that is the way I do it also, because when it does not function in a typically user friendly way, I have to resort to the same workarounds you do. How else would I do it?

I have no magic way of  doing this any other way than you all do. 

I am just saying we should not have to workaround this problem, it should be tailored to workaround us.

If this is a problem with the ARA 2 specification then the ARA 2 specification needs to be updated. And bring us different colored blobs per track while you are at it. Tracks should pass on color information to Melodyne. Orange blobs all superimposed are ugly and impossible to distinguish. Yes, I can work around and make some tracks gray, but color would be better. Then I would not have to click 15 tracks to find the right one. And Melodyne should work with 32 bit tracks and not resample them down, maybe it would not distort so many of our transients and syllables then I might not have to re-sing so many takes. ?

"If", this is all due to a lack of specification in the ARA 2 standard, then every producer in every DAW is dealing with these issues. All the more reason why this is vital to the music industry to fix this and not simply put our hands over our eyes and walk around it.

The waiter can walk up to your dinner table, grab a hold of the tablecloth and yank it out abruptly so as not to disturb anything on the table. The hard part is replacing the cloth in the same instant.

Or, Melodyne, you are no longer standing on a rug.  Oh no, where did it go?! lol

Inertia... 

On another note...

Some might say, "Rex you are simply not happy driving a Pinto, you need to drive a Maserati... A Pinto will get you to the store and back just the same."

No, it is not that. I just don't want a vehicle no matter what brand it is with two break pedals so I need to each time I have to stop, guess which break pedal is the right one. And yes, I can drive a standard and can do okay parallel parking. ?

Edited November 12, 2023 by RexRed
typo

[Glenn Stanton]

18 hours ago, RexRed said:

Some have taken this opportunity to suggest vocal training and repetitive bouncing and reapplication of Melodyne over and over. Fix a note bounce it and then resample and resample and resample...

yeah but what happens when someone else is the singer? and the tracks are done and dusted and now need to be mixed? so vocal training is nice, vocal caching is also nice, and having only / nearly perfect vocals arrive from parts unknown is also nice. ?

but -- if i may -- i tend to do call the comping and then apply the pitch fixes as it's faster for me to get the emotional content and timing first, then adjust pitch if needed. i guess it depends on the workload and time pressures that its not always feasible to go back to fix a take's pitch but rather get it all in one shot as much as possible.

Edited November 12, 2023 by Glenn Stanton

[Mark Morgon-Shaw]

4 hours ago, Glenn Stanton said:

i tend to do call the comping and then apply the pitch fixes as it's faster for me to get the emotional content and timing first, then adjust pitch if needed. 

This is the way. 

The OP's issues spring from using Melodyne as a crutch. 

[RexRed]

12 hours ago, Glenn Stanton said:

yeah but what happens when someone else is the singer? and the tracks are done and dusted and now need to be mixed? so vocal training is nice, vocal caching is also nice, and having only / nearly perfect vocals arrive from parts unknown is also nice. ?

but -- if i may -- i tend to do call the comping and then apply the pitch fixes as it's faster for me to get the emotional content and timing first, then adjust pitch if needed. i guess it depends on the workload and time pressures that its not always feasible to go back to fix a take's pitch but rather get it all in one shot as much as possible.

The problem is you don't know if Melodyne is going to corrupt a line or not until you apply it and see how it sounds.

If the singer is behind the mic, then is not the time to talk about them taking more voice lessons.

If Melodyne corrupts the line then you have the singer re-sing the line, if the singer is not available then you put the line out sub par and try and hide it with the music, effects or anything else available. The line maybe quite fine it is Melodyne that is unable to interpret it correctly.

The whole thing is, if the singer is there you never leave a Melodyne corrupted line in the song. You make another take, apply Melodyne, hear how it sounds, and another and another till the line is right. That is the workflow.

Edited November 13, 2023 by RexRed

[RexRed]

7 hours ago, Mark Morgon-Shaw said:

This is the way. 

The OP's issues spring from using Melodyne as a crutch. 

Melodyne is a crutch if you want to insult the singer, Melodyne is a tool if you want to respect the singer.

[RexRed]

When Bruce Springsteen sped up the tape to the song, "Hungry Heart" to make his vocal pitch sound higher, maybe the producers might have suggested to him to go take some voice lessons to improve his vocal range...

See how inane this logic is?

Edited November 13, 2023 by RexRed

[Mark Morgon-Shaw]

16 hours ago, RexRed said:

The whole thing is, if the singer is there you never leave a Melodyne corrupted line in the song. You make another take, apply Melodyne, hear how it sounds, and another and another till the line is right. That is the workflow.

It's a poor workflow. You are doing a post production during the actual recording. 

I'm not sure what you mean when you say " Melodyne corrupted line "  ?

16 hours ago, RexRed said:

Melodyne is a crutch if you want to insult the singer, Melodyne is a tool if you want to respect the singer.

I have no problem insulting anyone if it makes them get it right. It's a crutch if you're using it on every line just after you've recorded it. No legit singer would want to work like that, it's an art and a craft to coax the best performance from a vocalist.  Maybe call up a local pro studio and ask them if you can sit in and observe a session so you can see what a proper workflow looks like when working with singers.