One of a series of posts dealing with issues I encountered when trying to edit a long and complex documentary with Kdenlive - the problems I outline cost a lot of time, and require repetitive manual tasks where new features and shortcuts in the software could do the job better.
Other posts:
Independent length adjustment of synched audio.
Fixing inaccurate camera sync

Export for Sound Mix

The Problem:
You can't do a proper sound mix within Kdenlive, of course - that's not what an NLE is designed for. In order to do sub-frame adjustments of audio, editing of envelopes, and complicated effects like noise reduction, fixing clipping or other problems, and using a range of different EQ and dynamics tools, you need a Digital Audio Workstation. I currently use Audacity, but there's no easy way to bring the sound from my edit into Audacity and back again without lots of repetitive manual work.
I need a second or two of extra audio on almost every audio clip in my edit, in order to be able to crossfade outside of the cuts, and in order to find exactly the right part of the waveform to edit with the Envelope edit tool - this comes in particularly handy when tidying up dialogue which I have chopped together word-by-word from a rambling, 'um' and 'ah'-filled interview.

Current Workaround:
Without being able to export an OMF, I have to manually create 'handles' on almost every single audio clip in the film. Not just dialogue, all of the B-Roll shots when I want atmo from them too. This means extending each clip either with the mouse or using the 'extend edit' shortcut.
Every audio clip needs to be carefully checkerboarded to ensure that there is enough space either side to extend it, and this makes it tricky to adhere to my 6-track organisational rule of where clips go in the timeline:

A1 - Dialogue 1
A2 - Dialogue 2
A3 - Atmo/FX 1
A4 - Atmo/FX 2
A5 - Music 01
A6 - Music 02
(screenshot will come later)

Of course, once I have created my handles, my edit sounds horrible as there is now 2 seconds of raw audio jutting out from many clips.
Luckily I did a git commit before starting the sound mix

OK, so I have prepared my timeline for export - I do this scene-by-scene, as I have a different project for each of the 27 scenes in the 60min documentary.

Now I export 6x 48kHz WAVs, manually muting the other tracks and manually naming & rendering each WAV.
Then I import the WAVs into an Audacity project (I should really learn Ardour as I believe it has non-destructive editing and a video track).
Of course, now my separate audio clips are just anonymous waveforms in one long track, with no tags or names, so it's hard to orientate myself in the mix, and I can't just grab individual clips and apply effects, add fades, etc - instead I have to manually select regions and play them back to find out what they are.
Once my mix is done, I export either a full mix as FLAC, or more commonly 3 D-E-M stems as individual FLACs: Dialogue, Effects & Music. That way I can do simple adjustments of levels in Kdenlive without having to go back into Audacity.

Potential Solutions:
- Some kind of an exchange format. I know this isn't likely any time soon. AAF and OMF are hardly 'open' or 'standards' and neither Ardour or Audacity will open them anyway. Having this functionality would have saved me literally days of work. Days of manual fiddling about. In fact, it would have saved me from doing the sound mix at all - if I could provide an OMF/AAF I would have been able to get help from an audio pro to mix it. Back in 2006 I was using OMF in Final Cut and it was like magic. It's disappointing that the post-production industry hasn't come any closer to a real functional standard since then. Is there some kind of XML exchange format which could work?

-Automatic Checkerboarding - an feature where you could select a track and run a 'Checkerboarding' command, where you enter your choice of handle length. (eg. 2 seconds). Every audio clip on that track would be extended by 2 seconds before and after their current edit. Where there is no space for an extension of 2 seconds (i.e. because there's no more time in the original clip to extend it, or it clashes with another audio clip on the same track) the audio would only be extended the maximum length possible (this may be zero frames).
As an optional addition, you could specify a fade length for these handles, eg if I want every clip in Dialogue 1 & 2 to be extended by 12 frames with a 3-frame fade, and I want every clip in Atmo 1 & 2 to be extended by 2 seconds with a 12-frame fade, and I want every clip in Music 1 & 2 to be extended by 5 seconds with a 2-second fade.
I just found a discussion about an implemented Checkerboarding option in the Lightworks forum.

- A 'Stem Export' option to export all audio tracks at once, individually. This is a common feature in DAWs, and I guess it is probably one of the easiest of the features I'm suggesting to be implemented in Kdenlive. I still need to manually create handles but it simplifies one aspect of the job.

Other Ideas?
Is this a problem for other people?
Does anybody have suggestions for other current workarounds?
Thoughts on how potential solutions might be able to be implemented in Kdenlive?
Other relevant or related bugs/problems?

Kdenlive already integrates with audacity. Does edit clip on audio files open audacity for you? What is missing from this feature? https://userbase.kde.org/Kdenlive/Manua ... /Edit_Clip

And a "stem export" option for audio output was also added, also available in the KDE4 branch (posteriror to 0.9.10 release => MLT's builds are on this state).

Many thanks for these posts, enlightening us on many aspects. They definitely should be on website: I will enable a blog for you on kdenlive.org, do you already have an account or should I create one for you?

Maybe others would be interested to tell stories, tips & tricks etc, don't hesitate to ask for an account! (qubodup, TheDiveO, capslock, CorrosiveTruth...?)

Great, thanks for the tip Vincent!
A login for the blog would be great, I will write a more general, less technical blog outlining and summarizing the interesting and challenging aspects of tackling a project like this with Kdenlive.

ttguy, as far as I understand it, 'edit clip' opens a single audio clip, right?
What I'm talking about is a scene with dozens, sometimes hundreds of audio clips, which need to be edited and mixed in position. If I am cutting a sentence like "the mirrors reflect the light on a pipe, through which water flows" and I have actually chopped it together from many other parts of the interview (this was often the case in the film) it may in fact look like this in the timeline:

Dialogue 01:----------| the mirrors |---------| the light |--------------| through which |----------------------
Dialogue 02:----------------------------| reflect |------------| on a pipe |----------------------| water flows |---

In this case, it doesn't make sense to open only the clip for '| the light |' in Audacity, and then '| on a pipe |' etc.
I need to have them in the correct timing so that I can edit the envelope of the waveform and cut around other words or sounds.
I need to see where I have created gaps in the sentence which I will need to fill with atmo from the same interview.
Then, once I have done the fine cut to tidy up my edits, I need to apply EQ, amplification and compression to the whole section where this person is talking, then mix and render the whole scene's dialogue tracks to a single track.

I can't check the usage of this feature right now but as far as I remember from last time I tried it, it doesn't allow me to do all this.

Basically in my imaginary post-production dream world, what I am looking for is not the option to right click on a clip in the timeline and choose 'open in Audacity' but rather the abilitiy to right click on a .kdenlive project in my file browser and select 'open with Audacity'.
Then every audio track in the project becomes a track in Audacity, and while opening, Audacity would politely ask: "Would you like handles on your clips? how about fades?"
...it would also fetch me a beer, and do the sound mix for me, of course. And in this world, Audacity has a beautiful GUI.

Yep. You can only edit one track at a time in audacity when you launch it via kdenlive.

Have you checked out the "stem export" vincent mention yet?

no, I haven't done another sound mix since then, but I'm sure the stem export option will come in handy. I'm currently getting to grips with Ardour which might improve my workflow somewhat - I'm looking forward to testing it in a production environment!

sammuirhead you re speaking out of my heart,
I m just struggling with audio in kdenlive very much (even right now in the middle of the night )

because of this thread i just found out about the stem export option, thats great!! Although it seems it just renders the stems from audio track, and not the sound from embedded video tracks. I think for every kind of workflow it would be great to have the embedded sound also exported with the stem export.


Actually learning Ardour 4 for sound post production is not really as complicated as it appears in the first moment, and it will improve and speedup your workflow tremendously. There is also ways to get all kind of exchange formats into ardour via ArdourXchange or AAtranslator. These programs need wine which needs some configuration, but then you can import aaf, omf, and more or less whatever you like into ardour! Like ardour is open source (and reads xml files) it would be fantastic to have a aaf like exchange format between ardour and kdenlive, not sure if this is utopic but then there would be nothing missing in the linux world in terms of audio video postproduction! I m not a programmer, so no idea how complex and time intensive this would be.. but maybe there is something like an open standard already that can be used?

so soundwise my main workflow problems in editing are:

when i open scripts as clips , the clips just have the embedded audio. so when i want to make a huge project with scripts for every scene, it doesnt really work because i loose the possibility to edit the single tracks in audio. It would be fantastic to have scripts that when imported into the timeline would be able to split audio into different tracks again. so my cumbersome workflow for my project now is to arrange scenes more or less and render them out to lossless and make stems and reimport these files in a "master session". this is not very flexible at all and costs time and diskspace, but i couldnt find another solution ....


and this is a mlt thing as i understood, but it would be great to align audio and video in subframes, because syncing dialogue without the accuracy of subframes is impossible.

and theres more i cant remember right now...


EDIT:

jack support would also be fantastic, like all professional audio software and most video applications (lightworks, shotcut, blender i think..) supports it. this means when you work on some audio and then want to edit video next to it you have to shutdown the jack server and everything to be able to edit video with the alsa / oss sound ... or you have to install cadence and the snd_aloop deamon , you see it s getting complicated.....