Ducking how-to?

I’m working on some videos from my motorcycle trip in the Alps. There are two sound tracks: the sound recorded while riding, and a music track that overlays the bike sound. The result is mainly music, with motor noise coming through when I really got on the throttle, or when the music quiets down.

On a couple of occasions I rode past cattle wearing big cow bells. I want to duck the music track (the bike sound includes the bells)for the duration of hearing the bells. And I want the duck to be gradual, not sudden.

At the moment, the only solution I can think of is break the music sound track by removing a piece of the track slightly longer (to make handles) than the needed change. The piece of the music track is then copied to three new audio tracks (call them 1, 2, 3).
On 1, I do a fade out from the start of the new clip into somewhere near the start of the clip,
on 2, I fade in from somewhere near the end of the clip to the end of the handle,
and 3 I drop the gain, to the desired level, of the entire clip.

The results are the clip gradually fades from the normal track level to the lower level and back up again. But that’s a lot of editing and guessing to get it right.

Is there a better, easier way to do ducking?

Nope, that (and variations of it) is about the only way.
BTW: I wasn’t aware of the term ‘ducking’ for that process.

Here’s a definition for simple ducking. It can also be triggered automatically - source A ducks a track. For example: voice ducks background. And there are other versions. But, in the end, track A controls track B.

BTW, I can do my hacked version of ducking with two audio tracks: Create the clip to be ducked on a new audio track. Edit a clip from the start of handle to end of fade and fade it out. Cut the remaining clip at the start of fade in, remove the intervening audio, and fade in from the start of that clip. The result will be two short clips with a fade out and a fade in. Copy the clip to have less gain on the second track.

Effectively this is two audio dissolves to a clip. The fade out track will eventually be covered by the low gain track. The fade in will eventually cover the low gain track. Shazam!

OK, maybe I should try to put together a video of how this works. [/grin]

In the meantime, here’s a sample of what I’m looking for. Stay with the video and the ducking will show up. (This was done with Resolve - which I’ve since dropped for reasons I think belong only in PM) And yes, this is essentially a dashcam with music and titles. [/smile]

In all my years being involved with various areas of electronics technology I had never heard the term ducking. And yet, after reading the Wikipedia entry, I see it describes very well the effect you are seeking. We really do need keyframes don’t we? I do hope that they are delivered as promised by the end of the year.


Here’s a video sample of ducking. This is how it was done.

The first track is “as is”. The second track was boosted to about 2 db, to cover the bike sounds. The fades begin or end at that level. The third track was dropped to about -17 db.

All of the bike audio comes from the camera’s internal mic. SJCam SJ7 Star - 1080p60 fps. Smugmug, of course, compressed the video into near trash.

See the sample above for what the output can look like. It was also 1080p60 fps, but I transcoded it to 4K60 fps. This gets around the worst of YouTube’s compression. The bad news is a 30 minute video runs to about 125 _G_b. From starting the upload until YouTube has processed the video to 4K takes 12-24 hours(!). Transcoding takes 6-8 hours, The upload, via a fast FiOS connection, takes 2.5 hours or more. The rest is YouTube’s work. Processing takes longer on weekends. Gee… I wonder why? [/wink]

The toll road leads to Ötztaller Gletscher, in Austria, (Ötz Valley Glacier) noted for being where Ötzi, the Stone Age man, was found, and being the 2nd highest paved road in Europe.

125Gb is just 3Gb shy of YouTube maximum file size. There are plenty of 4K videos running well over an hour on YT with exceptional image quality those videos will have needed to be under 128Gb.

Hmmm… I forget where the 125 Gb number came from. I went through all of the Alps videos to date. The “record” length is 120 Gb, but some are as low as 75 Gb.

The reason for transcoding to 4K is only to keep YouTube from making a hash of the video. Transcoding 1080p to 4K does not make the video sharper, crisper, etc.

i managed to achieve a nice stereo ducking effect with the soundtrack from one clip blending in with the soundtrack from the following clip by using the filter in the audio section and applying a slow fade out for about three seconds, (it had a rather nice taper), and then transitioning to the next clip for about a second and a half by dragging the following clip and watching the time code tick down to where i wanted it…the soundtrack on the following track was pretty constant so ducking was a piece of cake…if i want to do an actual layered mix-down, typically i’ll use audacity and do my mix-down there. once i have it the way i like it in audacity, i’ll save as a .wav and bring it into Shotcut by opening an audio track, opening the .wav in the playlist, and then append it to the audio track. i’ll then sync it up by moving it back and forth under the video track until i have it just right…sometimes it’s a one and done…sometimes i have to play with it a bit to get the desired result. this may be one of those times when experimentation does the best service. but when it comes to actually ducking within Shotcut…i think so far i’ve just gotten lucky.

Consider this for making re-syncing something from Audacity: Add an audio track and place 1 second of tone (Open Other in the File menu goes to generators including a tone generator) so the start of the tone is at the start of the audio going to Audacity. Do your Audacity magic and bring the resulting track back into Shotcut. Align the start of the Audacity file with the start of the tone.

OK, let’s say the Audacity track starts at dead quiet. Now what? Audacity tells you where the start of the track is. That’s not driven by the signal level. However identifying the start, once the track is in Shotcut, might be difficult. Add a one second tone in front of the Audacity track and align its end with the start of the tone in Shotcut. Done right, the track is in the correct place. Discard/edit out the tones and life is good.

actually that sounds like a pretty nifty technique…i’ll give that a go should i ever find myself in a similar situation…but one second of generated tone is more than i’m really wanting to listen to…i think i’ll try it with about a .03 second burst, (normally i knock about 13dB off the tone generator in Shotcut just to make it listenable. run as is it pegs every meter i have), definately enough for a pip that would cover the span of about three frames,and just long enough to lock onto for a good syncing…thanks, i’ll try that some time.

Pick whatever interval works for you. Keep in mind a longer tone will be easier to find, and aligning to it won’t be as difficult.

i get what you are saying, and it makes sense…

Glad to help - party on! [/grin]

Ducking isn’t done manually. Ducking uses one track to drive compression on another. Shotcut doesn’t have the capability. I can’t think of a video editor that lets you apply compression to audio let alone drive the compression with another track. Automation of the volume is what most video editors do. Shotcut has a volume/gain filter but I don’t see a way to automate it. I would do the ducking in a DAW then export the finished audio to Shotcut. There are good free DAWS that will accomplish it. Manually ducking more than a couple of spots would be cumbersome. Shotcut will get there. I like it, but it is not fully mature yet. Not far from it though.

Agreed that Track A automatically ducks Track B. But lacking “automatically”, one does what one can. [/smile]