Using Ardour for Video post production fps question.


I’d like to do a music video project where I am going to record a live performance in multitrack with Ardour and a Camcorder simultaneously and edit the Audio and Video together, I notice there are fps settings within Ardour. Do I need to set up the Audio recording frames per second to match the Video capture fps prior to recording the performance? Secondly is 29.97fps (NTSC) a valid and possible fps setting within Ardour? I’m not on my Ardour box right now and I seem to recall only 25 and 30 fps being options but I could easily be wrong (too easily usually ;-))

Anyone done this? Any pointers appreciated!

On my cell right now so sorry about short and likely poorly typed reply.

You don’t need to set fps while recording, but it also isn’t necessarily a bad idea for your own reference. It really only affects Ardour’s clock display when set to SMPTE. Ideally your video and audio should be TC locked but if you are using a camcorder this is not likely to be possible.

And just about any commonly and not so commonly used fps setting is available in Ardour.


Does this often


If you want to record audio and video for a longer period (longer than a few minutes) the clock drift will become very visible. Just be aware of this and take precautions by either synchronizing your clocks (probably not possible with consumer camcorder and simple sound card) or by using multiple cameras and a lot of work in the edit room :slight_smile:

Here’s how one user solved the problem of audio/video sync :

Correct, but its one thing to do that with speech, doing that with complex musical sources is going to be an entirely different beast all together. Even with speech it isn’t ideal and will leave artifacts, the question is how noticeable those artifacts will be.


ntsc sync for me also never was a problem with ardour… but the cam sync… maybe you have a soundcard that can send a smpte signal to your camera? just an idea… of course depends also strongly on your camera, usually better hd / dv cameras will can sync as slave to a external clock/timecode

The solution peder suggested will create artifacts, but they should be barely audible if you’re experience about the same amount of drift as the OP did. A 10 second drift over 40 minutes is about 0.4% of a time drift. This will only be noticeable if you have to stretch the audio (rather than shorten it), and even then would be negligible. However, why not do the opposite of this? It’s a bit more work, but you could follow all the steps in that post until the audio stretching. At this point, find out how much drift there is by dividing the amount of drift by the total running time (in the example case, 10 seconds divided by 40 minutes, or 2400 seconds). Write the result down. Now import the video clip into Cinelerra and apply the reframert effect to the clip to stretch it rather than the audio. In the example case, depending on which way the drift goes (can’t think off the top of my head, just woke up), you would set reframert to a value of 1.004 or 0.996. Try importing one of your tracks from the multitrack audio (or a stereo mixdown) into CInelerra for a reference just to make sure it’s working right. Once you’ve got the correct value, export the video file from Cinelerra and import into xjadeo to continue your mix. Alternately, make sure your mix is done before doing anything in Cinelerra, and you can get your video editing done all at once as well.

Keeping things in sync is not all that complicated once you get a plan of attack. I used the audio stretch method while doing drum tracks for my first album. I didn’t have a multitrack audio card, so I used four old computers with different stereo on board cards to mic the drumkit. Afterward, I imported all tracks into my main DAW PC, time stretched (syncing to the overheads), and used triggers on everything but the overheads. At first I had no idea how it was going to work, but in the end it really only took a few more minutes of thinkng and working, and I made enough money on the first day of sales to get a proper multitrack card ;-).

Good luck.


Thanks for all the input and answers guys, I’ve got a lot of trial and error ahead of me…

On a semi-related sidenote as someone who’s tried pretty much every Linux NLE video editor I’m constantly surprised at what can be done with Cinelerra, I’m using the latest Heroine Virtual 4.2 release and it’s surprisingly stable compared to the prior 4.1. It’s a pity that JACK support seems to be a real back-burner issue with every NLE but LiVES which is not quite as advanced in other areas. I hear rumblings of JACK transport in the next Kdenlive but MLT doesn’t support JACK so that may not get fully implemented.

Obviously macinnisrr is well acquainted with Cinelerra and Cinecutie, what else are people using? There seems to be a real drought of user discussion on music related Video production in Linux.

The Blender NLE is often used in the groups I work with actually. And that should have Jack support. Obviously I do a fair amount of sound track work for animation, but even for non-animated things i have seen that used(And in fact have been using blender a bit for some green screen work for kids videos recently, so I am looking seriously at the NLE for an easier time composting the effects rather than doing it in FCP).