Sound System 'POP' after homemade trailers

[Jacob Tutt]

Hi Everyone,

At our cinema we've got into the practice of creating trailers, and fun home made 'snipes' that we play during our regular trailer reel before the feature starts.

Something I have noticed, and it is not always the case, in the transitions between homemade (DCP-O-Matic) trailers or content and studio create content like trailers or the feature, there will be a pop through the entire sound system. It's like a little blip.

I seem to think there is a relation to these elements being created with DCP-O-Matic, but possibly there is something going on with how our server plays back this content.

I would be grateful for any insight or experience others can share.

JT

[Mike Blakesley]

What kind of server / sound processor do you have?

I wonder if it's caused by switching from Interop content to SMPTE. I don't know if that would cause such a pop. We use homemade content here from time to time and we never have an issue like that. (We have a GDC SX3000 server, and Trinnov Ovation 2 sound processor)

[Pete Naples]

What are they sourced from?

I wonder if there's DC offset going on?

[Carsten Kurz]

It is easy to open the audio MXF from trailers, snipes, etc. with Audacity (with the FFMPEG extension installed) and inspect the start and end of the audio files. If there is nothing there, there IS nothing there. Some servers/media blocks may create that kind of artefacts at clip transitions, I guess, possibly when frame rates change, which I assume happens often with preshow content that you create your own. Also different numbers of audio channels in a DCP could cause this. If you rip content from non-professional sources, the edits may not be done at zero crossings or with fades applied, that may cause pops as well. It is certainly not a technical problem caused by DCP-o-matic.

[Leo Enticknap]

I was wondering about audio pops caused by a change in sampling rate between CPLs. Agreed with Carsten that by listening to the audio file on a PC, you will be able to determine if the artifact is baked into the recording, or being generated in playback.

Some servers/media blocks may create that kind of artefacts at clip transitions...

...and I have a vague memory of reading in firmware release notes for both the IMS3000 and the Alchemy that some recent updates fixed them (in some specific circumstances/combinations). Updating your server, media block, and audio processor to the current release versions might be worth a shot. Jacob: can you tell us what model of server, media block, and audio processor is in use, and what firmware versions they are on?

[Marcel Birgelen]

Early Alchemy revisions did have some issues with sound artifacts on CPL transitions, most likely due to the software forgetting to clear some buffers. Having written some audio decompression and playback routines myself in the far-away past, I remember how important it was to properly clear all memory buffers when switching into another mode, to avoid sound artifacts, especially during mode changes like sampling rate changes.

[Randy Stankey]

If you rip content from non-professional sources, the edits may not be done at zero crossings or with fades applied, that may cause pops as well. It is certainly not a technical problem caused by DCP-o-matic.

That's what I was thinking, too.

See if applying a cross fade over your audio transitions helps. It's probably a lot easier to go back and check your editing before you go tearing into things.

[Jacob Tutt]

We are using a Dolby IMS2000 server, software version 2.8.26-0 and CP750 processor here.

Carsten, what is a 'zero-crossing' or 'fade'?

Randy, what is a cross fade, and how would I apply it over transitions? Would this be done before I create the DCP version? We don't have much in terms of editing software here. Actually, DCP-o-matic is the most sophisticated software we have on our booth computer!

I'll see if I can get a little more specific for you on the IOP / SMPTE / channels / frame rates between offending clips this eve.

JT

[Leo Enticknap]

That version is 2-3 years old - current is 2.8.30 with a hotfix. The firmware and security manager (which have to be updated separately on the IMS2000) are likely also of a similar vintage. I would suggest updating all three to current as a first step. The procedure is not completely intuitive: you have to update the software, then security manager, then (media block) firmware, and the latter two need a KDM (downloadable from Dolby's FTP site at no charge). If you have remote access and a service contract with a support vendor, they should be able to do that remotely for you.

Applying a fade in or fade out is something that would have to be done in your video editing software (e.g. Premiere Pro, FCP, or Resolve) in the process of creating the video file that is then put through DCP-o-Matic to create the CPL. I don't believe that DCP-o-Matic has that capability, unless its fade in/fade out feature applies to both pix and audio.

[Randy Stankey]

Jacob;

If you look at your video in whatever software you use to edit, you can see the individual media clips that have been joined together to make your program.

If you look at two adjoining clips, each should have a picture track and an audio track. An audio track is often represented as a Cartesian graph where time is on the X-axis and intensity is on the Y-axis. This is where the crux of your question lies.

Let's say you have "Clip 1" joining with "Clip 2." You have imported into your timeline, trimmed them to your desired cut point and butted them up together to make one sequence from two clips. Mentally zoom in on your timeline and examine the splice between those two clips where the audio tracks join.

If one sound track is very loud and the other is very soft, you're going to have a jump in the sound levels where the transition is. When you listen to that transition in real time, it will seem like your sound is "jumping." In certain cases, that "jump" can cause a click or popping sound.

There are two ways to solve this. First, you can put a cross fade over the transition to blend the two sound tracks together and hide that popping sound. Second you can re-trim your edit point so that you don't have one loud sound joining another.

Imagine that Cartesian graph of your sound track again... X = time. Y=intensity. Your audio waveform can be positive, up to some theoretical maximum and, of course, it can go negative with your zero line in the middle.

What if the soundtrack of "Clip 1" is very loud/intense in the positive direction and the soundtrack of "Clip 2" is equally loud in the negative direction? When your video plays, there will be a sudden jump in the sound from, say, +10 units of intensity to -10 units. If you look at a magnified view of your soundtrack's graph, it will appear like the line is dropping off a cliff. This is the very cause of the pop you hear in the sound track.

The way to solve this is to re-edit your transition point (splice) so that both soundtracks are at or very near the zero line. That place is called the "Zero-Crossing Point." Now, when your splice between two clips passes by, you won't hear that jump in sound.

A "Cross Fade" is where you gradually lower the intensity of the last few moments of your outgoing sound while simultaneously increase the intensity of your incoming sound. Imagine an audio engineer manipulating the controls of a sound mixing console, gently moving one fader lever down while moving another fader up in order to create a smooth transition between two sounds. That's what your video software does when you apply a "Cross Fade" transition over the splice in your timeline.

I don't know what software you use to edit video but I use Final Cut Pro. If I was to edit such a video as we are discussing, I would import my clips to the timeline, trim them to the place I want to make the splice then apply a cross fade over the transition point. In my timeline, it would look like a Band-Aid stuck over the joint. (Maybe a piece of virtual splicing tape.) Now, when the final product plays, there will be a smooth transition and no jumps of pops.

What I am suggesting is that, when you troubleshoot your problem, there are two ways to proceed. You can delve into your projector's operation, the server, the sound system and other things. When you do that you will have to expend a certain amount of time, trouble, expense and sustain a certain amount of risk. You could also just go back to your computer where you made the video in the first place and check to see that your editing is good.

If your editing is faulty, you simply fix it, re-export and play the video again. Problem solved! If your editing is not faulty, you'll know that there is something in your projection system that needs looking into. Which is easier? Sitting at your computer for an hour or two or tearing your projector apart?

When I was just learning how to troubleshoot problems in movie theaters, I was taught to look for the easy solutions first then move to the more difficult, expensive or risky solutions only when the easy things have been eliminated.

[Carsten Kurz]

Jacob - identify two trailers between which an audio pop occurs. Tell us their properties, as far as you can identify them - frame rate, number of audio channels, etc.
Download the free audacity, together with it's FFMPEG plugin. The two DCPs contain two MXF files - a larger one, and a smaller one. The smaller one can be loaded into Audacity. You should look for signal right at the beginning or end of these audio MXFs. Look at the attached screenshot. Something like this could cause an audio pop. Professional trailers, ads, etc. would all have silence at the beginning and end, or a clean fade to silence. Depending on your sources or the way you use them in DCP-o-matic, this may not always be the case for the DCPs you create yourself.

This may not be the reason for your issues, but, it is something I would check. Which version of dcp-o-matic are you using?

If you do not always have these pops between trailers, you should try finding a pattern. This could be DCP/clip technical properties, or content, or, as Leo suggested, a firmware issue. Some people also experience a visual 'pop' between trailers - a green flash, etc., if two clips with different properties are played next to each other. The Dolby DSS200 is known for this when the frame rate of two successive clip changes.
You could also try to put a short piece of black (server black) between clips and check wether the pop still occurs, and if, wether it's at the end of the first clip, or the beginning of the second.


Bildschirmfoto 2021-08-30 um 00.59.02.png

[Randy Stankey]

Carsten reminded me of something. I wasn't sure whether you were talking about pops within video clips or between them.

If you are making "mini-movies" and the pops happen as they are playing, we're talking about "within."

If you are stringing several items together in your show program and the pops happen when one item ends and another starts, we're talking about "between."

I just assumed "within." Reading again, I see that you might be talking about pops "between." Regardless, the concept of "zero-crossing" applies to both scenarios. The difference is only where the problem occurs and how to deal with it.

If your sound pops happen "within" then go back and check your editing.

If your sound pops happen "between" then do like Carsten says.

[Randy Stankey]

BTW: This is probably the most famous occurrence of an audio "click" in a movie soundtrack that I can think of.

It occurs immediately after the word, "sir."

That sound clip has been my incoming message sound in my e-mail program and for text messages on my phone for as long as I learned to make custom alert sounds.

I'm sure it happened because the sound editor(s) on Monty Python either missed it or didn't think it was bad enough to warrant the time and effort of going back to fix it but that click has bugged the crap out of me, ever since the first time I tried to make that movie clip into an alert sound.

I can't "not hear" it!

And, now, I share it with you so that everybody can tear their hair out whenever they hear it!

[Paul Drage]

This happened to me last week too. It did it on two video only DCPs I created from static images. I restarted my machine, rebuilt the DCPs, and it went away.

Now I'm checking output on DCP-o-matic player after the creation, and hope it doesn't do it again.

[Jacob Tutt]

Hello everyone,

It's been a busy month at the theatre, and my attention has been drawn in different directions. Only now just sitting down to take a closer look at your suggestions, and I thank you all for your feedback!

I'd like to point out again that this popping problem does not occur at every run-through, but it does show up repeatedly on our main 'feature presentation snipe' before 3 seconds of black. Here's what it looks like:

---

Feature Pres: Scope - 7.1 - 2k - 24 frames per second - roughly a 20 second clip

End of clip: *POP*

Next clip: Black (3 sec)
- CP750 5.1 mode (from beginning of clip)
- CP750 5.0 level (from beginning of clip)

---

Opening up the track in Audacity as Carsten recommended, it looks like all 8 channels fall to '0' before the clip ends.

I see a lot of suggestions for cross fading or fade outs/ins. You'd think that the Dolby server would be capable of doing this automatically on all clips to prevent this sort of thing? Or why not?