• Wondering which camera, gear, computer, or software to buy? Ask in our Gear Guide.

Question about crossfading audio.

I finished reading the book Dialogue Editing for Motion Pictures, by Purcell. It has some good stuff in there, but one thing I cannot figure out, that it does not quite explain, is the crossfading. In ADR when matching up the words, there are popping sounds in between, and if I try to crossfade them, the audio gets quieter, and sometimes it will fade into after the take has been cut, and you will here things you are not suppose to here. What do I do about that, or how do I crossfade it correctly? Thanks.
 
You shorten the cross fade. Start the cross fade after the dialogue is over and end it before the next bit or word starts. Don't cross fade into silence either, rather into room tone.
 
When you say popping sounds, I take it you are referring to editing clicks? You can avoid these by editing at zero crossing points (which is always good practice when editing audio). Also, the Purcell book is referring to editing with professional audio editing software (ProTools), which has customisable crossfades, Equal Power or Equal Gain crossfades for example. Your problem sounds like what often occurs with equal gain crossfades (rather than equal power). Obviously video editing software does not contain all the audio editing features of dedicated audio software so I don't know if your software contains these crossfade features.

G
 
A lot of times though there is zero crossing, since the actors say one word after the other without pause Sometimes I like to mix audio takes, depending on if a line is said better in one, and put it with another, and that's where there is zero crossfading and a pop occurs. If I got them say every word individually, it would sound unnatural of course. I tried shortening the clips and then crossfading before, but the problem with that this, is that the dialogue volume turns down quiet, then gets loud again, and you can totally tell. Even with constant power, it still doesn't seem to work most of the time, and I can still hear the volume change. What I am doing wrong?
 
Last edited:
I tried shortening the clips and then crossfading before, but the problem with that this, is that the dialogue volume turns down quiet, then gets loud again, and you can totally tell. Even with constant power, it still doesn't seem to work most of the time, and I can still hear the volume change. What I am doing wrong?

This is one of those skills that takes the right tools, a LOT of doing it, and a great set of ears. Once again I recommend that you hire someone to do the job and sit with him/her for hour after boring hour. I'm not trying to be insulting, but you still don't get audio. Working with a professional for a while will hopefully set off the light bulb.
 
I know it's just I can't find anyone to do it. Maybe there is no one where I live who has that expertise. I've been looking for the past few months, but fear I will have to do it myself. Aside from kijiji or craiglists, any other sites I should be posting adds up? I also have not finished all the sound FX and Foley, since some is too quiet for me to know how to record properly as of yet. So if I do hire someone, I will still not be finished everything for them to work on. But as long as they are okay with just the dialogue for now...
 
Okay I could try Regina. I didn't loo that far. What if I send them the movie though, and a lot of the audio has been cut incorrect, for them to work with? Is there anything I should know? Thanks.
 
Ask what they want. I want the raw OMF as well organized as possible and all of the audio from the unused takes. I don't want any fades or volume automation It's usually not organized properly, so I spend quite a few hours just getting everything into a usable state. The client gets charged for that.

It always comes down to budget. The more work you want done, the more it's going to cost. It depends upon the project, but I usually charge by the linear minute for the four major audio categories - dialog, Foley, sound effects, mix. Score and music spotting only takes a few hours. I also charge for out-of-pocket expenses, like vegetable mangling for gore and fight effects, or any extensive custom recording like vehicles; it's all agreed upon in the contract prior to my beginning to work on the film. If the director and/or producer is local (I've done projects where I've never met anyone involved, working via phone, email, FTP and FedEx) we have an initial walk-through/talk-through (yes, I've also spent hours on the phone doing these) and a few more as the film progresses, usually after the dialog edit, when the ambient BGs are completed, when the Foley work is done, when sound FX are finished, and of course the mix.

What is the film about? Not the plot, but the "message" so to say (i.e. it's a film about making choices) or something as basic as "it's a comedy," "it's a bang-bang-shoot'em-up," "it's a horror film," "it's a mystery," etc. If you can't do this you're already in trouble.

What mood do you want?

Tell me about the characters. Character "A" is moody and withdrawn, character "B" is surly and aggressive, character "C" is is ditzy but lovable, and so on. This helps me when doing Foley and sound FX. You don't just put in footsteps when you do Foley, the footsteps need to reflect each characters personality, as do the sonic atmosphere of their living spaces. The same applies to sound FX. Example: on one film the protagonist was making choices throughout the film. I noticed that most of the time he left a room after making a choice; so bad choices had big, heavy door slams, good choices had "light" door closes. It's subtle, but it's there.

Get me in touch with the composer. I really like collaborating so the score and sound design don't conflict. This is especially important for action scenes.

There's more but you get the idea. The more I know about what the client wants the happier the client is with the final result. I hate working with "I'll know it when I hear it" types, except that they usually spend a lot more money, which is okay, but I do get frustrated. It's all about COMMUNICATION; and that's your job as a filmmaker. You need to communicate with your cast, your crew and your post team so you can ultimately communicate exactly what you want to communicate to your audience.
 
A lot of times though there is zero crossing, since the actors say one word after the other without pause Sometimes I like to mix audio takes, depending on if a line is said better in one, and put it with another, and that's where there is zero crossfading and a pop occurs. If I got them say every word individually, it would sound unnatural of course. I tried shortening the clips and then crossfading before, but the problem with that this, is that the dialogue volume turns down quiet, then gets loud again, and you can totally tell. Even with constant power, it still doesn't seem to work most of the time, and I can still hear the volume change. What I am doing wrong?

Firstly, I want to second everything Alcove has said!!

Secondly, I take it you meant in your first sentence that there is no zero crossing point? If this is what you are saying, then you are wrong. All sound we can hear is made of sine waves, sine waves are to sound what atoms are to matter. Have a look at this image:
Zero_crossing.svg


In effect, the blue line represents the in/out movement that your speaker cone will make to reproduce the sound wave. The black, zero line is where the speaker cone will sit when there's no sound or when the speaker is switched off. Zero crossing points do not therefore exist only between words (because words aren't individual sounds) but within words, syllables and even individual letters. If you edit anywhere other than the zero line, the speaker cone will have to instantaneously move to that point to start playing and of course it's impossible to move anything anywhere instantaneously, so the speaker creates a click. If you can't see the zero crossing points it's because you are not zoomed in far enough. Editing audio is often done while zoomed right down at the sample level and as there are 48,000 samples per second you have to be zoomed in virtually as far as your software will allow, providing your software will zoom down to the sample level.

Your issue with the dialogue getting quiet and loud again is two-fold: As I said before, you are probably using an equal gain xfade rather than an equal power xfade and what you are hearing may also be compounded by your xfades being too long.

Editing dialogue (inc. xfading) is a skilled job and the best dialogue editors are very highly paid, same is true of re-recording mixers, who mix the dialogue. If it were easy to learn these jobs just from trail and error and reading books, there would be so many capable people doing them well that market forces wouldn't demand paying people highly to do these jobs. From your posts H44, it seems like you are not yet "getting" even the basics of audio editing, so you need to take Alcove's advice and watch someone who does.

G
 
Okay sure. I hope that books though, can still teach me in the future. I know what zero crossing is, what I meant is, is that there is no silence in between words, when I try to put sentences together. That's where I have trouble. Right now I will concentrate on putting the rest of the foley in, then send it into them and they can get rid of unwanted noise hopefully.
 
Last edited:
I hope that books though, can still teach me in the future.

Yes and no. Books can teach you the theory of how digital audio works and of the functionality of the various tools and processors but they can't teach you when or how to apply these tools because that's a judgement call. There are a number of things you need to know in order to create decent or good sound for Film/TV and without exception you need all of them:

1. You first have to have an understanding of what sound is and how it works.
2. You have to have a good understanding of how humans perceive sound.
3. You need to know the audio tools and how they work.
4. You need to train your hearing to focus in on detail and to be objective about what you're hearing.
5. You need to put together all the above to both manipulate the audience's emotional response and to create a convincing illusion of a fictional reality.

You can find books or info on the internet which cover the first three items but you can't learn items 4 or 5 this way. Both of these final two items present additional difficulties: Item 4 for example, how can you train your hearing to focus on detail or to be objective if your speaker system/monitoring environment can't reproduce the detail or can't reproduce it accurately? Even if you're an expert in items 1-4, there are so many tricks and methods which have been developed over the 80 years of sound for TV/Film pertaining to item 5 (which also depend on the distribution format), not to mention that item 5 is ultimately dictated by artistic judgement. It would take many decades to learn all this through trial and error and books alone. Taking a degree in the subject speeds the process along a little but still doesn't give you enough to enable the production of competent mixes. Taken along with the enormous cost of accurate audio reproduction equipment (essential for item 4), this is why there is no alternative but to learn audio post from experienced pro's in a professional audio post facility and then practising/honing these skills for several years.

It's certainly important for filmmakers to learn about the possibilities of sound and sound design as story telling tools but unless you intend to specialise in audio there's not a great deal of benefit trying to learn how to do it yourself. If you are looking to become a professional filmmaker at some point in the future, you will obviously need to create commercial quality products and that will mean hiring audio specialists/professionals to do the sound. So, much of what you are trying to learn currently about the mechanics of audio post is likely to be of limited/no use when you get into the profession!

I know what zero crossing is, what I meant is, is that there is no silence in between words, when I try to put sentences together. That's where I have trouble.

Whenever you record any audio, there is never silence between words, sounds, musical notes, etc., just less sound or, occasionally, you might even need to edit right in the middle of a sound, which is sometimes possible because zero crossing points exist even in the middle of sounds. How you edit and mix effectively depends on what you've got to start with and what you want to end up with. If you have two pieces of dialogue, say Dialogue A and Dialogue B, do you match A to B or B to A or as is frequently the case, do you match both A and B to some illusionary concept, say Dialogue Z? You need all 5 of the items I mentioned previously to answer this question successfully and this can only be done on a case by case basis.

G
 
Okay thanks. I just feel I didn't learn anything useful from the books because of all the variables they are not able to get into. I'll send it into some pros. So I am guessing I should sends them the whole Premiere Pro file of the movie, or can I just send them a copy of the movie on hard disk? Do I have to send them all the original audio files, or can they just work with what I have this is already cut into the movie?
 
I just feel I didn't learn anything useful from the books because of all the variables they are not able to get into.

As I explained, that's always going to be the case and bare in mind that the Purcell book is far and away the best book written on the subject of dialogue editing! It's particularly useful for history, workflows, session layout, etc., but as you say, can't actually cover the practicalities of editing specific dialogue.

I'll send it into some pros. So I am guessing I should sends them the whole Premiere Pro file of the movie, or can I just send them a copy of the movie on hard disk? Do I have to send them all the original audio files, or can they just work with what I have this is already cut into the movie?

You'll have to contact them and ask. The usual workflow is a quicktime movie with BITC (Burnt in Time-Code) at whatever resolution/codec they specify, an AAF (or OMF) preferably with all the audio linked rather than embedded and all the original audio files but again, you'll need to check all this with whoever you are using.

G
 
Okay thanks I will. One thing I've noticed is that the room tone, is different than it was when the dialogue was recorded in it. This could explain why there is more popping sounds than maybe their should be in the dialogue tracks, but that's just my guess.

There was one scene, where we were not able to get a boom operator from 10-4 on weekdays, which is only when the location was available. That scene we shot then did complete ADR after, but maybe I should reshoot it with recorded live dialogue, so it will cost less money for them to clean up the ADR. But I feel like just keeping the scene as edited too, so perhaps I will spend the extra cash, since the scene is already fine.
 
One thing I've noticed is that the room tone, is different than it was when the dialogue was recorded in it. This could explain why there is more popping sounds than maybe their should be in the dialogue tracks, but that's just my guess.

The fact that they sound slightly different is not the reason you are getting pops. If you know what you are doing you can crossfade just about anything.
 
Actually I made three successful crossfades today, you just have to turn down the volume, at the part of the fade, where it's about to play more than what is intended to heard. Or at least it's one way I figured out how. A lot of them will not work for that though, and still trying to figure out the exemption.
 
Back
Top