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

Working with Audio Post

Those new to filmmaking or those working with no budget will likely have to do all the audio post themselves, using whatever audio tools are available in their video editing software. At some stage though, if you are a professional or aspiring professional, you will want to hand over the audio post to audio post professionals. However, handing over your project to audio post professionals requires additional knowledge, to avoid a number of common pitfalls. So I'll go into a fair bit of detail and explanation in this post, to help you understand and avoid these pitfalls.

Firstly, it's worth understanding a little about audio files: The best choice for all concerned, the production sound mixer, picture editor and audio post team is the Broadcast Wav format (.wav), although MXF format is OK too. The broadcast Wav Format (BWF) was specifically designed for use in broadcast and film applications. The BWF has two storage areas, one for the digital audio data ("essence") and the other for numerical and text data ("metadata"). The most obvious use of metadata is to store the audio file name but there are many other fields of information stored in the metadata which, if used properly, can massively aid the audio post workflow. Some of the metadata fields are written automatically when recording, depnding on the recorder settings, for example; the sample rate, bit depth, file creation date, frame rate, duration, etc. Other fields need to be entered manually, for example; File name, Project, Sound Roll, Scene and take numbers, Circled, etc. This information should be added by the production sound mixer during recording. The Time Stamp fields are also incredibly useful! Ideally, the camera/s and the sound recorder are being fed time-code from a single common source. This will result in the original time stamp field of the BWAVs precisely matching the time-code of the corresponding footage.

When you start editing your film, one of the first things you will do is to import all your footage and sound recordings and here is our first big potential problem because, when you import the sound files, your editing software will likely overwrite or delete much of the metadata. The first thing you should do before you import anything is to make a copy of your production sound recordings and work from the copies. This guarantees you have a copy of all the files with all the original metadata. Once you start editing, make sure you don't change any of the audio file names. When you've finished editing you will need to export your sequence for the audio post professionals, this is where things can get tricky!

On the plus side, virtually all audio post professionals use AVID's ProTools software, on the minus side, there are a few different ways of exporting your sequence and getting it wrong could cost you dearly. The two most common export container formats are OMF (pronounced "oom-fy") and Advanced Authoring Format (AAF). AAF is the most recent and slightly preferred of the two containers. AVID systems can export AAF natively whereas FCP or FCPX requires an addon like "Automatic Duck". Other less common editing programs may provide AAF export functionality but they often don't strictly adhere to the AAF specifications and result in ProTools not being able to open the AAF. The general workflow is to:

1. Export the video separately, adding Burnt In Time-code (BITC), using the video codec/resolution requested by the audio post team and including your sound mix in the video as a guide.
2. Export an audio only AAF (delete the video tracks if necessary), linking to the audio files, rather than embedding.
3. Send the video, the AAF and ALL the original production sound recordings to the audio post team.

Exactly how you achieve the video and AAF export will depend on your editing software and would normally be handled by the Asst. Picture Editor. If you haven't done this before, expect the odd problem and to have to do some research in order to produce an AAF which ProTools will open. But watch out for a few obvious problems: 1. Start your time-code at 01:00:00:00 and if you're working in reels, start each reel on the hour (EG: Reel 3 would start at 03:00:00:00). 2. Don't mix frame rates in your sequence. 3. Some NLE's allow mono, stereo or multi-track material on the same audio track. Don't do this though, make sure each audio track only contains audio clips of the same channel count. 4 Include a 2 pop, 1 frame of 1Khz tone at -20dBFS exactly 2 seconds before the first frame of action (FFOA) ie. 48 frames before the FFOA, if your frame rate is 24fps and add a corresponding (1 frame) of white screen to the picture at the same location. If you're working in reels, you need to do this for each reel and add a tail pop 2 seconds after the last frame of action (LFOA) for every reel except the last.

What should happen is all your audio tracks and edits are imported from the AAF and linked to the original audio recordings. If you have embedded (rendered) the audio in the AAF this can cause problems because if the audio team need to change an edit or look for room tone or additional material outside your edit boundaries we can only do this up to the limit of the handles you have set when rendering the AAF (usually 2 secs). A referenced (linked) AAF on the other hand allows us to change your edit boundaries up to the limits of the entire original audio recording.

Now let's take an extremely common situation: We're editing dialogue and come across a problem which can't be adequately fixed with processing. The solution is to find alternate, replacement dialogue by going through the audio files to locate all the alternate takes, wild tracks and different mic recordings (lav, boom, etc) and importing and synchronising them. Baring in mind a feature film usually generates thousands of production sound files, this task can easily turn dialogue editing into a nightmare, taking weeks or even months to sort out (and costing a fortune). If there isn't time to go through all the files looking for alternate dialogue, problems will have to be solved with ADR, which lowers the performance quality and again costs a fortune! However, here is where our metadata can save the day, providing of course all the relevant metadata was correctly entered by the production sound mixer and that metadata has been preserved by linking the AAF to the original audio files (rather than embedding). A couple of mouse clicks on the offending dialogue and ProTools will examine it's metadata and compare it with the metadata of all the production sound files and in seconds present a list of all the alternate takes and mic recordings. An additional second is all it takes for one or all of those alternates to be automatically imported, edited and synchronised. A task taking potentially an hour or longer maybe reduced to just a few seconds, multiply this situation a hundred or more times in an average feature and you can see why accurate metadata is so important!

Feel free to ask any questions all this info might raise.

G
 
In re-posting a question from another thread because it was off topic for that thread and is more relevant to this thread:

So, asking APE's advice with this idea.

Say as an Indie what CAN I do to reduce the expense of post audio?
How much "DIY" upfront can I do and not adversely impact the ability of the mixer?

Do I deliver the audio as separate time aligned audio tracks, a separate track for each line? Or just stems for dialogue?

Stems for sound FX
Stems for Foley
Stems for music
stems for atmosphere\sound scape.

There are two main stages to film mixing, the pre-mix and the final mix. The pre-mix is where the stems are created from the individual channels and the final mix is where the stems are balanced against each other, tweaked and mixed together. In larger budget films: Both the pre-mix and final mix are done in the same commercial mix facilities by (or at least supervised by) the re-recording mixer. The stems are generally not recorded down during pre-mix, usually there is an individual ProTools rig for sound FX, music and dialogue. These ProTools rigs are then sync'ed together each outputting one or more stems. At the end of the final mix both the complete mix plus the stems are recorded simultaneously, plus any other required stems (M&E for example). With really big budget features the stems maybe broken down further and supplied by as many as 10 or more synchronised ProTools rigs. This workflow makes it relatively easy to go into an individual rig and change something during the final mix.

To save the cost of paying for a mix room for both the pre-mix and the final mix, the indie filmmaker can supply the stems already pre-mixed (although this may not be allowed for certain distribution formats). The downside here is that it could be even more time consuming/expensive because any changes identified during final mixing could require recording one or more new pre-mix stems. To avoid this, I would want the stems broken down as much as possible but of course the more stems and channels I have to deal with the longer it's going to take, a bit of a catch 22. The other consideration is that the further away the acoustics are in the pre-mix room to the final mix room the more problems you are going to encounter during the final mix and the same is true for the level of skill/experience of whoever is creating the pre-mixes. With all these (and a few other) factors, there's no hard and fast rules about what pre-mix stems you should deliver to the mixer. Far better for you (or your sound designer/audio post guy) to have a conversation with the re-recording mixer and ask what he/she wants delivered.

To put audio post costs into perspective, a large budget feature for theatrical release with 5.1 surround could cost $10m or so. For a low budget indie feature (again theatrical release, 5.1 sound) the audio post is likely to take 3-4 months and you could possibly get away with as little as $80k-$100k (depending on the genre), with maybe a third or so of this budget going on just the last 2 weeks or so in the mix room. Obviously, most of the people here have nowhere near these sorts of budgets. Getting an experienced pro with a really good sound design room and then checking the mix for a few hours in a Dolby room could reduce the costs (of the low budget feature example) by a further factor of 4 or so (under certain circumstances!) and still result in a decent mix. Once you start going below this point though you are looking at very low level professionals or more likely amateurs/semi-pros/recent graduates/students, less appropriate audio post equipment/environments and therefore the chances of getting a decent mix decrease rapidly. I'm talking about features here, obviously shorts would be way cheaper.

Except for the last paragraph, I've spoken about mixing. If we're talking about audio post in general, there aren't really ways of reducing costs directly but there are ways of getting better quality for the same cost. This post is already long enough and my transcoding has just about finished so back to work for me. If anyone is interested, I'll try and elaborate in another post?

G
 
This was great. I knew you had written something like this.

So in audio post, when doing noise reduction, or some other correction, is the AAF file being adjusted, or are the reference files being adjusted, or is it some other duplication of corrected files going on?

This really was a great post by the way. Those little tidbits on avoiding common mistakes... just makes me want to give you a hug APE
 
So in audio post, when doing noise reduction, or some other correction, is the AAF file being adjusted, or are the reference files being adjusted, or is it some other duplication of corrected files going on?

An AAF is recognised as and will look like a single file but it's actually a container format. You don't actually change the container itself (the AAF "file"), you change the files it "contains". In the case of an embedded AAF, the files the AAF contains are all the audio clips in your timeline (usually with handles). In the case of a referenced AAF, the AAF doesn't actually contain any audio, just references (links) to the audio files you've used in your timeline.

When you import an AAF it extracts all the embedded audio clips and places them on the same tracks in the same place as they were in your NLE. When you import a referenced AAF, you are asked for a folder location (where all the audio files are stored) and then imports them into the timeline as they were in your NLE. Generally the audio post team/person will prefer a referenced AAF to an embedded AAF but it depends on a number of factors.

G
 
An AAF is recognised as and will look like a single file but it's actually a container format. You don't actually change the container itself (the AAF "file"), you change the files it "contains". In the case of an embedded AAF, the files the AAF contains are all the audio clips in your timeline (usually with handles). In the case of a referenced AAF, the AAF doesn't actually contain any audio, just references (links) to the audio files you've used in your timeline.

When you import an AAF it extracts all the embedded audio clips and places them on the same tracks in the same place as they were in your NLE. When you import a referenced AAF, you are asked for a folder location (where all the audio files are stored) and then imports them into the timeline as they were in your NLE. Generally the audio post team/person will prefer a referenced AAF to an embedded AAF but it depends on a number of factors.

G

I see. So working on the AAF and adjusting the audio files in the tracks just affects the AAF file and not the underlying audio files it references or links to, just like a video NLE? (edit: the reason I'm trying to confirm this, and clear my confusion, is because when working with Adobe Audition on Premiere, it keeps creating duplicate audio files with the changes in it. But I'm assuming that higher end audio programs such as pro-tools or Logic pro doesn't work like that)

------------
Any advice on what to look out for as I decide whether to go with an amateur audio post production person, someone who actually has an audio engineering background etc., but is an amateur. There are some of these guys around here in Toronto.

vs.

If I decide to do the audio post myself, what advice would you give me as I start the video editing/audio syncing process? I've been going through Pro-Tools and izotope RX tutorials, and am considering taking a shot at it. I'm extremely nervous that I'll overlook some basic thing since I don't have any meaningful experience with audio post. But I'm weighing my options and throwing myself into the mix as not doing that will mean spending a lot of money on some other amateur (since I cannot afford a pro), and the amateur might as well be me.
 
Last edited:
@trueindie,

For practical purposes its just a file, like APE said, AFF is a container, but its not really a container in anyway that is useful for you. When you change the file, you change the file.

Audition and Premiere work very differently. Premiere is "non destructive" meaning that your original files are not overwritten. They should remain unchanged through out the entire process. When you save the pro project, the original audio and video files are NOT overwritten. Now, audition I believe is "destructive" in some modes. When you save in audition you generally OVERWRITE the file. Im not sure if Audition in multitrack mode is non-destructive. Other multitrack programs, like cubase, pro-tools etc are typically non-destructive.


Looking for amateur? Look and listen to their work, verify that they "really" did what they say they did and that they still have that type of equipment\setup.


Try it your self for a while, you only have time to waste.. youll screw up, so what? At least youll be able to talk to your future post audio guy with a bit more knowledge.
 
I see. So working on the AAF and adjusting the audio files in the tracks just affects the AAF file and not the underlying audio files it references or links to, just like a video NLE?

When we import an OMF/AAF into our DAW, all the audio clips contained in the (embedded) AAF/OMF are copied into a folder and it's those copies we work on. Once the import process has completed, the OMF/AAF is stored away and not touched again. In other words, we never work on the AAF file itself, it's just used for the initial exchange of materials edited by the Picture Editor.

How those work copies are affected depends on the type of processing we are applying. Some types of processing causes a duplicate of the clip to be made and processing applied to that duplicate which allows us to return to the original copy if needed (like Audition), while editing and many other types of processing are completely non-destructive (like your NLE). This is essential as some types of processing we use routinely cannot be effectively applied in real-time. Different types of DAWs and plugins handle this in different ways though. BTW, Logic Pro is a good DAW but is much more suited to music/MIDI creation than to audio post, I don't know any audio post pros who use Logic Pro. The only serious competitor to Pro Tools for audio post is Nuendo but in many places Pro Tools completely dominates the audio post market. LA for example is, as far as I'm aware 100%, Pro Tools.

Any advice on what to look out for as I decide whether to go with an amateur audio post production person, someone who actually has an audio engineering background etc., but is an amateur. There are some of these guys around here in Toronto.

Tough question, because there are so many variables to consider. Virtually exclusively, those with an audio engineering background who are trying to get into audio post have experience, education and/or training of music engineering/production and typically know absolutely nothing about audio post, to the point of commonly not even realising there is a very significant difference!! Ideally then, you want someone who has trained/interned for a considerable amount of time in a commercial audio post house, not someone who thinks they can do audio post because their 2 year music production course included an audio post exercise or two.

Obviously it's not easy to find an amateur with real audio post knowledge/experience, so most likely you are going to be looking at doing it yourself or someone with a music production background. The advantage of someone with a music production background is they will likely have more and better audio equipment than you. However, they are not going to have any better idea than you about how to employ that equipment for audio post purposes. In fact, there's a fair chance they maybe worse than you because their ideas on how to employ those tools are likely fixed to how they employ them for music production!

Of course all of this is a generalisation, there are exceptions, but those exceptions are very rare or at least very much rarer than most lo/no budget indie filmmakers seem to realise!

If I decide to do the audio post myself, what advice would you give me as I start the video editing/audio syncing process? I've been going through Pro-Tools and izotope RX tutorials, and am considering taking a shot at it. I'm extremely nervous that I'll overlook some basic thing since I don't have any meaningful experience with audio post.

Chances are that you will overlook something basic, most likely quite a few somethings, either due to lack of knowledge/experience or due to translation issues (how the mix sounds when played by sound systems other than the one you mixed on). However, the probability that someone with a music studio and music production training/experience will overlook some basics is not, in all likelihood, significantly lower than the probability of you overlooking something. So I agree with Wheat, if you have access to Pro Tools, RX and plenty of time, then DIY is an attractive option. In other words, yes, it's unlikely to be very good but unless you can afford a pro or can find someone with a considerable amount of specific audio post training and experience for free, your only other option (music producer) is not likely to be very good either.

As to what advice I can provide if you're going to DIY, that's difficult! If we're talking about what you do in your NLE during picture editing? IE. before audio post, then the best advice it to keep the audio clips well organised. I often get AAFs from Picture editors which are a nightmare, there's no logic to what clips are on what audio tracks and I can spend hours or even days finding out what's there and organising it all before I can even start work. This of course costs the filmmaker one way or another, either in fees for additional time or in a lower quality final product (if cash or time are fixed). If you're talking about advice on actually doing the audio post, all I can say at this point is give it a try and go by what your ears tell you. Try say 30 secs of something, post it on youtube and start a thread on here, either Alcove, myself or both of us will critique it and try and give you some pointers. Here's an example which carried on from the Sound Design thread which you might find useful.

G
 
Thank you!!

Try say 30 secs of something, post it on youtube and start a thread on here, either Alcove, myself or both of us will critique it and try and give you some pointers.

G

That sounds absolutely brilliant. You guys will both be hearing from me.

Edit:
translation issues (how the mix sounds when played by sound systems other than the one you mixed on).
I'm going to be using these sony mdr-7506 headphones I purchased. Is there something I can do to guard against these translation issues physically (such as using different headphones to mimic different setups)? or is there nothing short of playing the audio on various systems to find out if there's anything wrong with them? I've got a couple of friends who have recording studios, and I was thinking of abusing my friendship with at least one of them.
/Edit
Best,
Aveek
 
Last edited:
when you import the sound files, your editing software will likely overwrite or delete much of the metadata. The first thing you should do before you import anything is to make a copy of your production sound recordings and work from the copies.

Is this still an issue? I haven't worked with an NLE in years that made any changes to the source material, video or audio. The last time I remember encountering a system that did this was probably in the mid- to late-90's.
 
Is this still an issue? I haven't worked with an NLE in years that made any changes to the source material, video or audio. The last time I remember encountering a system that did this was probably in the mid- to late-90's.

I think what APE is saying is that it depends on what kind of changes you're making. If you're cutting or fading sounds, it's not an issue, but I'm assuming if you're doing noise reduction, or corrections where you actually remove information about unwanted clicks or whatever, it's going to create a new file. It makes sense to me. You're not adding, say color information, as in a video file. You're actually removing information from the audio file, and so it creates a new file. Maybe in the future with superfast computers, removal information can be applied in real time, but I think right now, it's not possible.
 
Is there something I can do to guard against these translation issues physically (such as using different headphones to mimic different setups)? or is there nothing short of playing the audio on various systems to find out if there's anything wrong with them? I've got a couple of friends who have recording studios, and I was thinking of abusing my friendship with at least one of them.

No, unfortunately, there's essentially nothing you can do. There is no way of mimicking different setups with headphones, playing your mixes on various different systems can be helpful but can just as easily lead you further astray and the vast majority of music recording studios are commonly the worst places to try and reference a mix!

Is this still an issue? I haven't worked with an NLE in years that made any changes to the source material, video or audio. The last time I remember encountering a system that did this was probably in the mid- to late-90's.

If anything, the issue is worse now than it was in the mid-90's! Some NLE's strip some of the metadata on import, plus as trueindie states, there are various processes which can either only be applied offline or are better quality/more efficient to process offline, causing a new file to be created (and usually not preserving all the metadata. Although if you are handing the audio over to audio post then the Picture Editor should NOT be applying any of these types of processing! Lastly, creating an embedded OMF/AAF is inherently destructive because all new audio files have to be generated which are the length of the clips used in the timeline (plus handles). What tends to happen is things like the NLE writing to metadata fields (such as the "date created" field) the date when the NLE created new/duplicate files, rather than the date the original file was created and just changing the "modified" field. Most NLE's just ignore some of the extended BWF (Broadcast wav) metadata fields altogether and therefore don't write this data to any newly created files. Even selecting whole files to be embedded in the OMF/AAF doesn't solve the problem. The various metadata issues is one of the main reasons why we tend to prefer linked/referenced OMFs/AAFs to embedded ones, although again, some NLE struggle to create readable linked AAFs. And finally, particularly at the no/lo budget level, is the problem of roles not being fully understood or simply not being fulfilled at all. For example, a PSM's role is not just to record production sound, it's also to manage, organise and log all the production sound recordings and commonly there is no Assistant Picture Editor or anyone fully capable of fulfilling that role.

G
 
Back
Top