I thought I'd write a basics guide to audio post, for relative newbies (or more experienced filmmakers) who are serious about filmmaking and want to avoid their films feeling like no/nano budget amateur films, which is an unavoidable consequence of poor sound! It's aimed at those with a budget for audio post (albeit a very limited budget) and what you can do to maximise the effectiveness of that budget.
Most audio post professionals are extremely wary of (or won't touch) no/nano budget amateur films because we know from experience that it's going to be a frustrating and unrewarding experience: We're going to have to put in a great deal of time/work to even achieve the best hope of a relatively poor result and most of our time is going to be wasted dealing with basic, silly and entirely avoidable filmmaking errors. It's just not worth it, either for us or for you! It's not worth it for you because all that wasted time is going to cost you either: 1. A lot more money to pay for that wasted time or, if you don't have a lot more money, 2. A much poorer end result/film because you're going to have to hire a student, amateur or semi-pro whose time is far cheaper because they don't have the experience, tools or skills to do as good a job. A vicious circle of poor sound (and therefore poor films) which can only be broken by eliminating these basic, avoidable filmmaking errors.
The first stage of audio post is importing all the audio post deliverables from the picture editing dept, a job which should take just a few minutes but which frequently takes days! Traditionally these deliverables are created by the Assistant Picture Editor. I realise that nano budget indie filmmakers can't afford a professional Assistant Picture Editor but, you can't just ignore the role because you can't afford one! You (or your pic editor) is going to have to fulfil that role yourselves. Do you know how to create a video of your edited film with appropriate, continuous BITC, in the format requested? Do you know what an AAF file is, what the different types are and more importantly, how to create an AAF on your NLE which can be imported into Pro Tools without errors? If your answer to any of these questions is not an emphatic "Yes", then expect days rather than minutes of audio post costs for this task! My advice would be to find someone local with a Pro Tools rig and practice making AAFs until you can import into Pro Tools accurately, without error, every time and to master this knowledge/ability well before you ever need to hire an audio post person! I'll come back to AAFs again, later.
The next task of audio post is to edit the production dialogue recordings. Production sound is always full of little errors; crew noises, background noises, cast noises, props noises, costume noises, etc. Even those noises which might appear almost imperceptible during filming often become painfully obvious during sound editing/mixing. Some are easily avoidable, others are not. There are a number of threads here on indietalk relating to avoidable noises but in summary: Turn off all electrical appliances on or near the set; air con, fridges, cell phones, radios/TVs/computers, etc. Call for silence from everyone on or anywhere near the set. As much as possible, make sure your actors only move or handle props in the little pauses between sentences/phrases, not exactly at the same time they are speaking. Watch out for bed, chair, sofa or floor creaks during the dialogue. Obviously, make sure that whoever is acting as Production Sound Mixer PSM) is actually capable of recognising when the dialogue is packed with these sorts of errors. Better still, both you (the director) and your PSM should be monitoring the sound recordings. Remember, what happens between the words is irrelevant, we can cut that out in a second but when those noises occur over the word/s we can't, we're going to have to spend time finding another solution to the problem. The normal procedure in this case is to replace that word/s from an alt take. Failing that, the next option is ADR but that's more expensive/time consuming and more difficult to match. Last resort is clean it up as best as possible, try to disguise the problem a bit and just live with the fact that it's crappy. All of this raises several issues, issues which can either be dealt with quite quickly or which can again waste huge amounts of time:
1. Alt Takes: Obviously, if there aren't any alt or safety takes then the primary and best solution doesn't exist! Even if there are alt takes, if the actor does not deliver the same words in the same way, then again, we're plum out of options. If you/your actors decide to change the dialogue or it's delivery from one take to the next, fine ... but make sure you have a safety take of the different delivery/s! Also bare in mind that it's not uncommon for me to encounter a problem noise, say a clothes rustle or prop noise, only to find that exact same noise in exactly the same place in every alt take!
2. Fill: Even with interior scenes, the pauses between sentences is usually full of little breathing, movement or other noises. So we routinely cut it all out and replace it with "fill". This fill is created by stitching together all the little bits of clean pauses we can find. This is often a very time consuming or even a near impossible task as there are so few bits of clean pauses. You (the director) can greatly help this by say holding your hand up and demanding silence/no movement for 5 or 10 seconds before you shout "cut" or after you call "Sound" but before you call "action". Specifically recording 30 secs of room tone for each shot can also be very useful. With exterior shots we're going to get much more background sounds and often of much longer duration, distant car, truck, train or plane passby's, dogs barking, etc. This produces another set of problems because not only do they impact the dialogue more but because of the duration, they often fall in the middle of a subsequent picture edit, so some relatively clean fill is essential. In those cases where we just can't find or make enough clean fill, all we can do is carefully lower the noise where possible and then add back in some canned sound FX to try and disguise the problem.
3. Wild Dialogue: Where you're pretty sure that you've got noisy dialogue, most of the time for exterior shots for example, plan on recording wild dialogue. This is the lines recorded again, in a less noisy environment with the mic in the perfect position, yielding more dialogue and less noise. The sooner you can do this after calling "cut" on the actual shot, the better, while the style and delivery of the lines are still fresh in your actor/s minds.
4. ADR: On professional shoots we always use lavs + boom. Being so much closer to the actor/s mouths, the lav recordings will often give us a usable recording when the boom doesn't, either due to framing restrictions, a noisy location or both. Where neither boom, lavs, alt takes or dialogue wilds are usable, the last alternative is ADR. There are a lot of disadvantages with ADR though: Your actors will not be able to get the same delivery/emotion as they achieved during filming. Unless it's very carefully/well recorded we're going to have proximity and other issues to deal with. It's difficult to match to the original dialogue or impossible unless we can layer it with some good clean fill. And lastly, this all takes more time and money!
5. File Management: Yes, I know this sounds boring but in some respects, it's more important than all the above points put together!!! It doesn't matter what alt takes, wilds, ADR, room tone/fill you've recorded, if I can't find any of it! The WORST CASE SCENARIO is unfortunately also an extremely common scenario with most indie filmmakers: Hundreds of audio files (or thousands in the case of a feature) often spread over numerous folders, with filenames which give no indication of which scene, take or shot they belong to or which only make sense to the editor or director who created them, plus hundreds of duplicates with different and even more obscure names, created by the editor/NLE. The worst case scenario is arriving right back where I started when I first listened to the sync dialogue but having wasted a huge amount of time trawling through countless files for alt dialogue and/or usable fill. It's just not practical to go through this for every shot of every scene unless: A. The filmmaker has enough money for double or triple the amount of time budgeted or B. Has the resources (actors' availability/skill + facilities/personnel) for me not to waste time trawling through hundreds/thousands of files and just immediately list any problems I come across for ADR. On professional shoots, PSMs use pro field recorders which record bwf (Broadcast Wav) audio files which are timecode, scene and take stamped and locked to the cameras. Providing this metadata is preserved, this workflow eliminates virtually all file management issues but for those who can't afford to hire this pro equipment, far and away the best and often the only realistic solution to these file management issues is to completely avoid them!
5a. Audio Post Deliverables: OK, so what exactly do we need? In no particular order:
A. A sound log. A log written by the PSM, listing the filenames, scene, shot and take numbers, with comments about faults or good takes. A sound log is not absolutely essential but can save a lot of time in the event of a disconnect between filenames in the timeline and the original production sound filenames and can also save considerable time if what I'm looking for is mentioned in the comments and I don't have to physically find and go through every take.
B. An EDL (Edit Decision List). Similar to the sound log, an EDL helps me lookup exactly what scene/shot/take I'm looking at and marry that up with the audio filenames. Again, not absolutely essential but it takes little time to create one and can save a lot of time finding files. An EDL is invaluable in the event of any editing which occurs after the edit has been delivered.
C. An AAF. While I can (and often do) work with embedded AAFs, the limited handle length restricts our opportunity to find clean fill. Ideally we would want a linked/referenced AAF, linked to the original (unprocessed, unedited) production sound files. If an embedded AAF/OMF is the only practical option then make sure you don't render mono audio files as stereo. In any event, avoid the use of MP3s or other lossy audio formats, which are all designed for distribution and not for further processing. BTW, mixing audio formats is a common cause of errors when exporting or importing AAFs. Stick to one of either; standard wav, bwf (broadcast "wav") or aiff formats at 48kHz 24bit. Another useful tip is for your FFOA (First Frame of Action) to start at timecode 01:00:00:00, not at 00:00:00:00, which is a common newbie error.
D. Even with an embedded AAF, we'll still need all the original production sound files, for unused alt takes, wilds, tone/fill and in case the embedded audio files contain some processing applied by the pic editor. ALL the files need to be named meaningfully, IE. Scene, shot and take numbers, followed by type and/or character. For example: "Scn001a-tk2-Wild-Peter.wav" or "Scn014c-tk3-tone.wav". Naming your files this way means that at a glance I can see all the audio files which pertain to a particular scene/shot and what they contain. Naming a file "bedroom scene end.wav" doesn't help much, particularly if there's more than one bedroom scene/shot! Worst still is if you haven't maintained a consistent, logical file naming convention. If for example you name your next audio file "Take2 bedroom scene end.wav", this file will appear in a completely different place in the file list to "bedroom scene end.wav", I can't see at a glance which files pertain to which scene/shot or what's in them. I'm going to have to trawl around listening to likely candidates for what I'm looking for. Now, it might only take 20 mins to find and audition all the possible files but multiply that by 10 scenes, 3 shots each scene and 3 takes per shot and that's 30 hours of work for what should have taken about 5, that's 3 full working days wasted, probably about a third of your audio post budget down the toilet! For a feature we could easily be talking 2 weeks completely wasted and if some or most of your filenames are completely meaningless ("SAC_000-1492.wav" for example) then we're into worst case scenario territory!
E. A rendered locked-off video with BITC (Burnt In Time Code), with the FFOA starting, as with the AAF, at 01:00:00:00 and obviously in the same time code format as your frame rate. Personally, I prefer a H.264 at full HD res and bit rate around 15,000kbps for online delivery or full HD ProRes 422 or DNxHD 120 for physical delivery but different audio post people will likely have different requirements.
G
Most audio post professionals are extremely wary of (or won't touch) no/nano budget amateur films because we know from experience that it's going to be a frustrating and unrewarding experience: We're going to have to put in a great deal of time/work to even achieve the best hope of a relatively poor result and most of our time is going to be wasted dealing with basic, silly and entirely avoidable filmmaking errors. It's just not worth it, either for us or for you! It's not worth it for you because all that wasted time is going to cost you either: 1. A lot more money to pay for that wasted time or, if you don't have a lot more money, 2. A much poorer end result/film because you're going to have to hire a student, amateur or semi-pro whose time is far cheaper because they don't have the experience, tools or skills to do as good a job. A vicious circle of poor sound (and therefore poor films) which can only be broken by eliminating these basic, avoidable filmmaking errors.
The first stage of audio post is importing all the audio post deliverables from the picture editing dept, a job which should take just a few minutes but which frequently takes days! Traditionally these deliverables are created by the Assistant Picture Editor. I realise that nano budget indie filmmakers can't afford a professional Assistant Picture Editor but, you can't just ignore the role because you can't afford one! You (or your pic editor) is going to have to fulfil that role yourselves. Do you know how to create a video of your edited film with appropriate, continuous BITC, in the format requested? Do you know what an AAF file is, what the different types are and more importantly, how to create an AAF on your NLE which can be imported into Pro Tools without errors? If your answer to any of these questions is not an emphatic "Yes", then expect days rather than minutes of audio post costs for this task! My advice would be to find someone local with a Pro Tools rig and practice making AAFs until you can import into Pro Tools accurately, without error, every time and to master this knowledge/ability well before you ever need to hire an audio post person! I'll come back to AAFs again, later.
The next task of audio post is to edit the production dialogue recordings. Production sound is always full of little errors; crew noises, background noises, cast noises, props noises, costume noises, etc. Even those noises which might appear almost imperceptible during filming often become painfully obvious during sound editing/mixing. Some are easily avoidable, others are not. There are a number of threads here on indietalk relating to avoidable noises but in summary: Turn off all electrical appliances on or near the set; air con, fridges, cell phones, radios/TVs/computers, etc. Call for silence from everyone on or anywhere near the set. As much as possible, make sure your actors only move or handle props in the little pauses between sentences/phrases, not exactly at the same time they are speaking. Watch out for bed, chair, sofa or floor creaks during the dialogue. Obviously, make sure that whoever is acting as Production Sound Mixer PSM) is actually capable of recognising when the dialogue is packed with these sorts of errors. Better still, both you (the director) and your PSM should be monitoring the sound recordings. Remember, what happens between the words is irrelevant, we can cut that out in a second but when those noises occur over the word/s we can't, we're going to have to spend time finding another solution to the problem. The normal procedure in this case is to replace that word/s from an alt take. Failing that, the next option is ADR but that's more expensive/time consuming and more difficult to match. Last resort is clean it up as best as possible, try to disguise the problem a bit and just live with the fact that it's crappy. All of this raises several issues, issues which can either be dealt with quite quickly or which can again waste huge amounts of time:
1. Alt Takes: Obviously, if there aren't any alt or safety takes then the primary and best solution doesn't exist! Even if there are alt takes, if the actor does not deliver the same words in the same way, then again, we're plum out of options. If you/your actors decide to change the dialogue or it's delivery from one take to the next, fine ... but make sure you have a safety take of the different delivery/s! Also bare in mind that it's not uncommon for me to encounter a problem noise, say a clothes rustle or prop noise, only to find that exact same noise in exactly the same place in every alt take!
2. Fill: Even with interior scenes, the pauses between sentences is usually full of little breathing, movement or other noises. So we routinely cut it all out and replace it with "fill". This fill is created by stitching together all the little bits of clean pauses we can find. This is often a very time consuming or even a near impossible task as there are so few bits of clean pauses. You (the director) can greatly help this by say holding your hand up and demanding silence/no movement for 5 or 10 seconds before you shout "cut" or after you call "Sound" but before you call "action". Specifically recording 30 secs of room tone for each shot can also be very useful. With exterior shots we're going to get much more background sounds and often of much longer duration, distant car, truck, train or plane passby's, dogs barking, etc. This produces another set of problems because not only do they impact the dialogue more but because of the duration, they often fall in the middle of a subsequent picture edit, so some relatively clean fill is essential. In those cases where we just can't find or make enough clean fill, all we can do is carefully lower the noise where possible and then add back in some canned sound FX to try and disguise the problem.
3. Wild Dialogue: Where you're pretty sure that you've got noisy dialogue, most of the time for exterior shots for example, plan on recording wild dialogue. This is the lines recorded again, in a less noisy environment with the mic in the perfect position, yielding more dialogue and less noise. The sooner you can do this after calling "cut" on the actual shot, the better, while the style and delivery of the lines are still fresh in your actor/s minds.
4. ADR: On professional shoots we always use lavs + boom. Being so much closer to the actor/s mouths, the lav recordings will often give us a usable recording when the boom doesn't, either due to framing restrictions, a noisy location or both. Where neither boom, lavs, alt takes or dialogue wilds are usable, the last alternative is ADR. There are a lot of disadvantages with ADR though: Your actors will not be able to get the same delivery/emotion as they achieved during filming. Unless it's very carefully/well recorded we're going to have proximity and other issues to deal with. It's difficult to match to the original dialogue or impossible unless we can layer it with some good clean fill. And lastly, this all takes more time and money!
5. File Management: Yes, I know this sounds boring but in some respects, it's more important than all the above points put together!!! It doesn't matter what alt takes, wilds, ADR, room tone/fill you've recorded, if I can't find any of it! The WORST CASE SCENARIO is unfortunately also an extremely common scenario with most indie filmmakers: Hundreds of audio files (or thousands in the case of a feature) often spread over numerous folders, with filenames which give no indication of which scene, take or shot they belong to or which only make sense to the editor or director who created them, plus hundreds of duplicates with different and even more obscure names, created by the editor/NLE. The worst case scenario is arriving right back where I started when I first listened to the sync dialogue but having wasted a huge amount of time trawling through countless files for alt dialogue and/or usable fill. It's just not practical to go through this for every shot of every scene unless: A. The filmmaker has enough money for double or triple the amount of time budgeted or B. Has the resources (actors' availability/skill + facilities/personnel) for me not to waste time trawling through hundreds/thousands of files and just immediately list any problems I come across for ADR. On professional shoots, PSMs use pro field recorders which record bwf (Broadcast Wav) audio files which are timecode, scene and take stamped and locked to the cameras. Providing this metadata is preserved, this workflow eliminates virtually all file management issues but for those who can't afford to hire this pro equipment, far and away the best and often the only realistic solution to these file management issues is to completely avoid them!
5a. Audio Post Deliverables: OK, so what exactly do we need? In no particular order:
A. A sound log. A log written by the PSM, listing the filenames, scene, shot and take numbers, with comments about faults or good takes. A sound log is not absolutely essential but can save a lot of time in the event of a disconnect between filenames in the timeline and the original production sound filenames and can also save considerable time if what I'm looking for is mentioned in the comments and I don't have to physically find and go through every take.
B. An EDL (Edit Decision List). Similar to the sound log, an EDL helps me lookup exactly what scene/shot/take I'm looking at and marry that up with the audio filenames. Again, not absolutely essential but it takes little time to create one and can save a lot of time finding files. An EDL is invaluable in the event of any editing which occurs after the edit has been delivered.
C. An AAF. While I can (and often do) work with embedded AAFs, the limited handle length restricts our opportunity to find clean fill. Ideally we would want a linked/referenced AAF, linked to the original (unprocessed, unedited) production sound files. If an embedded AAF/OMF is the only practical option then make sure you don't render mono audio files as stereo. In any event, avoid the use of MP3s or other lossy audio formats, which are all designed for distribution and not for further processing. BTW, mixing audio formats is a common cause of errors when exporting or importing AAFs. Stick to one of either; standard wav, bwf (broadcast "wav") or aiff formats at 48kHz 24bit. Another useful tip is for your FFOA (First Frame of Action) to start at timecode 01:00:00:00, not at 00:00:00:00, which is a common newbie error.
D. Even with an embedded AAF, we'll still need all the original production sound files, for unused alt takes, wilds, tone/fill and in case the embedded audio files contain some processing applied by the pic editor. ALL the files need to be named meaningfully, IE. Scene, shot and take numbers, followed by type and/or character. For example: "Scn001a-tk2-Wild-Peter.wav" or "Scn014c-tk3-tone.wav". Naming your files this way means that at a glance I can see all the audio files which pertain to a particular scene/shot and what they contain. Naming a file "bedroom scene end.wav" doesn't help much, particularly if there's more than one bedroom scene/shot! Worst still is if you haven't maintained a consistent, logical file naming convention. If for example you name your next audio file "Take2 bedroom scene end.wav", this file will appear in a completely different place in the file list to "bedroom scene end.wav", I can't see at a glance which files pertain to which scene/shot or what's in them. I'm going to have to trawl around listening to likely candidates for what I'm looking for. Now, it might only take 20 mins to find and audition all the possible files but multiply that by 10 scenes, 3 shots each scene and 3 takes per shot and that's 30 hours of work for what should have taken about 5, that's 3 full working days wasted, probably about a third of your audio post budget down the toilet! For a feature we could easily be talking 2 weeks completely wasted and if some or most of your filenames are completely meaningless ("SAC_000-1492.wav" for example) then we're into worst case scenario territory!
E. A rendered locked-off video with BITC (Burnt In Time Code), with the FFOA starting, as with the AAF, at 01:00:00:00 and obviously in the same time code format as your frame rate. Personally, I prefer a H.264 at full HD res and bit rate around 15,000kbps for online delivery or full HD ProRes 422 or DNxHD 120 for physical delivery but different audio post people will likely have different requirements.
G
