Personal View site logo
Cinema gear deals, direct from factories - Gear deals and Gear deals section. Also check Cameras, lenses, software, gear deals.
You support is vital for us. To keep this place ad free and independent, select one of the options below.
Donations are going to community support costs, hosting, etc. Your support allows to improve and expand this site.
Post Production Sound Tutorial (+ a bit on location sound)
  • Workflow

    1. Location sound
    2. Transfer & backup
    3. Sync
    4. Edit
    5. Export
    6. Import in DAW
    7. Organize tracks
    8. EBU R128 & Monitor levels
    9. Create roomtone & clean tracks
    10. Noise Reduction
    11. Crossfade tracks
    12. Spotting session
    13. Foley
    14. Ambiance
    15. Sound Design
    16. Music
    17. Mix
    18. Export

    Wiki Link

  • 63 Replies sorted by
  • Hello all and welcome to the Post Production Sound tutorial. My goal is to expand on all 18 points of the workflow listed above keeping in mind that some of you have limited knowledge of the subject. This means that the overriding principle guiding my approach is to discuss concepts/techniques that have a PRACTICAL use and to not get involved in, or encourage tedious discussions about, the minutiae of theory or one product's merit over another. I will make every effort to remain gear/software agnostic so that whatever we cover here can be implemented on whatever gear/software the reader prefers or has access to. However, when I do upload a video to emphasize or demonstrate a point, I will necessarily have to use the tools (i.e. software/gear) I have at my disposal; in no way should this be taken as an endorsement of the tool being employed.

    The 18 points of the workflow above will take time to write about and elaborate on. My aim is to update each item on the list as the core overview of the subject is written and follow up with posts containing videos/illustrations/samples to illustrate those points I think might be more readily grasped by visual demonstration. I plan to cover the subject in the order of the workflow so would ask that you refrain from asking questions about topics not yet covered as I do not want the thread to bounce back and forth and becoming a chaotic mess; it will be much easier if the conversation progresses linearly relative to the workflow.

    A note about posting etiquette: Should anyone have a disagreement with any of the content of my posts, please present your opposing view and supporting arguments in a rational and civil manner; don't jump into the conversation with guns blazing. This has happened in the past and I find this behaviour unproductive and unnecessarily hostile, doing nothing to advance our collective knowledge base. I have no issues being challenged and will happily modify the wiki linked to in the OP to add important new info/methods or to rectify my errors. That being said, don't expect me to answer or address your post if it is rude or obnoxious in tone and/or spirit. The wiki link will always contain the latest updated version of the tutorial since the timeframe to edit posts in this thread is only 24 hours.

    OK, let's get this ball rolling. Have fun learning, and fire away any and all questions as topics start to populate. Don't forget that I'm doing this on my spare time and simply cannot dedicate the time required to cover each section in a thorough and thoughtful manner within a short period of time. I strongly encourage those of you working in the industry or with relevant knowledge/experience to contribute to this thread to enrich its content and refine the information base with your expertise - I'm looking forward to learning myself as this tutorial evolves.

    Cheers!

  • Awesome. Thanks a lot for this.

  • Looking forward to it!

  • Very cool. Thanks @spacewig

  • Good list. The one thing I don't see is the dialogue re-recording and editing. In narrative films and series there's often a dedicated dialogue editor who finds the best sounding takes for various lines and makes them work together. The dialogue editor then suggests lines for ADR/looping.

  • Agreed. I perhaps mistakenly assume that indie filmmakers don't have budget or resources for ADR sessions.

  • @spacewig Nice one! been waiting for a while for someone to explain the audio. You are right mostly people drool over glass and sensors but the audio which is most important not many touches on that.

    I have noticed coursera has started a course on sound design as well https://www.coursera.org/course/digitalsounddesign. Would be complimentary to spacewig's knowledge transfer. Looks like the audio gods have been in a nice mood lately!

  • I guess I would add open source software to the list since if you use commercial software it does not generally speaking include a commercial license. It gets very complicated if you are using loops, samples, EQ, algorithms, convolution, impulses, NR, mixing, etc., etc.

  • It gets very complicated if you are using loops, samples, EQ, algorithms, convolution, impulses, NR, mixing, etc., etc.

    What you mean exactly? For samples and loops included in software they have license you need to read. All else does not make sense.

    Opens Source make no sense also, as it just tell that some developers can get the source.

  • You have to read the license carefully for each and every aspect of each software program. Sometimes a single program contains many licenses. Most of these allow you to use the software, but not commercially. It is a personal use license. By using open source software, you then are free to use the resulting video commercially.

  • Sometimes a single program contains many licenses. Most of these allow you to use the software, but not commercially. It is a personal use license.

    For example? I never really saw paid sound editing applications who directly prohibits using software itself to produce commercial products.

    I think you just confusing it with part describing that owner can't lend this software or otherwise make income similar ways.

    By using open source software, you then are free to use the resulting video commercially.

    No. Again, open source is just the way source is available to developers. License may have same or even worse restrictions. Making source available do not restrict author any other way.

  • @DrDave

    Small addition, I think it is also not really proper place for discussion.

  • Well, not sure what you mean, and it is quite possible I am mistaken. However, here is the license for Audacity "Audacity is free software. You may use it for any personal, commercial or educational purpose, including installing it on as many different computers as you wish." Note that the license includes the phrase personal, commercial or educational. Most Eula's do not. For this reason, I recommend using Audacity or similar products for your workflow. Or, at the very least, including it as a hassle free option in a workflow description.

  • For this reason, I recommend using Audacity or similar products for your workflow.

    I recommend to ignore such recommendations and use best tool the job, not the open source one. If you are a company single call to lawyer and sending him license text will get rid of all strange ideas.

    As I said, I never really heard anything you talk about. Who knows, may be somewhere it exist.

  • Small update: I'll soon be finished item 1, location sound, and apologize its taken so long but that topic alone is a massive undertaking as far as covering key PRACTICAL concepts. It feels like I'm writing a book sometimes but I don't want to make propositions without supporting them, even if only superficially. Stay tuned.

    @DrDave I suspect your audio expertise will find many opportunities to expose itself in this thread ;)

  • Worth checking some of Michael Coleman's videos

  • 1. LOCATION SOUND

    1. Gear: Mic, cable(s), preamp/recorder, batteries, memory card, boom pole, shock mount, earphones, slate
    2. 24/48
    3. Make sure your gear works
    4. Scout locations
    5. -20db
    6. Inverse Distance Law
    7. Sound perspective
    8. Lavalier microphones
    9. Slate
    10. Roomtone
  • 1) The gear list above is the minimum you need in your sound kit to properly record audio for your narrative film project, assuming your standards exceed what your camera’s internal mic can provide. The reasons have less to do with the quality of your camera’s internal mics than the fact that they are static, i.e. fixed to the camera, and not terribly directional. The problem this creates will become obvious once we cover recording levels.

    Do you need a preamp, a recorder, or both? This all depends on whether or not you plan on recording the audio to camera, or to an external recorder. If you record straight to camera (BMCC/PC/PCC, GH4, etc) then you obviate the need for an external recorder and the need to sync in post, making it the most economical option, time and cost wise, as all you need is an external preamp to amplify your mic signal to line level [excellent preamps exist to achieve this such as Sound Devices mix-pre, Juicedlink, etc] and your audio will be perfectly synced to your footage.

    If you are recording to an external recorder (or both) then you will need a recorder, and, if the recorder’s built-in preamps are not sufficiently quiet, an additional preamp. Ideally you want an all-in-one device [Sound Devices 702, Fostex FR-2, Tascam DR-680] as this reduces the overall footprint of your recording setup by half (i.e. preamp, batteries, cables, bags).

    Your microphone clip MUST be shock-mount. A mic used handheld or with a standard clip will transfer all movements, vibrations, thumps, slides, etc... to your microphone capsule resulting in low frequency rumble and other noise pollution over your precious dialogue which is an absolute pain to remove in post. Neglect of this simple attachment will incur a steep time penalty further down the production pipeline.

    Make sure you have plenty of spare batteries charged and ready to be swapped out when necessary. Have enough to cover 12 hours of recording with phantom power turned on at all inputs. The only penalty of having too many batteries is a little extra weight; not enough stops production dead in its tracks (npi) and has everyone on set wondering who hired the unprofessional bozo to handle recording duties.

    Cans: The most important aspect of your monitoring earphones is they MUST BE CLOSED-BACK. Open-back cans will bleed sound onto the set, which can be a distraction and might be picked up by your boom/lav mics. Avoid at all costs.

  • 2) Your recordings never have to exceed a 48KHz sample rate at a bit depth of 24 bits. Anyone who tries to convince you of the contrary is full of shit. As a matter of fact, you won’t hear the difference between that and a recording captured at 16/44 i.e. CD quality. If someone you know insists that 24/48 is substandard and they can hear the difference between that and audio recorded at 24/96 or 24/192 have them listen to file "Tone 1" below. If they can’t hear anything play file "Tone 2" for them. If they still can't hear anything and complain that the files are silent (and they’re not, just look at your meters while they’re playing) have them listen to file "Tone 3", which is as loud as files 1 & 2, then change the subject because you’ve just invalidated their pretentious claim.

    I am not trying to start a flame war nor will I entertain any discussion regarding this hot-button topic. I merely want to prevent my fellow filmmakers from being conned into buying pricier gear offering settings they will never need, or of recording tracks using sample rates that will create unnecessarily large files and will likely incur a CPU overhead penalty when it comes times to mix. The clear advantage of 24 bit over 16 bit is that you have ten times the amount of values to represent the signal you've recorded which means rounding error artifacts (quantization noise) created when processing the files (noise reduction, spectral repair, reverb, EQ, deverb, etc) will remain firmly buried in the noise floor. In other words, they’ll be inaudible. As for the advantages of 48 over 44.1KHz, the file sizes are the same so might as well take the extra, though mostly negligible, smoothing of the audio signal this gives you. To summarize: Audio captured at a bit depth and sample rate greater than 24/48 will NOT sound better.

    Tone One.wav
    2M
    Tone Two.wav
    2M
    Tone Three.wav
    2M
  • 3) A lot of people neglect to do this. But please, check your gear BEFORE you arrive on set, preferably some time during store hours the day PRIOR to the shoot. Wire it up and record a minute of yourself talking, singing, farting, etc. Something’s not working? You’ve got a few hours to troubleshoot then buy/rent/borrow/steal the items necessary to resolve your issues. Showing up to a set with everyone eager to start filming but forced to wait while you try to find out why you’re not getting a signal guarantees you get schwartz-listed.

  • 4) Prior to principal photography ALWAYS scout the locations you’ll be filming at to determine what background (BG) noise issues you’ll be facing. Fridges, ventilators, generators, clocks, computers, phones, neighbours, pets, kids, trains, planes, motorized vehicles, lawn mowers, construction, neighbourhood bars/clubs, the ceramic tile company running machinery 9 to 5 down the street, etc. need to be identified in advance as they all have the potential of being picked up by your boom/lav and ruining your takes. Is it impossible to shoot with this noise present? No, you can always attempt noise reduction (NR) and/or spectral repair (SR) during post-production or you can replace unusable dialogue with ADR sessions but this adds a major bottleneck to your workflow. Will your actors be available for ADR? Are they capable of reproducing the same emotion(s) and tone(s) of the takes you’ve decided were keepers? Be warned: ADR is nowhere near as easy as managing BG noises of your locations and even choosing locations in advance that have the least probability of suffering from these issues. Assuming BG noise sources are within your control, simply making sure you can turn off/unplug offending devices during your shoot will make post that much less of a pain. As for NR/SR, keep in mind that these processes degrade the quality of your audio, usually to a degree proportional to the amount of correction to be made. Less processing = greater fidelity.

  • 5) Find out where -20db is on your recorder/preamp’s sound level meter. If that spot is not already clearly indicated (and it usually is) make your own mark using whatever means necessary (tape, marker, scratch). Turn all your gain settings to zero, plug in your microphone (switch on phantom power if required) and gradually increase the gain (microphone volume) so that the average sound level displayed on the meter hits that mark (-20db). Why? Headroom and noise.

    Headroom, in this context, is the amount of extra loudness a signal can have before it starts clipping. In digital audio, clipping (distortion) occurs when the signal coming into the recorder exceeds 0dbfs, (zero decibel full scale). Clipping sounds awful, is just about impossible to fix, so must be avoided. How do you know your signal is close to clipping? Look at your meter, that's what it's there for. If the average level of two people speaking (John and Jane in a cafe) is -20db on your meter, you then have almost 20db’s of headroom (~4x the perceived loudness) your recording device can handle before the signal clips, i.e. Jane throws steaming coffee into John’s face, John jumps up, grabs his face and starts screaming. Please remember this: headroom is what YOU decide it will be. When you record with an average of -20db, you are giving yourself 20db of safety in case your actors become louder. You can choose an average of -12db, but if your actor really gets into his/her part and starts yelling his/her lines or moves closer to the mic, your signal might clip = ruined take. Further, you’ll never have a final mix with an average loudness of -12db, so what would be the point. Why not just record everything at -40db, you ask, and never worry about it?

    Noise. And here we’re not talking about background (BG) noise but the noise inherent to your recorder (and all electronic devices). This noise, often referred to as self-noise/noise-floor/signal-to-noise ratio (SNR), is typically quite low but if the dialogue you record is too low (i.e. under -25db), when you raise the level back up during post you’ll also be raising the volume of your recorder’s self-noise by an equal amount. Dialogue recorded at an average loudness of -40db then raised to -20db in post will have a noise floor 20db (around 4 times) louder than it would be if the recordings were made at the -20db target level to begin with. Your tracks will probably be plagued by an audible hissing that will be a pain in the ass to minimize.

    To summarize: You want to record dialogue at a level loud enough to mask your equipments’ self-noise, but not so loud as to breach the clipping point = distorted audio. Make -20db your target average recording level. Additionally, you’ll appreciate the benefit of this when you mix your tracks using the EBU R128 standard.

  • 6) Inverse Distance Law (1/r) is the physical principle used to describe the propagation of sound pressure level (SPL) from a sound source as it relates to the distance from that source at which the SPL is measured. In a nutshell: It states that if source A has an SPL (measured in db) of X from Z ft/m away, it will have an SPL of X/2 (½) from 2Z ft/m away, X/3 (⅓) from 3Z ft/m away, X/4 (¼) from 4Z ft/m away… etc. In other words, the SPL drops by half (6db) per doubling of our distance to the reference SPL measurement.

    What’s important here is not how loud the source is (the actor speaking his/her lines), but how much gain you have to apply to the signal (coming from the microphone capturing the dialogue) to get an average reading of -20db. Let’s go back to John and Jane for a practical example:

    You are recording an over-the-shoulder shot of Jane speaking with John. The shot is framed tightly enough to allow you to hover your boom approximately one foot away from her head without the boom dipping into the shot. You adjust the preamp gain so that your meters are reading at least -20db when Jane is speaking. Perfect. Now the director’s moved to a two-shot which is wider but because of the camera angle you cannot get your boom any closer than 4 feet from the actors without it getting into the shot. The inverse distance law above tells us that Jane’s voice will now be 12db quieter, i.e. the average level of her voice on the meters will be -32db. Your two immediate options are to either record at -32db then raise the levels in post which, as discussed above, means that you’ll also be raising your recorder’s noise floor by 12 db; or you can simply increase the gain 12db on your preamp which will bring the levels back up to -20db. While this might sound like the no-brainer solution to your sound level drop, there is a catch.

    More noise. But this time from 2 different sources: the rest of your recording kit and your location’s ambient sound. Your mic, like your recorder, also has self-noise that is increased in direct proportion to the amount of gain you apply to the mic’s signal. Same with your preamp (though most modern opamps have inverse noise/gain characteristics, but I digress). So, on top of the recorder’s SNR you also have to contend with the noise inherent to the mic AND the preamp. A remedy is having a mic and preamp with the lowest noise-floor possible (this comes at a cost) and choosing a mic with a hot output (produces a stronger signal than other mics for an equivalent sound pressure level) allowing you to apply less gain.

    Ambient sound, also known as roomtone. People exist and interact in environments that are teeming with ambient sound. While you might do your best to minimize blatant BG noise from contaminating your set, unless you are recording in an anechoic chamber there will always be ambient sound present and surrounding all environments you record in. The difference between ambient sound and BG noise, in this context, is that BG noise is usually directional and comes from a source over which someone has some kind of control. Ambient sound is simply there and cannot be altered by human intervention. i.e. rain, thunder, ocean, highway traffic, parade, city din, etc. Its not evil, its just there. Everywhere. Why should you care? Because the ambient sound of a given location exists at a level that is usually the same no matter where you place your microphone. This means that when you raise the gain by 12db for Jane’s vocal to average -20db on the meter, not only have you increased the noise but you’ve also just raised the ambient sound in the recording by an equivalent amount relative to the dialogue, i.e. the roomtone in the two-shot of Jane will be 3 times louder than it was in the over-the-shoulder shot. You can, and should, choose a location exhibiting the lowest ambient sound, but you will still get this disparity when your subjects are at different distances from the mic.

    This might all seem overwhelmingly confusing but the practical bottom line is this: Get your mic as close to the person speaking as humanly possible WITHOUT getting the mic in the shot. What if its a wide shot and there’s absolutely no way to get the mic closer than 5 m/yrd from the actors using a 15 m/yrd boom?

  • 7) Just as objects get smaller as they move away from the camera, sound will appear quieter as the source of the sound moves away from listener, i.e. the camera. When watching a scene with multiple shots, you’ll notice the level of the dialogue from the close-up shots will always be louder and drier than the level of the dialogue spoken from a character relatively further away from the camera (as in a wide shot). This is how we experience sound in reality. A person talking to you from another room will not sound as proximate as someone talking to you from a foot away. So don’t be afraid of allowing some loss in loudness from a character moving away from the camera or whose position within the frame implies a distance further to the listener than alternate shots (not takes) of the same scene. What is much more important is intelligibility, i.e. that the dialogue be clear enough for the words to be understood. How do we achieve this If we cannot get the mic close to the actors for logistical/optical reasons?