Anyone who has ever been involved with location sound recording will know that there are occasions when it doesn’t quite go according to plan. Whether it’s uncontrollable background noise, faulty equipment or just plain user error, things can and do go wrong! We spoke to a number of professionals to ask them what their worst location sound disasters were. Here’s what they have to say…
Paul Maunder
A few years ago we needed to record a voice over for a video. The artist was not able to get to a studio due to other commitments so we agreed that we’d bring some portable sound kit along to a hotel they were staying in so we could record the VO there. The kit consisted simply of a location recorder, a pair of headphones and a Sennheiser MKH416 shotgun mic because that was the kit we had available at the time. I wasn’t able to attend so I briefed my colleague on a few points, such as minimising background noise and trying to find a spot which didn’t have too many reflective surfaces. It never occurred to me that I might need to mention not to point the mic in the wrong direction but unfortunately that’s exactly what he did! When I got the audio back, it sounded like she was speaking from the other side of the room.
“How close was the mic?”, I asked. He assured me that it was about 2 feet away from her, pointing at her mouth from just above. I couldn’t understand how it sounded so distant. The mic had been in a Rycote windshield and it turned out that he’d pointed it backwards, making the audio totally useless. He asked me if there was anything I could do to salvage it, to which I replied “Not unless someone has invented a stupidity inverter”. It didn’t go down well.
Russ Hughes
I did a two day shoot at Abbey Road. My client sent me two of their own US pro crew to help on cameras. Mike Thornton was on sound. We had timecode and everything running via wireless from the cameras. All because their pro crew had said it was the only way to shoot. Their amazing system didn’t work and they both recorded footage mute so we didn’t even have audio for sync. It took me hours in post to shot match to get sync trying to match lip reading.
Jay Whittaker - Sound Editor/Dubbing Mixer
Not so much a location sound disaster rather an unnoticed side effect of production sound not listening too intently to what they were capturing. Working on a low budget horror feature, the scene being shot was in a hotel room, and in post not only was I trying to fix the usual: de-noise passing foot fall from the corridor, a horrendously rattly air conditioning unit, a fridge - which I’m told could not be switched off - but the couple in the adjacent room had taken it upon themselves to get amorous at the same time the crew were shooting. Unbeknownst to production sound their shotgun mic was capturing all the dialogue beautifully, but it was also capturing nextdoor's “goings on”. iZotope RX 8 had never been so overworked.
Anonymous - UK Location Sound Recordist
The usual challenges we have are becoming very common sadly as DOP’s, Directors and camera operators want to make a name for themselves and shoot exciting, experimental and visually stunning content that is often compromising the sound department. Often the Costume designers don’t want to compromise either. So we can end up not being able to get a boom in due to “creative” lighting so radio mics are often relied on when costumes are noisy, like armour, waterproofs, or stiff starched clothing with tiny ties etc etc. Sounds like a bit of a moan but when high end tv egos come into play, nobody will compromise so often ADR is the only real solution.
Paul Maunder
Another story which springs to mind is the one and only time we actually lost any footage. We were shooting a corporate video using two cameras, a close and a wide shot. It was fairly low budget so there was no dedicated sound recorder. The radio mic receiver was connected directly to an XLR input on one of the cameras. The other camera was just recording audio from its top mic. A member of the crew accidentally formatted the camera card before I’d transferred the footage off it. Of course, it had to be the camera with the proper audio, so we were left with just one camera with crap sounding audio from the small shotgun mic mounted on top. There was no scope for a re-shoot so I was forced to use this audio. I was able to improve it quite a bit using de-reverb and de-noising in iZotope RX but it still didn’t sound great. I still remind my colleague of this story to this very day!
Damian Kearns
Despite the scope of breathless stupidity printed (or not) in the many doomed recordings I’ve encountered, there is one project I had the misfortune to tackle that I recall with more retrospective anxiety than all others.
It was August, 2001, and at the time, I was on staff at the Canadian Broadcasting Corporation. TV seasons were seemingly etched on stone tablets back then and since it was a month before the commencement of the 2001-2002 Fall broadcast season, I was “cooling my heels” some days; spending the odd shift covering off vacationing colleagues or repairing stuff for CBC Archives. Summertime was a welcome respite, since my life was not my own during the Fall, Winter and Spring post production months. I didn’t know it at the time but a few weeks later, the predictable autumn workload I was anticipating would be completely upended due to the events of September 11, 2001. But that’s another story. This story is about salvaging a wreck.
The Open Door Conversation Behind Closed Doors
One day, I was playing with gear and waiting for lunchtime to come when the studio door opened and a production member from one of the series I regularly worked on came in and closed the door. There was a problem with an episode. Could I possibly help?
The scenario was bleak: They’d actually been kicked out of a local high-end film to video transfer facility because their dailies audio (‘dailies’, also known as ‘rushes’ is the term for footage shot on set each production day), recorded to digital audio tape (DAT), couldn’t be synched to the film footage. They tried other places around town and no one could seem to put sound to picture. As it turned out, the time-of-day SMPTE timecode (an address track used to tell what time of day the footage was shot) we expect to be recorded with location audio wasn’t present on the DAT’s. Also, the clapper they used visually was this new visual clapper that was barely ever actually physically clapped (critical for sound sync), owing to the ersatz notion on set that the visual timecode was all that was needed. An un-clapped clapper was one thing but as it turned out, the visual timecode was almost never easy to read and much of the time completely incomprehensible, due to the angle of the clapper and/or lighting issues inherent to shooting outdoors, documentary-style. So I had no clapper and little to no timecode and no visual cue and I had 20 hours of tape to synchronise somehow, maybe.
I enquired about a reshoot. It turned out the documentary was about a tribe, the Maisin people, from Papua New Guinea coming to Canada’s west coast to meet some local indigenous people, the Stó:lō, as a sort of cultural brain exchange. The Maisin were already back home in Oceania, about 10,000 km’s away. So yeah, I like to call these situations a Fuster Cluck.
The film footage had all been transferred ‘without sound’ (MOS) to the Sony Digital Betacam format. It was the best video format we had at the time and all our audio post studios had a machine for playback, mix reference and restripe (aka. laying back to tape). I placed one of the Digi Beta tapes I was given into my DVW-A500P Digital Betacam machine, what I thought was the corresponding DAT into a Sony PCM-7030– my favourite ever R-DAT machine- and took a look. No SMPTE timecode on the DAT at all. The Digi Beta had 29.97 fps non-drop frame time-of-day on it, likely generated by whoever transferred the film to video, by guesstimating what the time of day on the ‘very hard to read’ visual clapper might have been displaying at the top of each take.
This Is All Very Technical
Since the DAT had no SMPTE 29.97 non-drop frame timecode to match the Digital Betacam tape, I used Absolute Timecode (ABS on the 7030, often also known as ATC). This typically about 30 frames per second (fps) address track timecode is laid down by default on non-professional DAT recorders during the formatting process on the initial recording pass of any digital tape (video or audio) to help the machine figure out where it is and how long a second is. It starts at the beginning of the tape and ends when recording ends. It’s subcode data that is used to reference when and how quickly to playback a tape in the absence of SMPTE timecode. For this job, I was going ‘under the hood’, so to speak, using ABS to create SMPTE.
On the 7030 I was able to change the DAT’s ABS to 30 fps non-drop frame timecode because the 7030 had the ability to infer SMPTE from ABS in its ‘Tc bASE’ setting. I would have to capture in and out (aka start and end) timecode from the Digital Betacam and the R-DAT, manually, write the numbers down, and then using a little timecode calculator program I found on the internet, try to come up with a value for the varispeed I thought I might need to operate at. The offset between the DAT’s timecode and the Digital Betacam’s SMPTE could be captured by the 7030 using the method detailed in the manual (pictured below). Any varispeed would be a constant through every tape, I hoped, as the location recording devices were theoretically operating the same way through the entire production.
Now comes the fun part. I scanned the Digi Beta for a long visual stretch of dialogue and once I found one, watched it a few times, trying to lip read until I felt I kind of knew what they were saying. I also noted any sort of hit or other transient event like a table impact near the start and end of the film clip. I found something near the beginning and then something else near the end that I could use as absolute synch reference points. I marked the two numbers down. Then I listened to the DAT.
After listening for a while, I found what I thought was the matching dialogue for what I had lipread and the right two top and tail synch points and marked down the ABS timecode for each. I was treating it as 30 fps SMPTE since the frame rates were nearly identical. I plugged my numbers into the timecode calculator program and came up with a varispeed value of about .1% slower which was likely the result of 24 frame film being pulled down by .1% to match video reference, which is 59.94 Hz, rather than the internal typical 60 Hz sync that North American machines run at if powered by our 110-120 volt electrical outlets, without external sync. The film to video conversion would add visual frames to increase the frame count from 23.976 to 29.97 but the timecode would be at 29.97 non-drop frame, pulled down by video sync from 30 fps which was typical for film-to-TV projects back then. After these tapes were ingested by the editor, the editor would work in drop frame mode to ensure the show’s running length was precise. Drop frame timecode drops 2 frames per minute except for every 10 minutes, to bring 29.97 timecode back into time alignment with the clock on the wall. This is why we use it. Otherwise an hour at drop frame would be 3.6 seconds longer.
Since the 48 kHz,16 bit DAT tape would be running at a different sampling rate to my 48 kHz Digital Betacam (48 kHz is only applicable if running on speed, and remember and I was slowing the DAT down .1%), I used analogue audio out from the R-DAT into the Digital Betacam. Digital machines don’t like trying to interpret the wrong sampling rate coming into them and a lot of noise and hits happen if this is the case, as machines struggle to sync to the offspeed digital bitstream. Analogue would be the way to go here since once the audio goes out through D/A conversion, it’s audio that can be resampled by the next input A/D. I then took timecode directly from the Digital Betacam to the 7030 to supply the timecode reference. Varispeeding was simple because all I had to do was switch the 7030’s sync reference from video reference to internal sync, hit the ‘Chase’ button and the external timecode I was receiving from the Digital Betacam would cause the machine to slew to match the timecode frame rate.
It worked! And the timecode sync method I was using held for the entire 20 hours of tape. The hard part was finding each bit of sound to go with the various film dailies, then grabbing a timecode offset for each clip, then watching to see if my offset was tight enough for lip sync, then potentially playing around with nudging the DAT’s offset frames at a time, until things looked right.
Recording to the Digital Betacam was easy enough. DAT audio is only 2 tracks and a Digital Betacam has 4 tracks. Each tape had about 40-50 minutes of footage on it so in an 8 hour work day, I could maybe restore two tapes. When production heard the first tape, they were very grateful and platitudes were heaped on yours truly. They were finally able to start picture editing. I was a hero.
From Hero To Zero
As the days wore on and the impatient cries for me to speed up intensified, I had to push back because I was working as quickly as possible. It took two weeks to put all the sound to picture, faithfully and accurately. People were getting on my case about their budget and things like that but my pushback was always “You wouldn’t have anything to work on if it wasn’t for me.” The picture editor was upset by the slow pace of my progress but it’s worth pointing out that none of them was ever in the room with me except on that first day so only one other person ever had any idea how painstaking this all was. I had no clapper, no timecode, was reading lips and then there were scenes with no dialogue I had to sort out! For any Mit Out Sound (MOS) shots! I managed to find wild tracks of location recordings to sync to those too so I mean, everything in this show was covered. Sound was back with picture.
After two weeks I received a hearty handshake and a bottle of whisky from my friend in production who knew exactly what I had to do to save the project; a project that had effectively been deemed DOA until it landed at my door.
My colleague mixed the episode about 6-8 weeks later, after the picture was locked. It took him three days. Everything was totally in sync and the location sound was actually pretty good; the audio I placed on the MOS shots meant no searching for fill or location ambient sound effects. Editing was easy. He got his credit but when he showed me the credit roll, my name was absent. I was furious. I went to find out why.
Production’s line was something to the effect that they didn’t want to “underscore the technical issue” I had solved by crediting me with it. This really pissed me off. I’d spent three times longer on the job than the post mixer had but I walked away content in the knowledge I had done something I’d never seen done before: I had pieced a show together with no timecode or sync audio with which to work. Though I didn’t receive on-screen credit, it is a story I have retold on several occasions and have now, finally – no tape sync necessary - put to print.
The moral of the story is: Problems don’t solve themselves. If you walk onto a set without a fully realised technical scheme, you can expect to pay the price later on in post, if you’re lucky enough to find someone with enough experience to fix your problems.
In Conclusion
So there you have it. A collection of stories of bad experiences with location sound. Of course, we all learn from our mistakes but sometimes the issues are caused by the errors of others, and that’s often impossible to control. Needless to say, the problem always has to be fixed in one way or another, whether that’s a reshoot or some arduous process of manually re-syncing or having to remove all kinds of noises in post. Let us know your worst location sound disasters!
Photo by Kyle Loftus via Pexels