Seriously Remote Recording (Sound Devices 744t)

[dkatz42]

I just got back from our video shoot in and around Malambanyama, Zambia. Eight days in the bush and 19 hours of video later, I thought I'd follow up on my experiences with the Sound Devices 744t.

The executive summary--a very fine unit, which worked extremely well, highly recommended for folks that want serious recording capabilities in a very portable package.

The gear:

A full size Sony HDcam (uncompressed 1920x1080 HD, glorious indeed)
A Sony HDV camera (MPEG2, for late-night street shooting and backup)
Sound Devices 744t
Sennheiser shotgun mic (the model escapes me)
Lav mic (the video guy owns it, very nice)
Rode NT4 stereo mic (for music recording)
MDR7506 headphones
Zeppelins, booms, windscreens
Pile of Lectrosonics wireless transmitters and receivers
Batteries, chargers, cables, spares


The setup:

HDcam (at 23.972fps, ND) in Record Run mode, feeding timecode via wireless to the 744t, set to External TC/AutoRecord
Shotgun/boom into 744t CH1
Wireless lav into 744t CH2 (for sit-down interviews or simultaneous translation)
Wireless feed from 744t CH1 out to HDcam for cue/safety audio
Second wireless receiver from lav to HDcam for cue/safety audio
Wireless IFB from HDcam to video guy's in-ear monitor


The scenario:

A combination of sit-down interviews (almost always outside) and event/background shooting. Events. Lots of B-roll.


All the wireless stuff was a godsend. I was self-contained, with a PortaBrace bag containing the 744t and three Lectrosonics boxes, headphones, and a boom. In cases where the camera had to sit back at some distance, I could get in much closer with the boom. Things seemed to work well up to about 200 feet, where the timecode feed started to fall apart (more about this later.) My video guy was untethered from his camera, handy in the cases where it was on a tripod and he needed to chase away chickens and guinea fowl.

My job was to point the mic and try to stay out of the shot, which I managed to most of the time. Because of the timecode setup, the 744t would automatically start recording when the camera was shooting video. This was *very* helpful, as I did not have talkback from the camera guy. Ten seconds of preroll was tacked onto each clip (programmable on the 744t.)

I did primary recording to a 4GB flash card (in order to preserve batteries and so that I wouldn't have hard drive heads banging around as I careened around the location.) 4GB proved large enough to carry a day's worth of audio at 24/48K, though we were tight a couple of days. At the end of each day I copied the contents of the flash card onto the 744t hard drive (the unit automatically keeps track of what's been copied, in case you want to make safety copies at several points in the day.) I then plugged the 744t into a 12" AlBook via firewire and copied off the day's work. For paranoia's sake I then copied the files to an iPod (from which I had wiped all of the iPod software and turned it into a teeny external hard drive.) Then I'd wipe the CF card for the next day's shoot.

The three copies of audio came back with three separate people...

A single 6000mAh battery was enough to drive the 744t through an entire shooting day, even when not being careful to turn it off, and even with the LEDs cranked up to blinding brightness so that I could see the levels in the direct sun. I never had to use my spare battery.

There were very few problems with this setup. About the only minor irritant is that the 744t's algorithm in AutoRecord mode is "when in doubt, record" (a good thing, generally.) Whenever the timecode would glitch (such as when the camera was powered up, or when I wandered too far away from the camera and the radio link became dodgy) the 744t would shift into record briefly. As a result I have several hundred small BWF files to clean up. A slightly smarter algorithm would help, since the 744t has enough preroll buffering to be able to spend a bit more time deciding before starting to record.

Switching batteries requires something long and narrow (like a screwdriver) to press the release button, particularly with oversize batteries. While I understand the paranoia, it seems as though a recessed button or a two-finger system would be better.

Happily we actually had electricity this time, as we had a lot of batteries to recharge (the last trip, in Kenya, we had to buy a generator and run it for hours every night.)

About the only thing I'd do differently is to have talkback from the camera guy (he could hear me, since I had the boom.)

An amazing experience, all in all.

[Remoteness]

Very cool report. Did you take any pictures of your rig in action?

[dkatz42]

Originally Posted by Remoteness

Very cool report. Did you take any pictures of your rig in action?

There were three still cameras in the hands of our team, so I'm sure that there is at least one of the mzungu with the greying pony tail holding a dead cat on a stick. I'll post if I find one...

[FilmDingo]

Did the Rode NT4 get any use? Did you have a wind rig for it? Did you do any ambience recording, if yes what was your go to setup for that.

Thanks

[dkatz42]

Originally Posted by FilmDingo

Did the Rode NT4 get any use? Did you have a wind rig for it? Did you do any ambience recording, if yes what was your go to setup for that.

Thanks

I used the NT4 on purely musical stuff (some with video, some audio-only.) I'll post a sample once I get a chance.

I didn't get a chance to do any real ambience recording, as we were shooting pretty much dawn til dark (I did some with a couple of NT5s in Kenya a couple of years back.) Plenty of B-roll audio with the Sennheiser shotgun tho.

[dkatz42]

BTW, I ended up writing a program to pick apart the jumble of BWF files resulting from the shaky autorecord algorithm. I came back with 964 files, of which perhaps a bit more than 300 match with actual video shots (there are a bunch of blank ones because I was recording two tracks with only one mic plugged in, and then there were a number of audio-only files from musical performances, but there were lots of short nonsense files.)

Basically the program pulls apart the BWF files, pulls out pertinent information (timecode, scene/take numbers, frame rate, sample rate and size), walks all of the data once (searching for nearly-silent files; the max level is written back to the BWF metadata so that the data doesn't have to be read again) and performs some heuristic juju in timestamp analysis to make an edumacated guess at which files are good and which are bogus, so that my video guy has an easier time pulling in the audio.

I now know way too much about BWF file structure.

It also allows the start timecode to be offset by an arbitrary number of frames, as it appears that there may be an off-by-one bug in the 744t external timecode interface.

2200 lines of C in three days. I knew my day job would come in handy some day...

I'm going to draft a note to Sound Devices soon, with some suggestions on how they could improve the autorecord/Ext TC algorithm to make this easier.

[Remoteness]

Brilliant!

[dkatz42]

Originally Posted by FilmDingo

Did the Rode NT4 get any use? Did you have a wind rig for it?

Forgot to answer the second half of this.

Yeah, I had the full Rykote treatment for the NT4--it probably cost as much as the mic, but the Rode SDCs are utterly useless otherwise if there's even a breath of wind, and virtually all of the recording was done outside.

Being slutty and all, I'll look into the Pearl mics next. Hopefully they'll even fit my Rykote zeppelin!

[dkatz42]

I've attached a field recording (literally) I did with the 744t and the NT4. The singers are a group of about 20 kids, aged 8 to 18, from Malambanyama, Zambia. It was recorded in the open, under the stars. I still get chills when I hear it, but you can tell me if that feeling translates.

The song is religious, presumably sung in one of the two major dialects in Zambia.

The lack of health care in rural Africa is apparent in the recording...

The MP3 encoding did weird things to the stereo imaging, but it's still decent. The singers were arranged in a half circle around the mic, about 10 feet back, which was in a zeppelin but no sock. 24/96.

I added a touch of reverb (it was *very* dry, as you can imagine), some gentle compression, and a tiny tweak of EQ.

You can hear the crickets in the quiet parts, along with the sounds of impatient kids, late for dinner after a long day...

[Aston]

Just curious, what wireless system did you use to link the 744t to the cameras? Can this be done with the 702t as well? Also, do you know if the 702t can program a 10 second pre-roll as well?

[boojum]

Originally Posted by dkatz42

It was recorded in the open, under the stars. I still get chills when I hear it, but you can tell me if that feeling translates.

Same effect here. Marvelous singing!

[boojum]

Originally Posted by Aston

Just curious, what wireless system did you use to link the 744t to the cameras? Can this be done with the 702t as well? Also, do you know if the 702t can program a 10 second pre-roll as well?

The ten second pre-roll is so on the 722 and the 788T. I suspect it is an SD standard.

[Arthur Stone]

Originally Posted by boojum

The ten second pre-roll is so on the 722 and the 788T. I suspect it is an SD standard.

Yep - adjustable up to 10 secs on the SD702.

[Jaymz]

It's samplerate dependent, worth noting.

44.1/48khz = 10 seconds max

96khz = 5 seconds max

192khz = 2 seconds max

[npulsipher]

Originally Posted by dkatz42

I still get chills when I hear it, but you can tell me if that feeling translates.

I didn't get the chills but it does sounds 100% awesome and made me smile.

[dkatz42]

Originally Posted by Aston

Just curious, what wireless system did you use to link the 744t to the cameras? Can this be done with the 702t as well? Also, do you know if the 702t can program a 10 second pre-roll as well?

Wow, you blew the dust off of this post. I didn't realize how long ago it was...

Looking back through my notes, the wireless gear was primarily Lectrosonics UM400s and R1As. I suspect those models have been displaced by shiny new baubles since then.

All of the linking was done in the audio realm, so it's just a small matter of the right adapter cables. The 702t should work just the same; it's the "t" that's the key for the timecode stuff.

Shameless plug: the shoot was in support of the Firelight Foundation, which funds grassroots organizations in Africa that support children made vulnerable by HIV/AIDS and poverty. Firelight Foundation : Communities Changing Children's Lives

[dkatz42]

Originally Posted by boojum

Same effect here. Marvelous singing!

Thanks. It's not quite the same without the starlight under the darkest skies I'd ever seen, and the emotional overload of what we'd seen and experienced, but those kids sure could sing.

[Arthur Stone]

Originally Posted by npulsipher

I didn't get the chills but it does sounds 100% awesome and made me smile.

+1...a nice sense of presence.

[RScott]

Very nice work! I love the recording and it sounds like a great field project, quite the experience. What were you using to transmit timecode? A wireless audio pack or something more robust?

[dkatz42]

Originally Posted by RScott

Very nice work! I love the recording and it sounds like a great field project, quite the experience. What were you using to transmit timecode? A wireless audio pack or something more robust?

Just a Lectrosonics wireless pair from the camera to the 744t, spending a little time tweaking the levels at each end for stability. The range was more than adequate and the link was solid.

We had four separate wireless feeds going; the good news was that there was basically *no* RF out there, so no interference issues, according to the spectrum analyzer on the receiver (though presumably there was in the cellphone band; rural Africa skipped wired telephony entirely, and on the first trip our generator was very popular for recharging phones, as it was the only electricity for miles).

[RScott]

Glad to know that wireless audio transmission of timecode works. I've been debating using wireless sync with Ambient Lock-it boxes or a similar black-burst generators but if I can get away with using my current wireless packs for that purpose, that may prove to be simplest for low-end shoots. Perhaps not as a locked-on-all-the-time solution but as something that the camera op can bump every so often and ever battery change to ensure sync.

[Rolo 46]

I used a separate TC sender from the camera to recorder on different spectrum from my radios
UHF diversity for talent
VHF diversity for camera link
TC data link
Its a lot of RF in and out of a handbag ,but it worked.
Murder when globe trotting though....