audio for multi camera shoot
 

I hope this is the right forum to ask this question. We have been shooting an event every year with multiple cameras and one thing that is always an issue during post is the audio.

The sound is actually coming into a sound board so I was looking for the best (inexpensive) way to get a master audio track recorded (preferably straight to a HD).

Last year we used one of the DVX100s as the master audio (xlr from sound board to camera), however, that was still cumbersome as the event is long and required a lot of tape changes, not to mention it hindered camera placement.

Initially, I was thinking about getting an ikey or ikey plus (http://www.ikey-audio.com/ikeyplus.htm) and capturing the sound separately. However, it seems that someone that knows a lot about audio may have a better/more cost effective solution. Using a device to go from the sound board to a laptop with an external USB harddrive could be an option too.

Any suggestions would be much appreciated.

Also, is there an efficient way to "mark" all the cameras with some sort of audio spike that is inaudible to the human ear but that would show up in the NLE. It seems that would speed up syncing the cameras if there was something like that which could be fired off every time one of the cameras did a tape change.

You aren't live switching this event?

If not, why not?

what if you just did a free-run timecode?

I've noticed the iKey havn't got a XRL input,

Are you mic using XRL cable?

Wade,
We have always just recorded with 3 cameras and done all the capturing and editing in post.

Wade,
I still need master audio recording device.

SiuChung,
We have XLR capability coming from a sound board.

live sound
 

heres an idea....send XLR feed to a USB device with XLR inputs, like a Zoom H4, or a Firepod, or an Alesis IO, then just live capture with Vegas or some other NLE right to an external drive.....I have captured live with Vegas and it will just roll - record until the drive is full (200 gigs or so should give you HOURS of recording time)....I never tried to capture just audio, but I dont see why that would be a problem, just create a track and hit the record button...or, just capture with Sound Forge...be sure to do a hand clap or use some sort of slate at the begining so you can easiliy find synch in post....

On live shoots I usually do of concerts and things, audio is captured to a Protools system...

In your case, that is a good suggestion to an XLR to USB interface to capture audio right to the hard drive.

For sync, run the shotgun mic on each camera....sync up all the cameras in your timeline, then sync up your clean audio to that. Delete the dirty audio from the cameras...voila - you're synced.

Hi Brian,
Capturing audio direct to your NLE is a great idea. Some usb/firewire interfaces to consider are M-Audio's Firewire410/Mobilepre/Fastrackpro , Presonus Firepod/Firebox , MOTU Traveller, Tapco usblink.

Hi Brian,

Just something else to throw into the mix (and hopefully save you some time in the long run):

There is a potential problem when using multiple cameras and a single master audio track captured using some or the other external audio capture device.

Here is a link to a thread that I started some time ago:

http://www.dvinfo.net/conf/showthread.php?t=64559

I just read it know for the first time in a couple of months and I am quite impressed with the amount of testing I did at the time and the amount of technical information it contains that was furnished by fellow dvinfo users.

Basically there is an inherent problem when using multiple devices to capture video and audio seperately and then synching the whole lot together in post.

The thread appears to wander around a little but press on (even if you get bored). There is a lot of technical information in this thread that may save you weeks of frustration.

Although after rereading your initial post it seems that you have probably already come across this problem and either live with it or have found a solution that I could not in which case YOUR input would be appreciated.

Anyway - I see that there are a couple of relatively new posts regarding audio and video synch problems appearing so this will at least put my thread back in the spotlight and maybe save some people from pulling their hair out (or worse).

Regards,

Dale Paterson.

Sync!!
 

Hi Dale

You all seem to be hitting the target all around the "Bull's Eye" but just off.

The problem is the clocks not just the time code. What is needed is a way to lock the clock of the digital audio system to the clock of the Master Camera.

(I would use one camera on a wide shot as the Master Camera and if I could lock the clocks of all the other cameras to that but that will not be posible with the cameras in use. The least expenceive camers that I know of that will "Genlock to external sync are the Canon XL-H1 and the XH-G1)

Take a video out of the the Master Camera and use that as the reference for the clock in the audio chain. I hope the mixer, which is the device doing the audio A/D, can lock to an external video signal. That could be the a straight video signal or just the "Black Burst" portion of that signal. I.E. striping all the picture information off and leaving just the sync pulses.

I that high priced production facility they were locking the cameras and any audio A/Ds to house sync or black burst.

Hope this helps.

Grayson

Grayson;
Given a DVX and an iKey, or MiniDisc player, or similar device, how would you lock off the audio from the video master to the external, separate audio device?
I'm curious to know how you'd accomplish this.
It's very important to be aware of what tools the original poster has available.
Your last sentence:
has me a little confused?

Good Morning,

Grayson - thanks for the response. After all my frustration and testing I realise that it is the device clocks and not a timecode issue.

Douglas is right - it does depend on the equipment that you have i.e. prosumer or professional.

Having said that even being able to synch one camera (the master) using genlock for example to the external audio capture device would still not solve the problem i.e. the other cameras would still be doing their own thing. The only way that this can work is if all devices are genlocked to each other.

Put it this way - using only one camera (without genlock) you could take a master audio out from the mixer (or whatever audio capture device) to the master camera and the audio on this camera would be in synch with the video on the tape in this camera BUT the other cameras would still be / go out because of their individual clocks.

Anyway - let me just say - it was not my intention to hijack Brian's thread - I just thought that it was something that he should be aware of (if he was not aware of it already). This topic I have beaten to death in my own thread and to date have not found a solution other than using reference tones at the beginning and end each cameras tape or using clapper boards and manually aligning the audio in post (I use Vegas). How's that for hitech!

On the other hand if anyone has come up with a REAL hitech solution for solving this problem using equipment that does not have genlock then by all means your solution would be more than welcomed.

Regards,

Dale.

Dale, since you're using Vegas, it's exceptionally easy to bring the cams to sync'd audio, it just takes a few seconds, particularly on a multicam shoot. We're doing more and more in HDV these days, and less and less in Beta, in fact, almost nothing. So, nothing is locked. Long and short form media all falls nicely into place, as even over long form, you're at worst off by a few frames after an hour, and as you've figured out in Vegas, they sync fast. You've just gotta be sure the audio is all at the same sample rate, same bitrate. If it's not, then between the differing clocks and sample rate, it's a lot of cutting.
I start by putting all cams on the timeline with my master audio enabled, and all cam audio muted excepting one. I loop a small section that contains a percussive, and you'll know immediately which way you need to go. Disable "sync to frames" in Vegas so you can nudge audio in samples rather than in frames or half frames. Doing this while looping will allow you to immediately hear whether you're in or not. If you hear chorusing, echoing, flanging, or any other "ing" then you're not in sync. It should be invisible, except for a volume drop when you mute either the master or wild audio track.
Once one track is sync'd, then mute it and move on. Of course, VASST has syncing tools that can do this for you, too, but frankly, I find the method I've been using for a few years to be more intuitive for me.

Hello.

That is the only way to do it.

For my tests it was easy. I started three cameras and the external recorder in close proximity to each other and recorded (generated) five pager type beeps and then just toward the end of the hour (tape) recorded (generated) another five beeps. When all of the tracks were on the timeline it was a cinch to get them all in synch BUT in a live situation you just have to hope that somebody coughs or claps their hands or accidentally hits the mic stand at the beginning and near the end of each hour so that you have your percussive reference points!!! Using voice or music and listening for that flanging sound is not exactly an exact science particularly when you have four devices running each with there own agenda!!!

What is needed is either:

Some sort of device that transmits a clock or timing signal or something like that and receivers that can be plugged into our FX1's/Z1's or any other make of prosumer camcorder via LANC or something like that to ensure synch of all of these devices (and you would need an audio capture device that could work off of these receivers as well). I don't think that any of the prosumer cameras gives you the choice of whether or not to use an internal or external clock though.

or

'Capture' software (Douglas - I know you don't like the word 'capture' - see below) that somehow uses the internal clock of the PC being used for 'capture' as a reference and resamples everything that is 'captured' from any source on the fly i.e. as it is being 'captured'. That could solve the problem and would be an elegant solution. This would be different from 'copying' the DV file on the tape in the camcorder to the PC which is actually what we are all doing (see - I know the difference).

I know that over an hour there is only a small margin of error but it makes a huge difference at the end of each hour especially when you have five tracks on the timeline and remember that you are also (in my case) dealing with the timing difference caused between the difference between the speed of sound and the speed of light i.e. the difference between the distance of the subject to the 'master' camera and the distance between the subject and the mic that goes to the mixer i.e. the camera might be thirty metres away from the subject but the mic is not more than a couple of centimetres away from the subject.

Then there is also the problem of Vegas previews being slightly out of synch with the audio anyway (unless you preview with 'Print to Tape') so even attempting to synch visually is impossible.

The fact that this issue still exists in this modern day and age still amuses me though. Forget about audio sample rates and the like etc. etc. etc. Logically 25fps should equal 25fps should equal 25fps (get the picture) and the audio should follow but 25fps might average 24.999fps over an hour on one device and 25.001fps over an hour on another device and there is your difference (this is making my brain go fuzzy again).

Basically it is a pain but just think how boring it would be if everything just worked!!!

Regards,

Dale.

I do quite a lot of multi-camera shoots of live music events - the system I use is to record stereo audio onto my main camera - and if I can, the same feed to at least one other. The others record room sound - it actually is quite useful for adding 'crowd noise'.

You can do a sync point at various times if you want to, but there is always a small amount of drift in a long continuous shot.

I use premiere pro to edit, using the multicam feature which for me works very well. However, I tend to edit a song at a time, so that any render issues don't suddenly mean having to go away while 4 60 min sequences sort themselves out.

My sync is usually quite simple - lots of little tricks, like the drummer being the clapper board, lighting effects that can be used as the sync point, and the audio tracks themselves. It's quite simple to zoom in on the audio waveforms and slip the tracks until the waveforms match. This system works really well. I think sync is often overstated, and really complicated timecode sync isn't required. Lip sync is the obvious aim, and getting it right isn't that difficult.

Hi Paul,

You are right:

The problem comes in when you are not aware of the possible pitfalls (or that they even exist) of using external audio devices for audio capture or using multiple cameras for long shoots.

Picture this: my first (paid) multi-camera shoot (for eventual broadcast), notebook, PA system and mixing desk supplied, set up and tested, six wireless mics, lighting all set up, ready to go. The event ended up being just over about two and a half hours long. Imagine my surprise after getting back to my studio, downloading all of the footage from each camera tape, methodically synching the beginning of each camera tape / track with its right place on the master audio track that came from the desk, only to see on the video tracks toward the end that the last speaker / presenter had already packed up his things and gone home (and was probably sleeping) by the time his speech on the master audio track ended (I exaggerate of course but you get the picture).

Personally I think it should just work BUT although I did get a bit frustrated and add a couple of years onto my life at the time I did get a certain sense of satisfaction out of testing and finding out why this clock thing is an issue at all and that's the only way I learn!

Once again - it was not my intention to hijack this thread. I just think that it needs to be mentioned that if you think that you can go out and buy three identical cameras and an external audio recording device and expect them to all be in synch with each other over a period of time once pulled up into your NLE you will be sadly dissapointed BUT there is a workable although not elegant solution to the problem which has been detailed by everyone above.

Regards,

Dale.

Dale, if you're interested in working in a mid-high end broadcast environment, genlock is critical. Just because the small format cams don't offer it, don't fool yourself into thinking it's not necessary; it most certainly is.
clocks are clocks. There is no way to make two clocks run at exactly the same frequency for any significant length of time and have it be affordable. You can get devices that measure the difference between the clocks and constantly adjust the median as a result of those measurements, but that's about it. It's a problem that has been part of the world since the beginning, and is likely going to be a problem for some time to come.

Sorry Douglas, as always you are right.

What I should have said is that you cannot expect to go and buy three (consumer or prosumer) cameras and some or the other (consumer or prosumer) audio capture device etc. etc. and expect them all to operate in synch.

However (and this is not meant to be a joke although the question may come across as such) BUT how do we know that the time of day is actually the time of day??? I mean just think of how many clock crystals there must be in the world all suffering from the same problem and yet, somehow, world time seems to be able to stay in synch for years and yet my Sony's can't stay in synch for longer than a couple of minutes!!! I'd settle for two or three hours!!!

Dale.

Dale, the answer to your question regarding time of day is:
the Atomic Clock in Boulder, CO. By order of law, we know what time it is. :-) Amazing...politicians can't balance a budget but they can create a world order of time.

On a significantly more serious note, if your cams are literally out of sync after a couple minutes, then you have something incredibly wrong with either your cameras or your workflow or both. Even mixing Canon, Sony, and Grass Valley cams on a timeline, I have no issues with sync whatsoever over lengthy periods of time. Shooting both Z1 and V1U camcorders for a 24 hour event, all 24 V1 tapes and Firestore from the Z1 are of exact timing once the tape-changeout and hard drive changeout is extracted from the work sequence.

If you've got cameras with external TC in you can use GPS to lock but that only avoids the problem of running cables.
I have to agree with DSE, I've shot long multicam with a Sony and a JVC camera and not had any significant sync drift and when I do it's a piece of cake to compensate in Vegas. If you're getting out more than a few frames in an hour something is seriously wrong.
I record all location audio into our Edirol R4, can't undertand why people will spend zillions on cameras and baulk at spending half as much on good audio gear. Being able to record your audio at 24 bit can be a real plus, having a few spare tracks to record the room from a few extra mics of your own can also make a huge difference to the final sound of the show. Feeds from desks are almost always dry and mono, having a bit of stereo 'feel' in the final mix helps put the sound back into the space of the venue.

I have little problem in syncing the following cameras from three cameramen, PD150, two FX-1 and a GS400 and occasionally a TRV50. In a two and half hour show there is no real sync issues with the FX1's or the PD150, occasionally the GS400 and TRV50 may be out a frame in this time. Unfortunatley for these two consumer cams it is not consistent. It may go forward a frame in the first hour and then back a frame!!!!. I am sure this is well within the spec of 1/3 a frame but the drift is not always consistent, depends on when they are started after a tape change? Video sync has never really been a problem over these times. As mentioned earlier I sync/mix audio in Vegas though most of my editing is with Edius Pro3. Most of the time I don't use the audio from these two consumer cams anyway but it is there if needed. For interest, the PD150 and one of the FX-1' have shotgun mics and the other FX-1 one uses a AT825 single point stereo. Main audio comes from one of the shotguns with the stereo feed providing some "air". I too would love to find a nice easy way of getting sync free audio but have resigned myself to working as I do considing the equipment we are using. I seem to spend a lot more time on the audio these days than editing the video!!! Camera outputs are better, NLE's are more capable and faster but getting clean audio to match the quality of the video is another thing alltogether!!!

Ron Evans

"I seem to spend a lot more time on the audio these days than editing the video!!"

Spot has often said that audio is half or more of a motion picture. Perhaps he will need to append that in the near future to say that it is always MORE than half the production.

BTW, the atomic clock is adjusted every year to account for changes in the Earth's orbit. I remember that 2004 had a bigger adjustment due to the effect of the Tsunami in the Indian Ocean.

If one of your cameras has a clock faster than the other, just put it on a fast plane and time dilation will help correct the difference.

Seriously, I'm happy that I mostly work on productions where only a few minutes at a time need to match. I don't know how I would match up 24 hours of footage from multiple cameras. I doubt I could even keep the tapes organized!

Actually, Spot has a copyright on the phrase "Audio is 70% of what the audience sees..." I do believe audio is much more than half the image. It warms my heart that folks actually heard that. :-)

Hello again,

For the record, if I remember correctly when I was doing all of my tests in April this year, there is a one or two frame difference per hour between my three cameras i.e. they don't go out of synch after a couple of minutes but you can just start hearing that 'flanging' from the audio tracks at about thirty minutes or so and onwards.

And just when I was ready to (once again) put this subject to rest somebody goes and says something to draw me in again:

Bob. Please could you explain this further for me:

Regards,

Dale.

Hi,

Sorry, but I am really interested to know what Bob Grant meant by using GPS with cameras that have external timecode capabilties (see previous message).

Regards,

Dale.

GPS works by very precise time-of-day signals. With appropriate GPS receivers a time hack can be derived that fed into a camera would allow all the cameras in a multicamera shoot to have exactly the same clock setting with their clocks locked to the GPS signal. I seem to recall seeing something about Ambient selling equipment that can do that but I don't recall any detail

Steve, thanks for that.

Excuse my ignorance but does GPS not depend on line-of-site? You know - like a satellite phone? I mean the equipment (receivers or at least their antennae) has to be able to see the satellite(s)? As you can maybe tell I don't know much about (or have a portable) GPS!!!

Edit (Sorry - I forgot to Google before posting the above question):

This comes from http://electronics.howstuffworks.com/gps1.htm:

This is very exciting BUT correct me if I'm wrong but (obviously) you need cameras than can accept and lock to an external time code (in) and have a clear view of the sky? Cost wise are you not back to square one i.e. professional equipment with Genlock?

Regards,

Dale.

Yep ... and I can't imagine why it would be needed for any except very special situations and it certainly would be a budget solution. Genlock is something that I think a lot of people are confused on anyway - its main purpose is NOT to sync audio and video or to lock the clocks of two cameras together so they show the same values for Time of Day, though that might be a fringe benefit. When two cameras in a multicam shoot are both recording to their own tapes which will then be cut together in post, genlock is irrelevant. It's needed when you have multiple cameras and need to live switch them in real time. As you know, the video image is formed by scanning a raster. If you cut from one camera to the other, the position of the scanning point should be in the exact same position in the frame in both cameras otherwise there'll be a jump due to the momentary loss of sync in the resulting image. For example, if the scan jumps from being at 30 pixels in on line 22 of the frame on camera 1 to be at 400 pixels over on line 300 on the frame from camera 2, a noticable glitch will occur. Genlock locks together the horizontal and vertical sync sweeps of both cameras to the same clock signal, insuring this doesn't happen. Of course, it's also needed in on-line post editing, insuring the frames of the multiple playback VCRs are similarly synced together. In a broadcast studio it's vital, providing a house clock signal that all video sources sweep in lockstep to. But for typical DV multicam shoots where the source material is in the form of separate files that are cut together in post in an NLE, it's not needed.