MD recorder time accuracy?
 

One of the nice things about recording audio on multiple camcorders is that it is pretty simple to synchronize afterwards. The video makes it easy to sync off a flash, and the time accuracy of the audio tracks means that once synched, the audio tracks will stay in sync for very long periods of time. This means that you only need to sync the tapes once so long as you keep the cameras filming continuously throughout the event. With mini-disc audio I suspect that you have a little more difficulty synching up the audio due to the inability to visually sync. Do you have tricks that you use to make this simpler?
Perhaps even more problemmatic would be if the minidisk had some longer term drift relative to the videotape shot with the camera. This would mean that you could be synched in one spot,
only to find that you were no longer synched later in the event. Is this a problem with minidisk recordings that have been captured onto your computer via the analog output of the minidisk? Do some minidisks have this problem, while others don't?

All devices that aren't actually synchronized to a master signal will drift to some degree. This will vary due to many factors, including the ambient temperature during both recording and playback. So there's no hard rule, even for the same devices on two different occasions.
Usually you can cheat around this, especially during a multi-cam shoot, so it's not a huge problem.
The method of sync that roughly corresponds with the flash for visuals, is the clap. Obviously this requires recording audio on the camera too unless your clapper can be picked up visually by all cameras. I know it sounds obvious, but it is sometimes overlooked to actually record audio on the camera if you know you're also recording on an audio recorder. This is a terrible mistake to make. Always record audio to the camera. For that matter, if you're using a spare camera as an audio deck, point the camera at the scene as a backup wide shot.
Since hearing a clap without delay may be hard to do with widely spaced cameras and mics, sometimes you can only go by matching up the general sounds of the event as recorded on the camera versus mini-disc. Then confirm that it looks right in relation to the visuals if there is any great distance involved that causes delay.

I find it is easy to sync up sound track using this method:

1 I pan one track to the left, the other to the right.
2 Get them close by finding a readily identifiable sound/word/noise - whatever.
3 Kill the echo by sliding one track in relation to the other.


My Sony MD player stays in sync over long periods of time.

I've done this and a Sony consumer mini-disc seems to lose sync after 15 minutes. Time strecthing and pitch shifting fixed it well.

Sync to something with a big spike and a visual cue, like a drummer hitting a drum or objects hitting each other. I don't find syncing 2 waveforms to work that well (microphones respond at different speeds to sudden noises or something).

But you have to sync the sound tracks using the sound. I've yet to see a microphone that delays the sound once the sound hits the diaphram. I don't think microphones have much of a latency problem.

A common problem is that someone will tape an event from fairly far back with an on- or near-camera microphone. The sound arrives long after the visual event and will have to be slid up to make it look correct.

Same thing happens with, say, a wireless microphone which delivers sound to the camera almost as fast as the visual event, and another microphone back at the camera location. The microphone local to the camera will have to be slid in time to match up with the wireless sound (and the visual event).

With really bad consumer cam mics the waveforms will look different and I found it hard to match them up. Maybe it was because sound reaches the camera at different times and that I was using FCP (doesn't let you slide in increments less than a frame).

MD Time accuracy
 

Mics don't have latency.
The ONLY time latency exists is when there is a computer involved in the recording AND monitoring process.
Devices don't drift.
Devices with different clocks might not sync up, but as long as everything is at the same sample rate, it's easy to fix whether you are in Forge or Peak, or other audio tool. It's when you have a 44.1k sample in a DV project that is 48k that this becomes a problem.
FCP doesn't really allow for stretch, but Peak will make this simple to do.
Clocking is different from camera to camera too, but this is rarely a problem if all devices are recording a 48k/16 bit signal. (default for DV) some folks use the 4 channel tools found on various cams that giveyou 32k/12 bit signal, what a waste that is. Don't even bother if you want good audio...Use 2 channels of 48/16.
Sync is easy to maintain by having a clap before the project action begins. It provides a transient. You can also use a child's toy "clicker" to create a less noisy but more spiked transient, allowing audio to be sample-accurate-locked to picture.

Microphones differ in transient response (kind of like latency), which may be one reason why the waveforms don't look the same.

http://homerecording.about.com/libra...y/aa040897.htm <-- good picture there

MD recorder time accuracy?
 

No, not kind of like latency, latency is specifically related to signal lag between input and output. Transient exception is related to how a microphone shapes the sound, or the personality of the microphone, but there is zero time lag, it's simply how the microphone interprets the analog pressure coming in to the plate and how the plate translates that to electrical energy. But there is zero delay in how this occurs. The drawing you reference shows a different curve, which in the convoluted world of a drawn waveform, might appear to be 'delayed' where it's actually simply the difference between one frequency curve being drawn and another containing more information, resulting in a higher or lower drawing of the transient. The drawing also is a greater reference to how voltages are maintained in the mic's output. For instance, the ribbon mic by nature, cannot handle high SPL, Sound Pressure Levels, or the ribbon will break due to the method in which it moves. A condenser mic has positive voltage, or active voltage applied to the plate, and therefore can respond with a higher voltage output, as the figure A you refer to shows. Figure B is a standard dynamic mic which generates voltage based on the SPL of the analog signal coming in, therefore it's transients/voltage jumps, are less significant or powerful. None of these transient drawings relate to latency.
Transients are merely the difference between the most quiet point in a signal and the loudest point in the same signal adjacent to the most quiet point.
It is *possible* and even probable that an on camera mic, generally a dynamic mic, at a different distance from the sound source than another mic, would draw a waveform in an NLE differently than a mic closer to the source, or sound being mixed from the sound board. However, this can be offset in most any NLE. This is not latency, this is a study in Thiel's time paradigm. In the old days, we'd use a slide rule to calculate the differences to offset via varispeed gradually sped up with lots of preroll, or offset by quarterframes for a start point. In modern times, we find the attack point, which is invariable regardless of the mic used, and find where in the decay point it lines up with. If the source clocks are same sample rate then it will generally maintain sync for reasonably long periods of time. Of course, voltage and other issues can creep in here shifting the clock ever so slightly, and cause drift based on power pulses.
But to get back to point, microphones don't have latency in any explanation. Computers and processors do.

Hey Douglas,

Welcome. I hope you can stick around!

Glenn,
Use your ears, not your sight to line them up. Even with a one-frame resolution, you can get two tracks lined up. If they drift, they don't drift much and you can detect that very easily if the tracks are panned to opposite sides. It is a small matter to then separate one of the tracks and slide it.

I'll be sticking around, I was surprised to see how large this forum has grown!!
In lining up audio, the trick is to listen for a flam or phasing sound. If you hear that, then the tracks aren't lined up exactly, or they are slightly different clocks. If you can hear them being lined up on transients but not in between it might well be differences in the converter. If the sounds have a slight "chorus" effect to them that is consistent, or if the volume of the overall track seems to drop, then the audio is lined up correctly. If your NLE will allow you to zoom to the sample level, (many don't) then you can easily zoom in and find the attack point and line it all up there. Problem is in that event, if the cams were at different distances, you might have some unique artifacts. But they'll be so negligible it usually won't matter, because you'll be using the best of multiple audio sources.