Least worst drift in portable recorders?

[Noa Put]

Originally Posted by Richard Crowley

I'm with Mr. House. Buy a lottery ticket and you can buy full genlock/timecode equipment with the millions of €€€ you will win.

In another thread where something similar was discussed a few other weddingvideographers claimed the same as I do, which is normal because in our business we don't use very expensive external audio recorders but preferably small and cheaper ones. I could accept having 2 recorders that are so close as luck but not 3 (taken the tascam out of consideration), especially when other colleagues come to the same conclusion.

I believe you that "on paper" it is not possible but it appears that a larger percentage of these recorders can hold sync much better then you think.

Have you actually tested several recorders I use to be certain that I and other wedding videographers all have the possibility to win the lottery :)?

[Pete Bauer]

We shouldn't go too far afield from the original topic, but a couple quick replies to Richard's and Steve's comments.

Certainly, temperature extremes have some effect on crystal oscillation speed, but I'm doubtful that temperature differences between devices recording the same event explain the kind of differences we're talking about in this thread.

As far as the validity of comparing the oscillators in watches and clocks vs the oscilllators in low-cost AV gear, that's an open question...unless there is specific info available to share. It could be so, but I don't see why it would be, that inexpensive camcorders and portable audio recorders use very different crystals with orders of magnitude difference in frequencies.

The experience of many of us is that the device accuracy is in the same ballpark, and there would just be no incentive for manufacturers to use more expensive parts like a 10X or 100X more accurate oscillator. Whether a $200-300 audio recorder is sourced with the exact same crystal as clocks and watch manufacturers use, I don't know...and I doubt most of those manufacturers would care to state their parts list. I suspect the circuitry that counts and processes the crystal oscillations is the biggest variable at the low-cost end, and the oscillator itself is probably a commodity part. Anyway, if there is factual information about that, it would be interesting to discuss in a thread of its own.

For my part, I've done shoots when it was in the high 90'sF (upper 30'sC), humid, and sunny. And I've done similar shoots at night when it was jacket weather. I just don't detect any difference in relative accuracy amongst the cameras and audio recorders I use.

I entered this discussion simply to say that my experience of digital audio recorder accuracy has been about what Noa's has been. And we've both taken the bother to actually measure the relative accuracies of our devices. Personally, I'd be chucking any device in the recycle bin if it put my recordings off by any more than a couple frames an hour. But, I'm fussy about such things.

[Roger Shore]

It's interesting to read the kind of differences in audio 'drift' other folk are encountering. Pretty much in line, it would seem, from what you might expect from the differences in commercial grade crystals. (Only the very cheapest 'dictaphone' style audio recorders are likely to use ceramic resonators, rather than crystals, and they are likely to be much worse!).

One thing that hasn't been specifically mentioned is the fact that it is the accuracy of the camera oscillator that is the governing factor.
Say you have an audio recorder that is absolutely spot on. If it's the camera that is slightly adrift, then you have to adjust your 'perfectly timed' audio recording to fit the camera audio . That's the one in sync with the video -- so has to be the master.
Even if it's the track that's slightly wrong!

So it's not absolute accuracy that is the critical factor, but the relative accuracy of the different pieces of kit.........

Now whether cameras do generally use higher quality crystals or not?.....if not, then those that find their kit 'in sync' are indeed lucky.

It could be that the camera is adrift, and you're just lucky your recorder is adrift in the same direction -- and by the same amount!

[Dave Blackhurst]

Not to poke a hole in the boat (since we're a'drift), but you also must consider the physical distances between recording devices (most wedding venues for instance may have several hundred milliseconds of "delay" if your recorders are widely separated). If you move the recorders, that alone could account for "drift"! Under controlled conditions, you might be a little off, in "live" sound, you could be all over the place! if it's a "live" room with lots of reflected sound, even more fun!

I know that electronics now are very cheaply made, but I have trouble buying into huge inaccuracies due to electronic variables...

I typically sync multiple cameras and recorders, and have yet to see huge differences in "sync" - yes, there are likely slight ones, equivalent to short "reverb" type delays (which actually provides automatic "ambience" if mixed very carefully), but not "his lips are moving and the sound isn't the same" type stuff. I do a "fine tuning" sync off of physical motion though, rather than audio, as there can be some slight differences as noted above due to delay between audio source and recording device.

[Pete Bauer]

Dave, it is good to hear that your experience about drift between devices is similar to Noa's and mine.

A couple of frames variability is possible in a typical venue if you're on the move, but a couple second difference isn't unless you have a pretty massively large venue. So I don't really think that's at the root of the problem for the OP or Richard. No boats harmed in the writing of this thread!

At 30fps, audio will lag video one frame for every 37 - 38 feet of distance between source and the mic. Two seconds delay translates to about 750 yards, or 4/10 mile.

Shooting in our larger local stadiums, my high center camera's audio can lag by four frames from a mic on the field AND the video from that high center camera. That's expected since the sound has to travel an extra 150 feet past the audio recorder's mic and on to the high center camera. The light that gets recorded by the camera takes essentially no time, only nanoseconds, to travel the distance.

[Henry Kenyon]

The last wedding I did, (several years ago) was an outdoor event. The three cameras were at least 50 feet away. The groom wore an omni directional lav connected to a Sony Minidisc walkman in his pocket. Each camera had scratch audio.
In post all four audio sources were flawless after 35 minutes. All that was needed to do was adjust for delay, but the MD was rock solid with the video.
I'm considering a multi channel Tascam perhaps the DP24, with a number of wirelesses.
What I've heard many times over is that SD cards easily slow down as they fill up (usually after 50%). Could this be a cause for drift?

[Mark Morreau]

Thank you, everybody, for your contribution to this thread. It seems that real-world experience suggests that these recorders might be more accurate than the theory suggests.
Particular thanks to Noa whose real-world testing of his devices has influenced my buying decision.

[Noa Put]

Remember that my findings are not to be considered as standard deviation for such devices, from what I hear from other users (that are doing weddings) they all say their recording device is also pretty close, not always dead one but I guess like in my case several milliseconds off per hour. In my case that means moving it one frame per hour in general to maintain sync. Can we know what you decided to buy?

[Al Bergstein]

To the OP, I've shot a lot of video of live performances, over many hours in the last few years. I'll post a few examples at the end of this post. I've never had enough drift, if any, to worry about. I thought about jam syncing and get professional gear to really sync right, but have found I've never needed it. I use a Canon XF305, XF105 and sometimes a 7D or 5D Mkiii. I have used in the last two years, in order, a Zoom H4n (it died on me prematurely and was the least professional of the units I have used), the Tascam HDP2, a great recorder but somewhat dated, but if I was doing theatrical off camera Indie productions I likely would have kept it). My buddy who is a professional audio engineer up here claims the Tascam is the best cheaper unit he has used, and that it's pre-amps were the cleanest he has heard, and it does jam sync. The videos of the Brothers Four and Mark Pearson on my web site below were all recorded on the Tascam. I also borrowed a friends' non TC Edirol (the hard drive based one) and my current unit, a Marantz PMD 661. My earlier videos of the Brazilian Choro were all onto Edirol and were continuous have concerts, no stopping the camera, so like one hour of music. The concerts from the last six months are all PMD 661 recordings.

If I was still shooting tape, then it could be a problem of the tape transport, and that, yes, might cause you enough grief to look at jam syncing or actually running TC from a pro unit for doing so.

As I said, I've always synced up after the fact, and even stopped using Plural Eyes because it seemed easier to just find the sync point visually in my NLE's (I've gone through FCP, Vegas and now on Adobe Premiere/Audition).

I would say that if you are having drift problems, it's likely your recorder, and buying a new one, of the caliber of the ones I mention above, will likely solve your problem. As I mentioned, the Zoom H4N seemed to be the least capable of the ones I've owned/used. I think that out of them, if I bought one today, I would be likely to either get the Edirol R-44 (for four solid XLR inputs if you need them) or the PMD661, which I have liked the best of all of them. A great professional level tool. (used by radio field reporters around the world as I understand it). Certainly not a Sound Devices but we aren't talking about you spending that kind of money. If I was doing truly remote professional level work with a crew, I'd just save up and buy the SD. They are top quality.

To be clear, I am not a professional audio engineer. Just a videographer who cares about getting my sound as good as I can.

My videos are at
https://vimeo.com/mountainstone/videos

They aren't "perfect", some were shot with single camera, some multi and a mix of a lot of varied cameras that sometimes didn't match quality, but all the concerts you see often ran hours in length, and I would shoot continuous takes until the breaks between bands. I've shot nonstop for a half a show without experiencing any noticeable drift.

[Donald McPherson]

Could I through in another ball. Could the drift have anything to do with CBR or VBR? Shoot me down if you want. I know nothing.

[Richard Crowley]

Originally Posted by Donald McPherson

Could I through in another ball. Could the drift have anything to do with CBR or VBR? Shoot me down if you want. I know nothing.

No. There are at least three COMPLETELY DIFFERENT AND INDEPENDENT kinds of "compression".

1) TIME compression is where you take a recording and make it shorter. Modern digital techniques allow us to do this without changing the pitch. And conversely you can STRETCH a recording to make it longer.

2) AMPLITUDE (or LEVEL) COMPRESSION (aka LIMITING) is the process of reducing the amplitude (or level or loudness) of the audio signal so that it fits within the maximum value of the digital sample. Allowing the audio signal to go over FULL SCALE in a digtal recording will CLIP the audio. This is typically a rather bad thing to be avoided (unless you are doing fuzz e-guitar). Of course, there are corrolaries in video as well, where you can :"blow out" the whites by opening your lens too wide. In video-land "blow-out" means "clipping"

3) DATA COMPRESSION is where you take the resulting stream of data values (whether audio or video) and reduce how much storage space (and/or transmission time) they take up. There are both "lossless" and "lossy" methods of doing this data reduction. CBR and VBR are two generic descriptions of the results of this data compression. Note that data compression DOES NOT CHANGE the sample rate, so it doesn't change the length of the clip. Well-known examples of data compression include MP3 for audio and MPEG for video. Both MP3 and MPEG (and most other compression schemes) have both lossless and lossy variations, as well as CBR (constant bit rate) and VBR (variable bit rate).

Note specifically VBR (variable bit rate) does not imply that the sound (or video) speeds up or slows down depending on the bit rate. It just means that the amount of (compressed) data may be higher at times, and lower at other times (typically depending on the complexity of the signal/data). For example you can compress the snot out of silence or black without losing anything.

These all border on gross oversimplification. Each of these qualify to be the focus of a graduate-level course or a stack of textbooks.

[Mark Morreau]

Am just bumping this up to share the results of a couple of sync tests I've done with my recently purchased Roland R-26.

In two separate one hour sync tests, running the R-26 recording WAV at 48Khz alongside my Sony EX1, I could not see any drift at all. AFAICT the R-26 kept perfect sync with the camera.

I was surprised, but very pleased.

I'll be using the R-26 in real life situations on a number of shoots this week, and I'm curious to see if I continue to get such good sync.

[Steve House]

Originally Posted by Mark Morreau

Am just bumping this up to share the results of a couple of sync tests I've done with my recently purchased Roland R-26.

In two separate one hour sync tests, running the R-26 recording WAV at 48Khz alongside my Sony EX1, I could not see any drift at all. AFAICT the R-26 kept perfect sync with the camera.

I was surprised, but very pleased.

I'll be using the R-26 in real life situations on a number of shoots this week, and I'm curious to see if I continue to get such good sync.

How are you testing the sync?

[Mark Morreau]

Originally Posted by Steve House

How are you testing the sync?

Got my metronome out, set it to 60bpm

Pressed "Record" on EX1 and R26 at about same time.
Used clapperboard to mark audio and video start.
Let it all run for just over an hour.
Used clapperboard again to mark audio and video end.
Brought files from EX1 and R26 into Premiere and put both on timeline.
Zoomed into timeline at frame level.
Lined up starting clapperboard marks on both audio tracks.
Went to end of timeline.
Looked at end clapperboard marks.
At frame accuracy they were in the same place.

Then panned the EX1 audio track left and the R26 Audio track right and listened to playback to see if I could hear metronome clicks out of sync. Couldn't.

Then repeated the whole thing. Same result.

See any flaws in my methodology? Please say if you do.

I could compare them in Audition too, to get down to sample rather than frame accuracy, but that can wait: right now 1/25 second good enough for me.

[James Kuhn]

Mark...that is a pretty slick little audio test. Simple & elegant. Bravo, dude.

Cheers,

J.