Monitoring Media

[Joshua Anderson]

I had an issue with 4 CF cards the other day and it got me thinking about how we monitor the files. 

 

I use 2 788T recorders. I feed a split analog audio feed to them. If one recorder has freeze up or an other issue, the 2nd recorder is getting it's audio independently. Each sound roll then is 2 CF cards. On this particular day I had issues with 2 Sound Rolls - so 4 CF cards. I also record onto external drives and was able to get those to post to clean up the dailies. 

 

What struck me though as I've been trying to find the weak link is that I don't (and can't) monitor the media that I actually hand in. My recorders (and maybe others are different) only let me monitor the Internal Drive in record. I can play back and check files from the CF slot and the FireWire port, but I can't listen to them in record. 

 

With 1/4inch and DAT, we would listen to the playback head. Since I hand in CF cards that go to Dailies and then Post, I'm not actually listening "in confidence" like we could in the past. 

 

Just a thought I had about what we're actually monitoring in the digital domain. 

 

Josh

[Glen Deakin]

Don't really see how you can. With a 788 you can record to 3 separate media. Can't exactly monitor them all the time. Was there an error message on the 788? Anything go offline?

[Joshua Anderson]

Out of the 3 media I record (Int, Ext, CF), I can only monitor the Int during record, but I hand in the CF media for Dailies and Post. So why not monitor the CF while in record. But since there isn't a playback head on a digital recorder, does that matter?

In my case, there wasn't an error on my machine. At some point my CF cards corrupted, but I don't know if it was during the recording or after I handed them to the loader to put into the shuttle drive that gets sent to the transfer house.

[pkautzsch]

CF data rates not allowing for simultaneous reliable recording and confidencence listening?

As long as there is a backup created before you hand in the CFs, it shouldn't be a real issue. And that's what the machine does right while shooting.

[Joshua Anderson]

Unless I'm missing something on my 788Ts, after I cycle to the CF drive, it automatically switches back to the Int drive once I hit record. I was wondering if other machines were also so internal drive centric. My point was that in the workflows I typically work in, no one goes back to the Internal drive. I will send Post a drive of about 4 episodes on an external drive, but by that point dailies, editors and Post audio will have been working with the files from the CF cards.

 

My question was two-fold: If the CF cards are that central to the workflow, why not monitor them in record? But at the same time, we don't listen to anything like a confidence head anymore in the digital world, so does it matter? 

 

In my situation last week, I found that I was handing in CF cards that were corrupted. My Internal and External drives were fine and there wasn't any indication that there had been an error on recording. Maybe the cards went bad when our loader transferred the files on her laptop to the shuttle drive for the Transfer House. But I can't say with certainty that I handed her uncorrupted CF cards - because I listen to the internal drive. 

 

The sound was all good in the end, so there wasn't an "issue." Or is there an issue, because I then have to rush backup drives to Post so as not to delay dailies delivery to Executives by anymore time than this issue has already caused. 

 

Josh

[Simon Hayter]

Because you are creating a WAV file , I don't believe you can 'listen' to the file on the card until it is completed. I imagine all digital systems work the same way. All you can ever monitor will be the input. 99.9% of the time all this works fine for us. Just occasionally there will be a corrupt file. I just apologize and make post another copy.

[Joshua Anderson]

Because you are creating a WAV file , I don't believe you can 'listen' to the file on the card until it is completed. I imagine all digital systems work the same way. All you can ever monitor will be the input. 99.9% of the time all this works fine for us. Just occasionally there will be a corrupt file. I just apologize and make post another copy.

 

I was wondering if it was a "digital" thing. Like you're saying, I've only come across it one other time when I had a card (or recording) lock up. I run 2 recorders and I split an analog feed to them, so if one machine has a lock up it won't affect the other (rather than sending out of one recorder into the other). But this issue was spread across both CF cards - twice, as it happened again on the second roll of the day. Part of me thinks it's related to the download after the cards have left my hands, but in this digital world I have no practical way of confirming the integrity of the CF card. And like I was saying, it's become the source of audio for the entirety of the shows I work on.

[pkautzsch]

This topic bothered me more, so I started to re-read the manual.

I remembered that on the 702 and 744 there are the "M" headphone modes, which route a real playback signal from disk to the cans (aka confidence monitoring), introducing a delay of up to 12 seconds. The display will show something like "A,Bm". I also remembered that the manuals of these machines state that neither "input" nor "track" monitoring actually monitor the disk. "Input" monitoring routes only the selected input(s) to the cans (display showing "1,2", for example), while when "track" monitoring ("A,B") you'll hear all that is routed to the same track (which can be more than one input).

 

The "M" mode is not available on the 788T - so actually the machine doesn't have any sort of real confidence monitoring.

 

What struck me though as I've been trying to find the weak link is that I don't (and can't) monitor the media that I actually hand in. My recorders (and maybe others are different) only let me monitor the Internal Drive in record. (...)

 

Just a thought I had about what we're actually monitoring in the digital domain. 

 

Additionally I thought about how I'd have handled the situation.

Of course, the quick fix is to give them the data from the INHDD. Being able to do so in an un-bureaucratic manner is part of what gets us hired again.

 

7 series machines tell you if there is an error on any media, not just the highlighted one.

From 788T manual, p 41: "In the event of a storage media failure, the specific error will be indicated on the LCD and warning
bells in the headphone monitor. (...) If more than one drive is selected for recording, the remaining drives will continue recording without failure." I did have that with a badly maintained machine that belonged to the producer more than once.

 

So if you didn't get an error message while rolling, it didn't happen while rolling. It has probably happened during copying, or by removing the CF with the recorder powered on. I do power down the 788T when inserting or removing a CF - don't know of another way to unmount the drive. And I've learned about the importance of unmounting drives the hard way.

 

So, to avoid this happening again, I think the CF removal/handling/copying process on your shoot should be looked at. I'd have an informal talk to the data wrangler if they have any suggestions.

[Joshua Anderson]

I didn't have any warnings on my recorders.

 

But, I've never shut down my recorders before pulling CF cards out. I've had these machines since 2009-2010 and have been pulling cards as part of my reload process since 2011. 

 

So far it all seems like an aberration. But it made me think about how I can be more certain of the media format - if it's possible. 

 

Thanks for the history lesson on the 744. I skipped that machine, but always admired it. 

 

Josh

[JonG]

The 6 series machines (664 and 633) allow you to choose which card you monitor from (SD or CF). However the 788T is an amazing machine and I wouldn't replace it because of this one feature, so in your case I would verify that the CF cards you are using are reliable. I have CF cards that are approved media for my 788T, but I get missing files when I use them, however they work fine on my 633. So I keep those for the 633, and use previously approved cards that Ive had for years that have never given me a problem with my 788T. I use Transcend 133x 16BG CF cards for the 788T, again, never had a problem with them, and Ive had others tell me that they are more reliable than a lot of other recommended cards. However I am sure others here have had different experiences.

 

My two cents

[Joshua Anderson]

Jon, it's funny that you mention the transcend cards. I've been using 8 GB 133x Transcend cards for 3-4 years. No problems ever. Before starting this job, I thought about getting 16 GB cards so I wasn't against the wall sometimes with disc space. Production picked up 4 Delkin 16 GB cards that were on the approved list for 788Ts. I used them the first day, but they didn't sit well with my 788s: sometimes the 788s would look like they weren't seeing the external drive after I put in the CF cards. I would power cycle the recorders and it would be fine. But on the second day, I decided to go back to my Transcends and put off figuring out 16 GB cards. The problem cards ended up being 4 of my Transcend cards, on day 5. 

 

I still ended up using other 8 GB Transcend cards until Production switched my Delkins for 16 GB Sandisk cards. So far so good. 

 

I don't have any intention of switching out my 788s. I looked at the newer 970, but I like having 2 independent recorders, plus the 970 weighs a lot more and doesn't have any limiters. 

 

Josh

[osa]

The 6 series machines (664 and 633) allow you to choose which card you monitor from (SD or CF).

dont know how much this matters but i believe the term is "playback" when it comes to 6 series machines in that you specify what card any time a file is played back from the machine vs. monitoring in real time in the sense of confidence.

 

-Ken

[Jeff Wexler]

The issue of "confidence monitoring" was hotly debated at the time a small group of us started making file-based recordings in production --- for me this was 17 years ago with the first Zaxcom Deva I. There was great concern, mostly from all the people who were not yet using the Deva, that they couldn't "listen off the tape" (of course it wasn't tape but you know what I mean) while they were recording. For many, this was a deal breaker, and I heard from several people that they would not be doing this whole new file-based thing until they could have confidence monitoring. My take on the issue was this: for file based recording to be feasible, we were still going to need confidence, confidence in the integrity, stability and predictability of our tracks, but it became clear to me that we needed to re-define confidence. It was also clear that it was highly unlikely, even as other companies started to produce file-based recorders, that we would ever have a workable "listen off tape (disk)" audible confidence monitoring. One of the recorders (I think someone was saying it was one of the Sound Devices recorders), could give you a playback of your "recorded" file as you were recording, this was probably as close to audible confidence monitoring as we would ever have. Nobody used it as far as I know because this playback was something like 12 to 15 seconds after the real time recording, thus making it a feature that couldn't really be monitored while you needed to monitor the direct signal while mixing the recording in real time. Do any of the machines today provide monitoring of the recording off the media with little or no latency?

 

I found the solution, for me, was to re-define the notion of CONFIDENCE. After doing 3 movies with my Deva, I had all the confidence I needed and that confidence did not come from "listening" off the disk --- it came from the Deva operating system, checksum verification, the MARF file format, the use of highly reliable hard disks. My real world experiences provided me with the confidence that what I was monitoring during recording (the direct signal) would be exactly what the recording would be on playback. 

 

After 17 years of recording on the Deva I, Deva II, Deva IV and presently the Deva 5.8, the recorder has never let me down. This is not to say that the Deva or any of the other recorders have always been totally reliable for everyone, I have been quite lucky in this regard, and it also is not to say the whole file-based recording system is completely reliable. There have been, of course, many problems with various media (thankfully usually with our daily deliverable, not the primary master recording) and this is always troubling and often not so easy to sort out. To have full confidence in the recording you need to have confidence in the machine, confidence in its operating system, the file format, the choice of physical media, and the confidence that comes over time with all these elements.

 

I don't think we will ever have an exact equivalent of monitoring the recording as we used to "listening off tape" but I think it is clear that we don't need it either. Additionally, I'm sure quite a few remember that listening off tape, which we could do with DAT, was no guarantee and provided no sure confidence in the recording. In fact, there was one firmware version for one of the first DAT machines from Fostex that had a serious bug: if you switched to monitoring off tape it caused data corruption and what you heard "off tape" was garbage. Did not in any way instill confidence in the machine.

[Derek H]

Great post Jeff.

I think you're right the 744 can monitor off the disk but the delay is quite long as you said.

Not sure about other sound devices recorders.

[Philip Perkins]

The only SD recorder that can moni off the drive in real time is the 744.  With the others you are hearing some form of e to e.  I asked SD about why the "M" modes were removed from the 788, and they felt that the huge delay (many seconds) involved made it not worth the expense to add, since few people would use it.  I still drive 744s very often, and sometimes drop into an M monitor mode mid-take to see if a recording is being laid down, but that's only something that can be done if the scene being recorded is fairly placid.  You are hearing so far behind real time that it is pretty unusable for anything doco or dramatic.

[Joshua Anderson]

Maybe it's a stupid question, Jeff, but do you hand in drives at the end of a film that Post uses for all sound work?

I'll hand in a drive full of a few episodes to Post a couple of times during a season, but I think they'd use the audio that came from the deliverables without asking for anything else. So the deliverable can become the master.

[Jeff Wexler]

Maybe it's a stupid question, Jeff, but do you hand in drives at the end of a film that Post uses for all sound work?

I'll hand in a drive full of a few episodes to Post a couple of times during a season, but I think they'd use the audio that came from the deliverables without asking for anything else. So the deliverable can become the master.

If everything is working properly and as it should, the files that were on my deliverable media are identical to the master in all respects, so of course you could refer to those as being the master (along with the very same identical files that are still on the "master" drive/media). This is the fundamental beauty of digital files --- you can have an endless number of master files. At the end of a job I usually give sound editorial one drive with the whole movie on it but it is really just for further redundancy and backup to the files they already have.

[Tom Duffy]

We (TASCAM) put true confidence monitoring in our 4 channel rack-mount CF recorder (HS-4000) as it was a requirement for a big customer.   It reads back about 2 seconds delayed.     To get this to work we had to custom optimize the file system low level code, qualify the cards to a higher standard than for regular recording, and we ran out of processing power to do it at greater than 1x rates.

 

As Jeff explains, modern work flow deals with the issue differently now.

 

Tom (TASCAM).

[Constantin]

As long as there is a backup created before you hand in the CFs, it shouldn't be a real issue. And that's what the machine does right while shooting.

Well, it shouldn't be a real issue, but it is. The one time I had something similar happen (no, the 788T does not always warn of corrupt media), I got about 20 calls, from the transfer house, the assistant editor, the line producer, the production company, AD... All frantic about a missing sound file (one). I could calm them all down withthe fact that I had at 2 more copies, but still...

Additionally I thought about how I'd have handled the situation.

Of course, the quick fix is to give them the data from the INHDD. Being able to do so in an un-bureaucratic manner is part of what gets us hired again.

This unbureaucratic manner is sometimes not so easy. I keep my gear either in my van or on set and that applies to the backups, too, except for an additional backup I make every few days on an external drive.

So when the call comes I'm usually nowhere near the data they request, and they always need it "now!"

Fortunately it was only the one time for me, but it'll be the same next time, I'm sure

dont know how much this matters but i believe the term is "playback" when it comes to 6 series machines in that you specify what card any time a file is played back from the machine vs. monitoring in real time in the sense of confidence.

Same on the 788T, by the way.

I have started to make it a point to listen back to at least a few seconds of every other take or so, depending on time during takes. It's very fast to do on the 788 and it does actually provide confidence that a recording has been made that can be played back on the deliverable.

In addition, and this is a no-brainer, only use cards approved of by SD.

[pkautzsch]

Well, it shouldn't be a real issue, but it is. The one time I had something similar happen (no, the 788T does not always warn of corrupt media), I got about 20 calls, from the transfer house, the assistant editor, the line producer, the production company, AD... All frantic about a missing sound file (one). I could calm them all down withthe fact that I had at 2 more copies, but still...

Bad communication, I'd say - in the actual situation as well as beforehand.

In pre-production, an email to about all you mentioned, and additionally everybody involved in data copying, that you will have a backup on your recorder, and you will make another backup "at home". Won't keep everybody from panicking, but they do say "sorry" after being told they could have known instead of waking me up at 3 in the morning. Production should know who has copies of what anyway - they might have a top-secret policy where you might not be allowed to store stuff elsewhere.

In the actual situation, a quick email to everybody who needs to be informed helps if files are missing, and that they are recoverable.

 

This unbureaucratic manner is sometimes not so easy. I keep my gear either in my van or on set and that applies to the backups, too, except for an additional backup I make every few days on an external drive.

So when the call comes I'm usually nowhere near the data they request, and they always need it "now!"

Fortunately it was only the one time for me, but it'll be the same next time, I'm sure.

When a backup is in the same car as the original, it's not a backup.

I'd never leave the recorder in the van overnight.

And, frankly, it's not a big thing to hook the 788T up to a laptop via USB, and let it copy the files while having dinner.

If it needs to stay on set, I'll take a copy right at wrap time, and take the HDD with me. This is, actually, the safest way, and it might be just those 10 minutes that get you another hour of overtime. I really should start doing this as SOP.

 

I have started to make it a point to listen back to at least a few seconds of every other take or so, depending on time during takes. It's very fast to do on the 788 and it does actually provide confidence that a recording has been made that can be played back on the deliverable.

In addition, and this is a no-brainer, only use cards approved of by SD.

As long as you have the time to do so.

In a really fast-paced show, you don't.

[Constantin]

Bad communication, I'd say - in the actual situation as well as beforehand.

In pre-production, an email to about all you mentioned, and additionally everybody involved in data copying...

Those emails were sent and the multiple copies were mentioned in pre-pro discussions, too. That didn't keep anyone from panicking - at first. Like I said, everyone calmed down quickly. 

 

 

When a backup is in the same car as the original, it's not a backup.

I'd never leave the recorder in the van overnight.

A backup next to the "original" is still a backup with regards to corrupt media problems. And I tend to wait until the data wrangler has copied and verified my data, so... there is another backup. 

Where do you leave your recorder then? My (locked) van has an alarm and is parked in my garage, which has a special, heavy-duty protected gate and a fire alarm. Don't think my actual house is any safer. Insurance has certainly not disagreed at an on-site inspection.

 

 

As long as you have the time to do so.

In a really fast-paced show, you don't.

 

There is always time at some point. All it takes are a couple of seconds, no need to listen to the entire take. 

[Marc Wielage]

As Jeff often notes, the camera department never runs a backup copy while the shoot is happening -- it just goes to one drive, and if that drive fails or gets erratic, often many takes are affected. It's a miracle that many sound recorders today can make 2 or 3 simultaneous copies with relative ease. 

 

I think the CF cards just get flakey over time and you have to replace them every so often. At least now they're cheap -- I think the 32GB cards are around $30 or so, which is pretty affordable even compared to (say) 7 DVD-RAM disks back in the day. To me, this is a computer problem, not an audio problem.