D.A.T. vs Mini DV

[Alan James]

Is recording sound on mini DV just as good as recording sound on a D.A.T. tape? From what I have read D.A.T. can record 48khz and as far as I know so can mini DV (I might be wrong though). Given that the connectors and mixing boards are all the same is it worth recording sound to D.A.T. tapes if you are filming digital anyways or is it better to record to mini DV tape. I am only talking about quality here not convince or price. Is D.A.T. only for celluloid with no optical sound track?

[Bill Ravens]

IMHO DAT is no longer the audio medium of choice. Seems like most people are moving to CF card recorders. With some of the better choices being up to 192KHz, 32-bit at a lot less money than DAT, not to mention smaller and lighter, compact flash is the best way to go. HDV format compresses the audio and doesn't provide the same quality that uncompressed PCM (which is the standard for mini-DV) does at 44.1kHz. MiniDV works fine but won't give you the quality of CF.

[Chris Barcellos]

Check specs out here.

http://reviews.cnet.com/Sony_MZ_RH91...9.html?tag=sub

I have the 910 and it records at various sample rates, with PCM being highest level, and about the same level as CD quality as I understand it.

[John Miller]

Originally Posted by Alan James

is it better to record to mini DV tape

You are right. DV supports 48kHz/16-bit stereo PCM. It also supports 32kHz/16-bit stereo, 32kHz/12-bit dual stereo and 44.1kHz/16-bit stereo - all PCM.

Pro DV kit typically only supports the 48kHz/16-bit and 32kHz/12-bit.

Consumer kit *sometimes* only supports the 32kHz/12-bit.

Given the nightmare of trying to sync DAT and DV in post and the equivalence of the recording formats, using your DV for audio makes perfect sense.

This is true for DV and its brethren (DVCPro etc) - it is NOT true for HDV and similar MPEG2-related formats. HDV uses an MPEG audio compression format.

[Alan James]

Thanx for the info. What you have said gives me more reason to dislike HDV (which I already swear I will never use for my own projects). So lets say I want to record some sound effects for a movie separate from my camera (canon XL2) for mixing into the final soundtrack. Would you recommend a compact flash recorder and what should I be looking for in a compact flash recorder if that is your recommendation? In your opinion what model would be a good counter part to my XL2?

On a side note. What do the majority of the pros use on movies shot on 35mm?

Thanx

[Steve House]

Originally Posted by Alan James

Thanx for the info. What you have said gives me more reason to dislike HDV (which I already swear I will never use for my own projects). So lets say I want to record some sound effects for a movie separate from my camera (canon XL2) for mixing into the final soundtrack. Would you recommend a compact flash recorder and what should I be looking for in a compact flash recorder if that is your recommendation? In your opinion what model would be a good counter part to my XL2?

On a side note. What do the majority of the pros use on movies shot on 35mm?

Thanx

One of the common recorders you'll on the sound cart on feature sets these days is a Zaxcom Deva V. Very pricey, 8-inputs, 10-tracks, records PCM to an internal hard drive, often mirroring to an external DVD-RAM drive for dailys. Sound Devices recorders are used for a lot of film features as well as double system sound video shoots, ENG and doco shoots, etc as well - the 744 is a 4-track model, the 722 is a stereo recorder. Both record to internal hard drive and CF cards. Then there's the 702T and 702, CF cards only but with the latest firmware upgrades they have the ability to simultaneously write to an external hard drive as well. Some mixers are recording direct to a Mac laptop using software such as Metacorder. And we shouldn't forget the Fostex PD-6, the file-based successor to the PD-2 and PD-4 DAT machines.

[Ty Ford]

Steve's right.

DAT is dead. HD SMPTE based recording with a SD 702T (for timecode) is a minimum. I have Boom Recorder on my laptop but haven't really used it extensively in a critical situation. Others have and seem pleased.

I have an XL2, btw and have found it works pretty well for audio.

Regards,

Ty Ford

[David W. Jones]

Originally Posted by Ty Ford

I have Boom Recorder on my laptop but haven't really used it extensively in a critical situation. Others have and seem pleased.
Ty Ford

Like Ty, I also use Boom Recorder.
In my case it is installed on a Macbook Pro, teamed up with a MOTU Traveler audio interface. It's not the best solution for mobile run and gun style work, but for low budget indie style productions it offers great performance per Dollar, allowing for multiple inputs, timecode, multiple outputs to send to the boom-op, director and so forth, as well as embeded metadata.
And the files work great with FCP.

[Bill Ravens]

hi end is Sound Devices CF recorders like the 702, 722, 744
lo end is the zoom h4
mid range is something like the tascam unit

no matter what you choose, getting time code or word clock sync into an offline audio recorder is not really possible with prosumer cams..they have no TC or wordclock out. clapboard is the commonly used technique and it works fairly well.

I think I saw something from JVC that is coming out that will read composite video out black pulse and convert it to a TC to sync to an external recorder, but, I don't remember wwhere I saw that.

[Ty Ford]

There's a box that pulls timecode out of LANC signal, but I don't know of one that's portable.


Regards,

Ty Ford

[Steve House]

Originally Posted by Bill Ravens

hi end is Sound Devices CF recorders like the 702, 722, 744
lo end is the zoom h4
mid range is something like the tascam unit

no matter what you choose, getting time code or word clock sync into an offline audio recorder is not really possible with prosumer cams..they have no TC or wordclock out. clapboard is the commonly used technique and it works fairly well.

I think I saw something from JVC that is coming out that will read composite video out black pulse and convert it to a TC to sync to an external recorder, but, I don't remember wwhere I saw that.

Part of the appeal of the Tascam HD-P2 is that it is able to sync its sample clock to incoming video blackburst. If the camera has a composite video out that sends a signal in real-time for monitoring while recording, routing that to the Tascam's video input will do the trick as according to the Tascam manual it can extract blackburst from ordinary composite video. But note, this does NOT send timecode from the camera to the recorder. The timecode clock and the sample rate clock are actually two different clocks.

[Bill Ravens]

steve...

part of what I perceive as the problem is that there are about 4 different ways of syncing equipment:1-wordclock, 2-midi timecode, 3-SMPTE timecode, 4 black burst.

Out of all these standards, only timecode is an absolute time sync. All the others sync to a pulse without a reference starting point. Better than nothing, but, still not really adequate for video/audio sync. SMPTE, of course, is the best, but, few audio recorders read SMPTE. The Sound Devices recorders use wordclock or timecode. But my JVC HD100 doesn't output either wordclock or TC.

[Steve House]

Originally Posted by Bill Ravens

steve...

part of what I perceive as the problem is that there are about 4 different ways of syncing equipment:1-wordclock, 2-midi timecode, 3-SMPTE timecode, 4 black burst.

Out of all these standards, only timecode is an absolute time sync. All the others sync to a pulse without a reference starting point. Better than nothing, but, still not really adequate for video/audio sync. SMPTE, of course, is the best, but, few audio recorders read SMPTE. The Sound Devices recorders use wordclock or timecode. But my JVC HD100 doesn't output either wordclock or TC.

It's really confusing, I'll grant you. There's two kinds of problems to resolve for sync -one is to be able to line up the audio and video when they're recorded separately and the other is to prevent drift so that when the first frame of the audio and video recordings are lined up in sync and then played through, they will STAY in sync over the duration of the clip. Time code makes aligning the two easy by matching the frame numbers but in and of itself it doesn't guarantee they will stay in sync. NTSC DV calls for 29.97 frames per second, audio at 48kHz. That means that 100 seconds of video will have 2997 frames and exactly 4,800,000 samples of audio. But suppose your audio recorder's sample clock is a tad off - let's say 48.01kHz. That 100 seconds of audio will have 4801000 samples, 1000 extra. Now we combine the audio and video and play them back together on the same machine - after 100 seconds where 2997 frames of video have played we'll also have played 4,800,000 samples of audio - but that isn't the full 100 seconds of audio that should have played, we've got a thousand samples left over. The audio is running slower than the video and during the 2997th frame of video is played we're hearing the audio that was originally recorded during the 2994th frame of video - we're 3 frames off. The picture and audio have drifted out of sync over the course of the shot. So what we need to do is lock the sample clock of the audio recorder to the sample clock in the video camera so that it really does record exactly 48000 samples for every 100 seconds of video. That's what blackburst, midi timecode, and wordclock sync do - provide various means where the sample clock in the audio recorder can slave to the clock imbedded in the incoming signal coming from the video camera. Those methods don't help you line them up because there's no common frame reference and you still need some sort of slate to do the lineup. What they will do is once you've somehyow lined up the start of both audio and video, they'll stay in sync through the scene (which is the much more important issue anyway). Timecode does both, provides matching frame number references to aid in aligning the audio and video in post and also a sync signal to lock the sample clocks together.

There are devices like the MOTU Timeclock that can derive wordclock from composite video to allow your camera to drive sync to an SD recorder even though the camera doesn't output TC or wordclock and the SD doesn't accept composute video but they're pretty pricey and not very portable.

Something I'm hazy about and am looking for some answers to, is whether the fact that an audio recorder like the SD units accepts SMPTE timecode ALWAYS means that they also slave their sample clock to it. Since the TC frame references are stored in the Time of Day clock and the TOD clock and the audio sample clock really are two different circuits, simply slaving the TOD clock from the frame numbers in the incoming timecode doesn't in and of itself mean the sample clocks are locked. I've assumed that recorders that accept timecode do slave their audio sample clocks to it as well but I haven't actually found it in writing in any documentation that that lock is engraved in granite (except the Tascam HD-P2 manual says that recorder does).

I find it really interesting that the mini-DV specs allow for 'unlocked' audio. That means that in most miniDV cameras the audio sample clock is NOT locked to the video clock, even when recording in-camera. We're lucky that most present day clocks are pretty accurate. But that's why longer clips from some cameras can gradually drift out of sync over the course of the shot even though they were recorded single system. That's an advantage to the Sony DVCAM spec, it calls for 'locked' audio, and is also a plus for the Canopus line of A/D capture cards - they also output locked audio.

[Bill Ravens]

exactly why I opted for an SD 702 instead of an SD702T. The SD702T has a timecode generator, but, you can't sync it to the camera TCG. So both TCG's go off on their merry own and you're right back to the clapboard to sync the first frame. And there's no guarantee that the TCG's are not going to drift apart, either.

Since most all DV and HDV cameras output composite/component video signals, I'd be REALLY happy if someone made a PORTABLE black burst to wordclock interface. seems like the lack of quality in all these HDV cams has to drive people to external audio recorders, so there's gotta be a market for it.

EDIT: wait a minute! I just found out that Midi time code is the same as SMPTE. Which means that a DAW like ProTools should be able to read the SMPTE TC embedded in a video file and sync it up, both reference start time and linear time, and lock the audio to the video. hmmm...need to go look into this. This means I wouldn't have to care if the audio recorder and video recorder were in sync. I can let the software sync the TC's in post.

[Steve House]

Originally Posted by Bill Ravens

..
EDIT: wait a minute! I just found out that Midi time code is the same as SMPTE. Which means that a DAW like ProTools should be able to read the SMPTE TC embedded in a video file and sync it up, both reference start time and linear time, and lock the audio to the video. hmmm...need to go look into this. This means I wouldn't have to care if the audio recorder and video recorder were in sync. I can let the software sync the TC's in post.

Are you sure SMPTE and MIDI are the same? I think I read somewhere not too long ago, I think it was on the SMPTE website, that they weren't. The time information - hr:min:sec:frames - being encoded is the same but the manner it which it's encoded and transmitted is different.

Also, there's no continuous timecode associated with a WAV or BWF audio file. The BWF has a timecode based timestamp marking the first sample but that's it. So we're back to where we started, even with software that lines up the matching frames of audio and video, that only applies to the first frame and there's nothing to prevent them from drifting apart later in the file(s). For it to maintain sync it would have to have timecode extending throughout the audio file and stretch or compress the audio so its embedded frame numbers always matched up to the video numbers. Even the SD 702T and 744T don't record continuous timecode - the clock runs continuously but the only value actually recorded in each file is the start of file timestamp. During playback the timecode display is generated 'on-the-fly' by adding elapsed clock time to the start of file timestamp rather than reading code recorded alongside the audio.

Analog recorders and DATs actually recorded the SMPTE code continuously throughout the take parallel to the audio proper but file-based recorders don't. With film, when the recording was "resolved" in post and transferred to perfed magnetic media to be mated with picture for editing, the recorded TC signal controlled the speed of the resolving recorders so one frame of recorded timecode corresponded to one frame of mag perf and thus one frame of picture.