Bypassing HDV audio

[Carlos E. Martinez]

Contrary to some opinions, I think HDV is not great except for the sound, which is MPEG. And now that I have bought a Z1 I intend to do something about it.

What I plan to do would be in two stages:

1) Recording on an external non-TC recorder, like a Zoom H4 or Roland R9. My idea is to pick the audio output from the audio/video jack on the Z1, which should be line-level, and feed it into the audio recorder, recording in PCM. The Z1 would also work as a preamp.

2) Using a portable mixer, like a Sound Devices Mixpre, to better control the microphones. I have always advocated for external mixers for audio quality improvement. Not only because the preamps are better, signal metering is more clear and good limiters are on-line, but also because you have better control of the audio levels, better linearity and better headphone monitoring.

Has anyone used any of these and got good results?

My idea is to get an external bracket to hold this external stuff. The one I have is the Bracket1, but it seems rather tight packed if you will hold a recorder and a mixer.

I have seen small hollow boxes that you can screw to the camera, on the base, to hold HDD and wireless mics. But I don't think they will take a Mixpre and a Zoom H4.

This bracket and setup should be balanced in some way to use it hand-held too.


Carlos

[Steve House]

If you're going to use a mixer I would suggest giving some thought to bypassing the camera audio step altogether and recording directly to your separate audio recorder. While the pros and cons of the MPEG audio format are debateable, another issue is the audio circuitry in the camera itself. If you can cut that out of the loop, so much the better. You'd be better served, I think, if the mixer fed the audio recorder and the camera (now used as a scratch track) either independently from different outputs from the mixer or feeding the camera from a monitor out on the recorder (the reverse of what you said). Also, I'd suggest looking at something like the Tascam HD-P2 instead of the handhelds you mentioned. A bit more expensive but probably worth it in the long run.

[Carlos E. Martinez]

Originally Posted by Steve House

If you're going to use a mixer I would suggest giving some thought to bypassing the camera audio step altogether and recording directly to your separate audio recorder. While the pros and cons of the MPEG audio format are debateable, another issue is the audio circuitry in the camera itself. If you can cut that out of the loop, so much the better. You'd be better served, I think, if the mixer fed the audio recorder and the camera (now used as a scratch track) either independently from different outputs from the mixer or feeding the camera from a monitor out on the recorder (the reverse of what you said). Also, I'd suggest looking at something like the Tascam HD-P2 instead of the handhelds you mentioned. A bit more expensive but probably worth it in the long run.

Thanks for you comments, Steve. I have quite a large experience as location sound recordist and sound designer, so I am quite aware of all the points you raised. I also rent audio equipment for a living, so I need to know what's on the market. That doesn't mean you don't always learn something new. ;)

The pros and cons of MPEG audio are not really debatable, if you are working for a theatrical feature. And the cons are much worst than the camera internal circuits, which BTW improved a lot after the PD170.

If I use a mixer, it will feed the external recorder and camera, the latter if possible not going through the recorder. Perhaps I misexplained myself: if I don't use a preamp, I would feed the recorder through the camera audio-out. This I can consider as an option, because as I said the camera internal electronics have improved a lot. But I will carry on HiMD PCM tests before going further this path. BTW I consider standard MD audio to be superior to MPEG audio.

The Tascam HD-P2 looks like a very interesting recorder, particularly because you can record TC. But it's too large for what I intend to do. You also have to carry and use it from your shoulder when you are moving. I want something you can put on the camera.


Carlos

[Carlos E. Martinez]

There was a mistake on mistake on my initial mail, which I now corrected. What I wanted to say was that "HDV is NOT great for sound". Now the rest of that mail will make sense.

Sorry.

[Martin Pauly]

Carlos,

not looking at the quality impact for a moment, I can't tell from your post what the advantage is to keep the camera involved at all in the audio processing. Do you intend to use the built-in (or attached) camera microphone and just record its signal with higher quality, instead of putting it on the MiniDV tape? I am sure you have thought about this carefully, I just can't figure out what the reason might be.

- Martin

[Carlos E. Martinez]

Martin,

Originally Posted by Martin Pauly

not looking at the quality impact for a moment,

We can't do that. The reason for this whole task is to improve on the audio quality.

I can't tell from your post what the advantage is to keep the camera involved at all in the audio processing. Do you intend to use the built-in (or attached) camera microphone and just record its signal with higher quality, instead of putting it on the MiniDV tape? I am sure you have thought about this carefully, I just can't figure out what the reason might be.

Once again, sorry. Yes, I have been thinking over this for a long time, and I might be leaving lots of things unexplained. Let's try to answer to your questions and explain the whole process:

1) I will certainly not use the built-in microphones. If I get to use the camera as audio preamp, I will input with condenser and/or wireless mics through the Z1's XLRs. The Z1 has internal limiters, reasonable quality preamps (even if I will carry on some comparison tests), phantom power and external level pots to do adjustments. The on-screen meter is not bad. This setup might allow me reasonable control (and quality) which I can output through the audio/video jack on the left and go into the portable recorder. As I see it, the main problem is in the HDV audio RECORDING, not in the audio chain. Compressed audio recording is what I am trying to avoid.

2) The internal preamp may not be up to my demands. In comes a preamp/mixer to help us. The SD Mixpre is small enough to almost make a single unit with the Zoom H4, as well as holding it all on with the Z1 on a camera bracket. It has good mic preamps, bright meters, is very small & light, limiter, AA power, good heaphone amp. Mics go into the Mixpre and go out into the camera and H4. Audio recording is done simultaneously in Zoom and Z1.

3) To be investigated: TC out in the LANC out. I should need an interface, but it might allow me using SMPTE TC sync on the audio track. There's a technique, used by Ambient slates, that allows you recording a TC burst at the beginning of each take, but you must rebuild the TC track in transfer. If not I would have to record the TC track on one audio track.

[Steve House]

Originally Posted by Carlos E. Martinez

....

3) To be investigated: TC out in the LANC out. I should need an interface, but it might allow me using SMPTE TC sync on the audio track. There's a technique, used by Ambient slates, that allows you recording a TC burst at the beginning of each take, but you must rebuild the TC track in transfer. If not I would have to record the TC track on one audio track.

Let us all know if you manage to track down a converter that will access the timecode coming from the LANC terminal! I know a number of people have been looking for such a critter for quite some time and they just don't seem to exist. Likewise there aren't any compact portable solutions yet to generate TC from video - somethng like the MOTU Timepiece will do the job but it's not battery operated or easily portable in the field. That was one reason I mentioned the Tascam earlier - it can sync both its TC and its audio sample clock to incoming video / blackburst.

I'm sure you know this, but just in case, timecode capable file based recorders such as the Tascam or SD recorders do not record continuous linear code. The BWF file format specifies that code is used for a timestamp in the file header of the code associated with the first sample of audio and the code that is output during playback is generated by counting the recorded samples starting from that data point.

Why do you say you'd need to record code on a separate audio track, a la DAT or analog recording? AFAIK, the 'regenerated code' method is just as accurate, if not more so, than traditional LTC tracks, providing sample level accuracy in the output code.

[Carlos E. Martinez]

Originally Posted by Steve House

Let us all know if you manage to track down a converter that will access the timecode coming from the LANC terminal! I know a number of people have been looking for such a critter for quite some time and they just don't seem to exist.

Well, it did exist. It was this one:

http://www.spcomms.com/lanc_interface/

But I am beginning to think it's not made anymore.

There's this one too:

http://www.avitresearch.co.uk/lancmtc.htm

But I am not sure how would that work.

My idea was to use a Denecke generator, that I already have, to jam to the camera. That's why I need a Lanc/SMPTE converter. I don't think the generator will recognize a MIDI signal, but I could look into that with the Denecke people.

AFAIK you have to jam the generator every time you change a DV tape, because it's not absolute time TC on the Z1. But I think you can control it. Not familiar with that yet.

Then I would inject that signal on a TC channel or on an audio channel.

I'm sure you know this, but just in case, timecode capable file based recorders such as the Tascam or SD recorders do not record continuous linear code. The BWF file format specifies that code is used for a timestamp in the file header of the code associated with the first sample of audio and the code that is output during playback is generated by counting the recorded samples starting from that data point.

Yes, but that's a specific file type, that is recognized along the chain as such. ProTools, or whatever editing program you download the BWF file to, will separate the TC and put it along the audio track. I am not sure how that would be here.

Why do you say you'd need to record code on a separate audio track, a la DAT or analog recording? AFAIK, the 'regenerated code' method is just as accurate, if not more so, than traditional LTC tracks, providing sample level accuracy in the output code.

Yes, that's what I meant. What I am not sure of is the accuracy of the Z1 to keep sync with longer than 30 minutes takes. The regenerated code, which is also how the Ambient "burst" system worked, is based on a certain accuracy. You also need a generator that will start with each burst and regenerate the signal. I never used it, just know how it works.

Some time ago I made some tests because I wanted to use an MD based system, and the MD not being a TC based system could use that burst system. Denecke told me the equipment I should use and all that.

An interesting note is that in pro forums told me that the TC wouldn't be recorded on the MD, and it did. I played a continuous 3 minute take on my TC slate without missing a frame.

Of course, like on the small Sony DATs, in MD you needed to record the TC on one audio track. Using the "burst system" allows you to have two audio tracks.

Only once did I read of an European film, if I am not wrong, that was made in DV using TC.


Carlos

[Jeff Mack]

Originally Posted by Martin Pauly

Carlos,

I can't tell from your post what the advantage is to keep the camera involved at all in the audio processing.

- Martin

Carlos, I agree, no need for the camera to be involved. The mix pre can take your mic in and then output directly to the recorder. My opinion however is to incorporate a laptop and a Firebox by Presonus. You probably have a laptop and your NLE probably has all the editing power you need. If not, the Firebox comes with Cubase. You could use the camera mic audio simply for reference and run your mic straight into the firebox and firewire into your editing software where all of your plugins are available to mix with.

I'll beat you to the punch and say you probably don't want to lug the laptop around or worry about power. If that's the case, sorry to waste your time.

Jeff

[Steve House]

The concern is not whether the Z1 can keep sync with takes longer than 30 minutes but whether the audio recorder can maintain sync with the camera. Remember in video audio chases picture while in film the reverse is true and audio is the master timecode reference. To maintain sync over the course of long shots, the audio recorder's sample rate must be slaved to the camera's video sync so that during 1 second of video exactly 48000 samples of audio for each channel will be captured. Recording timecode on an audio track doesn't accomplish that but continually receiving TC or video from the camera and converting it into wordclock to slave the audio recorder will. The derived code works off of that basis - if the first sample of audio is recorded at timecode '00:00:00:00' either fed from the camera or generated with the audio recorder's own TC generator, 48000 samples later the generated timecode will be '00:00:01:00' and if that point corresponds to exactly 29.97 frames of video due to the audio clock being slaved to the video clock, you'll be in perfect sync. The Ambient system of Lockit boxes is based on that premis - each Lockit device is tuned to the Clockit master so they are running at exactly the same rate. A Lockit set to produce video genlock is fed into the camera driving its sample clock. Meanwhile another Lockit, this one set to provide wordclock, is fed to the audio recorder driving its sample clock.

When you import a BWF file (WAV file) into an NLE the code actual isn't regenerated nor does it need to be. Instead, the NLE's own timeline becomes the clock reference. During import, the timestamp in the file header lines up the first audio sample with the same time on the NLE timeline. Since the timeline represents 1 sec of time as 48000 samples, if the audio clock in the recorder was accurate sync will be maintained indefinitely.

It's a shame that Sony in their wisdom did not choose to provide either genlock or TC outs on the Z1.

[A. J. deLange]

The Sony does put out both Genlock and TC in the sense that any composite video out can serve as a Genlock signal and that the LANC port conveys time code (at least I assume it does as LANC is a Sony invention and this is a Sony camera). This is the principal on which the AVIT converter works. It takes the LANC time code and converts it to Midi Time Code. If you have a MTC to SMPTE LTC (Linear Time Code) converter you can use the LANC derived time code in the same way as conventional SMPTE time code. I suspect, but have not verified, that the LANC TC signal is a unipolar (i.e. TTL) version of SMPTE time code so that running it through a blocking cap with load resistor would convert it to a SMPTE signal but I have not tried this and wouldn't advise it unless you have scope to check levels with. Also, I could be all wet in my basic assumption. There may be more to the conversion than I think (NRZ to Bi Phase mark or something of the sort). Whatever the LANC signal it is of audio bandwidth. Plug in a set of cheap phones (or use a blocking cap) and listen. Thus it can, therefore, be recorded on an audio channel and, presumably, reconstructed later.

And no, the TAO device is no longer made.

[Gints Klimanis]

Originally Posted by Carlos E. Martinez

Contrary to some opinions, I think HDV is not great except for the sound, which is MPEG. And now that I have bought a Z1 I intend to do something about it.
Carlos

I have nothing to help your quest for quality. I just wish that there would be a video recording standard that would allow you to adjust the audio quality to allow for uncompressed audio (48000 Hz, stereo, 16-bit) and/or allow for additional channels (like, four) from say the iLink connector to be added to the HDV stream.

[Steve House]

Originally Posted by A. J. deLange

The Sony does put out both Genlock and TC in the sense that any composite video out can serve as a Genlock signal and that the LANC port conveys time code (at least I assume it does as LANC is a Sony invention and this is a Sony camera). ...
And no, the TAO device is no longer made.

I should have said "doesn't provide genlock INPUT (so one can drive it with a Lockit or similar) or TC output..." I understand what you're saying about code being present in LANC - the problem is that there's no way to actually get TO it because as far as I've been able to find there're no devices currently on the market that can read it. It's a shame to have it available right there in that little jack and not be able to get to it to do anything with it - so near and yet so far <grin>. I've heard rumors that Ambient is working on a battery operated portable device that will accept video and perhaps LANC and output code and/or wordclock but it hasn't been announced yet. AFAIK, the Tascam is the only affordable recorder that actually slaves its sample clock to incoming timecode or video burst. The SD recorders will slave to external wordclock but they DON'T derive wordclock from incoming time code or otherwise sync their sample clock to the incoming code - this comes direct in response to a question I asked of SD tech support by email a few weeks ago.

[A. J. deLange]

Steve,

If you recall back to similar discussions we had a couple of years back regarding the XL2: There is a solution in the MOTU Midi Timepiece which will take in the MTC from the AVIT adapter and video from the camera and produce word clock and SMPTE LTC code. The HD-P2 can slave to either and time stamp (as has been noted) with the code. Another possibility for recording is MOTU's Traveler which captures to a laptop (can be powered by battery or the Firewire) and accepts word sync and LTC. The Timepiece is not a very attractive solution in that it is 19" x 1 RU and only runs on mains power (though as it draws but 7W powering with a motorcycle battery and inverter is feasible).

Somebody ought to do this as it should be a simple thing. I've dropped a couple of hints to a couple of our hardware designers but they seem more interested in spending their spare time with their families than on my pet projects.

[Carlos E. Martinez]

You are forgetting that most simpler editing programs are not quite made to work with audio time-code, and it would also take a special DV player to output an audio track as TC. Too many complications.

Ok. Let's forget a little about time-code. If that had been my original idea, I would have mentioned it at the thread start. It came by four mails later.

What do you think about my proposal and how would you make it work better, how to improve it?

The reason I am willing to implement it is because I will take on a job that we will be shooting with probably three cameras, two being Z1s. I will be handling one of the Z1 and supervise the audio setup. I won't be micing, as there will be a boom operator.

It's quite likely that I will follow this project through to the editing, so I want to organize something that works well.

At the same time I would like to think of something I will be able to use on my other jobs, where I will be directing.

For instance, one simple idea that worked fine in film times was the bloop light. You hit a button: a light flashes for a fraction of a second and short beep is recorded. That's something that three cameras can pick (the flash) and I can record on the CF recorder. Whether that beep is 1KHz or a TC tone is no problem, if you see where I am getting to.

OK, we will have to sync cameras and sound later on, but the audio quality will be preserved.

So that's it. If any of you can think of a way to improve on my general idea (Z1 + CF recorder + small pre package), plus the syncing question, please feel free to talk.