Sample Rates and Digital Interfaces

[Jeff Babb]

Hi folks,

I'm seriously thinking of making the move to computer based recording.  I've been reading Jeff's site, ramps, and the manufacturer's pages to try to figure it all out.  But I still have many questions before I take the plunge.  (I still have my Deva II for TC generation and over the shoulder stuff.)

I'm looking at a firewire interfaces and see, for example, the focusrite and the traveller have available sample rates of 44.1, 48, 88.2, and 96.  So what's gonna happen when post wants 48.048 etc?  Is there a setting that I'm not seeing in the on-line literature?

Jeff Babb

[RPSharman]

Jeff,

I just bought the FireFace 800 - This is what I recommend after the research, or the FF400 if that'll work for you.  It has a setting for 48.048.

Now here's the question for anyone reading this... Metacorder apparently has a setting for 48.048 but it does not stamp the files as such, which creates some confusion.  Perhaps Take can chime in about Boom Recorder and its settings for 48.048.

Many TV shows shot on film require 30ND 48.048k - Does Boom Recorder record and stamp files with this sampling rate?

Robert Sharman

[tourtelot]

FWIW, many shows for TV are recording at 48.048.  It really might be a concern.  I know that the MOTU units will not record at "off-standard" sample rates.

D.

[Darren]

Jeff,

Now here's the question for anyone reading this... Metacorder apparently has a setting for 48.048 but it does not stamp the files as such, which creates some confusion.Robert Sharman

Metacorder does have a "Pull Up Timestamp" feature.  However, I have no idea how it is used as I have no experience with 48.048 workflows.

[Jeff Babb]

I had a request this past week for 30 NDF, 48.048 and since I was using an analog mixer with my NL recorder it was no problem.  But it got me to thinking about the digital side of things.  It's good to know that the MOTU/Boom Recorder technique isn't hamstrung (sp?).

Thanks for getting back to me.

Jeff

[RPSharman]

Just to let you guys know.  Regarding 30ND 48.048 and Boom Recorder.  The configuration I found that works for me is the following...

I set my 744T for 30ND 40.048 - Fireface 800 is set for 48.048 - Boom Recorder is set for 30ND (receiving TC from 744T into audio channel) - Sample and File settings are both 48k - I tried setting File for 48.048, as it is an option, but then the TC does not match the 744T or the slate.  Not sure why.  Perhaps Take can shed some light on what the 48048 setting does which affects the TC.  But clicking on the TC readout in Boom Recorder, you can see the sample rate is 48048 when setting are both 48k, and I have had no complaints from post on a couple of different shows this season, so this seems to work.

Robert

[Sergio Sanmiguel]

It might be a good idea to start building a database as to what are the matching sample rate/fps for film, HD & SD with the 7 series SD Recorders & Boom Recorder...

I'm sure is not an easy task for software developers to find a quick solution for this issue, but to Us as end users to know how to handle a situation like this.

[takev]

Hello everyone,

The file sample rate is the sample rate that is set in the header of the audio file. The audio itself is not resampled. First the audio sample rate together with the framerate settings are used to calculate the number-of-samples-since-midnight. When writing the file the number-of-samples-since-midnight is recalculated in reference to the file sample rate.

If you lie to Boom Recorder about the audio sample rate (setting the audio interface to 48048 and Boom Recorder's audio sample rate 48000). You will also need to lie about the frame rate (30ND in the real world, 29.98ND in Boom Recorder).

As the timecode in the audio file is encoded as number-of-samples-since-midnight, the framerate written in the header of the file is unused by an edit program (except for displaying the original framerate).

I probably have now confused everyone even more. I am planning a new way of setting the sample rates, which I hope will make it much more clear what is happening.

[Richard Lightstone, CAS]

Robert,

Here is a previous post of mine on the same topic from July 4th:

http://www.jwsound.net/SMF/index.php?topic=1305.msg8458#msg8458

Regards,

Richard

[RPSharman]

Richard,

I remember this posting, but when I set the BR interface for file sample rate at 48.048, the TC readout was inaccurate and did not match the slate or the 744T.  As Take mentioned, the TC calculation is samples since midnight.  Perhaps with file sample set for 48.048 it reads input TC as a relationship of time since midnight and then calculates samples creating a new TC.  The TC is "later" than the input TC.  Perhaps I will turn on my machines now, later, then in the morning and see if the relationship changes with time from midnight.  Maybe I will also try setting 27.79 and 48.048 on BR with TC input 30 and FF800 at 48.048.

Robert

Robert

[Richard Lightstone, CAS]

Robert,

Yes once you set it to 48.048 and 30NDF I too saw a difference in the timecode readout on BR. I ignored this as I was doing a test.

As I said in my post, as part of my DVD-Ram ID, I audibly read out the Master T.C. from my Denecke GR-1 which I feed to both Boom Recorder and the Deva for about 15 seconds and record that as part of my Reel ID. (The Deva was also set to 48.048).

I then took that sample DVD-R from Boom Recorder and the DVD-Ram from the Deva to Modern Video where we played both disc back in the pulldown mode (29.97).

The Boom Recorder tracks were in dead sync and my audible time code count was a match to the Time code readout from the disc.

(The same for the DVD-Ram)

So to sum up - 48.048 and 30NDF in Boom Recorder resolved perfectly.

Regards,

Richard

[RPSharman]

Thanks, Richard.  This is good to know.  I was nervous seeing the TC not match, but glad to know it was okay.  It seems the 48048 in the file menu has no actual effect other than marking the file as 48048 and making the TC readout not match.  But it does add some confidence to the post guys to see the 48048 sample rate in the file info.

Robert

[Noah Timan]

Now here's the question for anyone reading this... Metacorder apparently has a setting for 48.048 but it does not stamp the files as such, which creates some confusion.  .

Stamping a 48.048k file as 48kHz has become pretty standard industry practice, in my experience.  Most Avids don't recognize a 48.048kHz stamp, so the recorder stamps the file at 48kHz to "fool" the Avid into importing the file.  This was originally called "F" mode on some machines (Deva, SD) because it was Fostex who originated the practice, but I've never heard of anyone successfully using a file in post with a "real" 48.048 stamp. 

nvt

[Richard Lightstone, CAS]

Noah,

In our discussion, the pulldown is achieved in telecine transfer. The DVD-Ram is played back at 29.97NDF and this achieves the pulldown to 48K. So the files sent to editoral are at 48K, 29.97NDF. Courtney Goodin will have a much better explanation of this - especially the designation of the "f" mode. I also believe 48.048 was not invented by Fostex.

Regards,

Richard

[Glen Trew]

Hello Richard (and all),

The days of timecode controlling the playback speed stopped with the Nagra IV-STC. Since the use of DAT, playback speed has been, and is still, controlled by the sampling frequency. For location film/video recorders, the Fostex PD-2 was the first to have a sampling freq specifically for the purpose of the .1% pull down, but missed the bet with 44.056kHz. The PD-4 finally got it right with the option of 48.048kHz.

The "-F mode", coined by Zaxcom, stands for "Fostex Mode", where a 48.048 file is stamped as 48K. The Fostex idea is to stamp the file with the predicted playback rate (typically 48K) to achieve automatic pull-down. Zaxcom's idea was to stamp the file with the original recorded frame rate and then choose to pull it down or not. Fostex's method was ingenious for the linear system they pioneered (DAT), and it is logical that the idea continue - though not neccessary - with their nonlinear machines (DV-40 and PD-6).

Both concepts have their strong points and logic, but most of the major players (specifically Fostex, Zaxcom, Sound Devices, Metacorder, to name a few) have implemented user options to accomodate each others systems. The more these can be understood by the users, the better, but it also underlines the increased need to have communication between the prouction and post production sound teams prior to rolling the first take.

Glen Trew

[Richard Lightstone, CAS]

Glen,

Thanks for the explanation -- but more importantly, how are you? I'm assuming you are home and out of the hospital?

Have a great Labor Day weekend.

Regards,

Richard

[Noah Timan]

Noah,

In our discussion, the pulldown is achieved in telecine transfer. The DVD-Ram is played back at 29.97NDF and this achieves the pulldown to 48K. So the files sent to editoral are at 48K, 29.97NDF. Courtney Goodin will have a much better explanation of this - especially the designation of the "f" mode. I also believe 48.048 was not invented by Fostex.

Hello Richard,

Yes, the pull *down* is achieved in telecine transfer.  However, in a 48.048 kHz workflow, the mixer is first pulling the sampling rate *up* by 1% (from 48.000 kHz to 48.048 kHz) so that when telecine pulls the material back *down* 1% the resultant actual sample rate is 48.000 kHz rather than 47.952 kHz.  This avoids having to go back and forth from an analog domain in transfer, and is the benefit of the 48.048 kHz workflow. 

What is affected as such during this transfer is only the media transferred by telecine on the telecine deliverable (ie generally only the mono mix or 2-mix portion of the file for printed/transferred takes).  This does not affect the original recorded BWF files. 

Now, after the telecine transfer, many editors will want to load the original BWF files directly into the Avid.  This allows them to use other material that may not be included in the telecine transfer -- for example, iso tracks or alternate takes that were not printed and transferred.  If the *timestamp* on the file says "48048" in the stamp instead of "48000" (regardless of what the sampling rate actually is) this can cause the Avid to reject the file.

What I was indicating was not necessarily that Fostex invented 48.048 kHz, but rather that their systems (AFAIK) were the ones that began the practice of *stamping* 48.048 kHz BWF files as 48.000 kHz files -- hence the use of the terminology "F" mode.

I don't know if it is still available, but a good resource for understanding the ins, outs, and process of 48.048 kHz vs. 48.000 kHz is Matthew Schneider's (of Postworks) presentation at last year's Gotham Sound seminar.  I know it was available for a time as a streaming video on the internet and may continue to be.

nvt

[Noah Timan]

The Fostex idea is to stamp the file with the predicted playback rate (typically 48K) to achieve automatic pull-down.

Hey Glen,

Correct me if I'm wrong, but I believe the Fostex method doesn't actually pull anything down -- it just alters the listed sample rate in the metadata, effectively "fooling" the Avid or other machine importing the file into believing the file is 48.000 kHz and treat it as such.  My understanding is that this method has proved successful thus far in editorial workflows.

nvt