Standardized Audio Workflows with BWF, Metadata...

[Eric L]

Standardized Audio Workflows with BWF, Metadata and Non-Linear Editing Systems

It is becoming evident that there is a large amount of confusion between the various film and TV production departments when dealing with the new species of hard disc recorders and the Broadcast Wave File format.

I have attended the two most recent seminars in which this subject has been discussed, and it has become extremely clear to me, from comments and questions made by the audience members, that a tempest is brewing on the all-too-near horizon.

Location recorder manufacturers have implemented functionality in their respected devices that, not only differ from manufacturer to manufacturer, but also are not yet fully supported by the various post-production software manufacturers.

As is often the case, various different workflows have been developed, in order to achieve results that are required on a specific project. For instance, Mr. Larry Blake spoke at the last seminar held on Dec 7, 2005, in Hollywood, CA, in regards to the workflow that he and his co-workers had implemented on the production of “The Good German”. While this is beneficial when shared with the audience, it came at the cost of the crew doing various camera tests and having the luxury of being able to plan the audio workflow starting with the Production Mixer’s input all the way to what Mr. Blake knew was required for the final video delivery specs.

As we all know, not every production can be managed in this way. Large film productions have the ability and budget to work like this. However, more and more mid to low budget productions are beginning to move to the use of hard disc recorders. As some of us in post production audio already know, productions like these, often do not think all the way through to the final delivery. Often times, they rely on the sound department to deal with any issues that have arisen. Unfortunately, more times than not, the sound is already “in the can” before a post facility is even brought on board.

It has become imperative that a standardized workflow, or at least some standardized metadata and software functionality be discussed and developed. At the last seminar, the respective manufacturers made it clear that they do not want to force anyone on how to utilize their products. They put forward, as they saw it, four distinct audio workflows that have been developed by users, intentionally avoiding the promotion of any single one. (The Power point slides that accompanied this presentation were supposed to be made available on the Editors guild website,Editors Guild , but I can not find them) This is when I realized that, as in the past, it is not up to manufacturers to establish standards for anything. We, as the users, need to demand that the equipment and software developers provide the required functionality as it applies to our industry.

I put forth this post, as an attempt to start a discussion that, I believe is necessary. Regardless of budget, all productions should be able to benefit from the use of these new technologies. Reality TV to blockbuster film productions.

Here is a short list of a few requirements to help get the ball rolling.

1.BWF files should be stamped with the sample rate at which it was recorded, not the sample rate intended for playback.

2.Metadata should be standardized to contain the same information from device to device that all software applications can read and sort by.

a. Metadata should contain the sound roll number, scene, take, time-code, and track number (at least these if not more). Also any notes that the production mixer wants to add to a track, such as “Boom 1, Lav 1 etc”. So that the picture editor or dialogue editor can tell which files are which tracks once they are imported into whatever system they use.

3. Software support for the 10-track poly files that some of the recorders are capable of providing.


These are just a few of things I think should be standard. I do not consider myself an expert or an authority in anyway. I am just an end user who does not want to do things twice, when it could have been done correctly once. Or worse, pay someone to hand sync every line of a feature film! Yikes. I hope, those with developed workflows, will be willing to share them here, and be open to comments and questions. Everything from metadata input to digital cinema delivery requirements is fair game.

While there are no rules to this discussion, you may have noticed that I have left my comments platform and manufacturer agnostic. What will not be helpful, in my opinion, is any company or product bashing. I do, however, want to publicly thank the representatives of AVID, Digidesign, CAS, The Editors Guild, Zaxcom, and Coffey Sound who have already started this discussion, and made efforts to bring this subject to light. I believe that the DUC is a perfect place for this discussion. If anyone knows of any other forums, please feel free to post the link to this thread so that others can contribute. Let’s try to keep the discussion centralized. Hopefully, comments made here, will be influential in future product development. Thanks for taking the time to read this and to contribute.

[FlorianE]

Great initiative!

In the market I work in, production sound has almost fully changed over to disc recording, and the post facilities have already got their first wave of punishments this year, myself not excluded.
As I'm in PAL land, things in theory ought to be reasonably straight forward.
But without a standardized workflow everybody seems to feel free to reinvent the wheel. I've spent a ridiculous amount of time this year rounding off some of these new wheels, and some of my collegues here have been worse off than me.
Our place is trying to push the Titan/EDL route. Apart from metadata issues, I also had problems with production sound: naming conventions, track layout, even sound reports. Even with a recorder where the manufacturer has tooled and documented the whole workflow.
For a start, it is not easy to get detailed information from all sources involved.
Some essential tools also seem to be largely missing: namely a globally usable ingest software handling batch mixdown for offline edit, renaming, correcting metadata and logging to database - all in one, on at least the two main platforms.
It may be very useful if everybody made publicly available all the information they've already got hold of.
What has helped me most so far is The audio Post Chain, but then it is a bit manufacturer-specific.

Cheers
Florian

[Frank Kruse]

2.Metadata should be standardized to contain the same information from device to device that all software applications can read and sort by.

This already exists. It´s called iXML and supported by the HHB PortaDrive and the Cantar also Aaton´s InDAW Telecine-system supports it. Zaxcom and others have announced support for it. Only Digi and Avid anre going to go down the MXF route. AFAIK nop support for iXML is announced

a. Metadata should contain the sound roll number, scene, take, time-code, and track number (at least these if not more). Also any notes that the production mixer wants to add to a track, such as “Boom 1, Lav 1 etc”. So that the picture editor or dialogue editor can tell which files are which tracks once they are imported into whatever system they use.

This is already reality and can be done with the Deva 4/5, Cantar, PD6 and the HHB machines. The Nagra V for some reason does not write a TAPE ID but has only 2 tracks anyway ;-)

3. Software support for the 10-track poly files that some of the recorders are capable of providing.

Yes please. Mind that Digi has stated before that they only test and officially support stereo BWAVs. We all know PT loads 6-track poly files but this seem to be an un-official feature.

General comment on bext chunk and iXML: since the metadata in the BWAV chunkc isn´t standardized each mfg follows it´s own way of using the space. That´s why iXML was invented. It carries a standard set of additional info like "wildtracks" etc. etc. and can like in coexistence with the regular bext chunk. I really hope that Digi will support iXML one day...

Frank.




[/QUOTE]

[Tiago Silva]

MXF wouldn't be so bad if ProTools would gain a database-driven media management like an Avid. But the rest of the world is going with iXML. Are they mutually exclusive? Would it be possible to convert iXML to AAF and then import into ProTools or Avid? AAF is supposed to work hand-in-hand with MXF and is open-source.

BWAVs made with Sound Devices are imported into Avid bins with TC, scene/take and comments, but I don't think you can have a different comment per track (only per clip).

[Frank Kruse]

iXML, bext-chunk and MXF can all live together at the same time in coexistence. These are *not* actual file types they are types of meta-data.

The Avid already stores all the tape in info etc. inside every media files it creates but it´s stored in the OMF chnun k or wraper or what ever they call it and it is not completely read by PT AFAIK. Problem for 3rd party developers is that they need to get the OMF SDK to write software to de-crypt the avid´s metadata so it doesn´t compare at all to open formats like iXML etc.

fra

[Tiago Silva]

Of course they can coexist, as long as all applications can understand all of them. You said yourself that OMF keeps its metadata in special areas of the OMF file, but there was no way to export that metadata to other applications; besides, most or all composition and description metadata end up in the bins, not on the file.

So, what we need is both standards for metadata carried with file (AES31-level2), e.g. BWAV with BEXT and MXF; description information; and compositional information (AAF, OMF, AES31-3 and 4).

So, right now iXML is supplementing the capabilities of BWAV+BEXT with things the standard was not designed for, because standards are almost always a most common denominator affair to get everyone to agree to them. The rest of the information would be on the private chunks of each manufacturer. How does MXF handle this? Can you have more, or less information than BWAV+BEXT+iXML?

Would you call iXML description information? Is it something like a crossplatform "Master clip", in Avid language, but carried inside the file ? But I'm not sure if ProTools will ever be aware of that, because the concept of a "master clip" is alien to ProTools. (regardless of .L .R .C suffixes - yes, it's a start).

At least ProTools7 has folders and "region groups", but that's still 10 years behind the technology curve. Let's see how (if) Digidesign implements MXF.

[FlorianE]

Before I get confused:

Is this right:
MXF and AAF are streaming file types, which have to output raw data AND metadata concurrently,
whereas BWF and OMF are disk-based file types, which can separate these into different chunks?
And this:
the iXML metadata seem to be actually housed within the BWF bext-chunk (I took some of Mark Gilberts sample files and looked at them in a disk editor), so that iXML is a sort of implant extension of the bext definition?

But in any case:
The workflow problems I have had to deal with are on a much more human level. For instance:
sound reports (disk or paper) which do not link filenames to slate numbers, making it very tedious say to find alternative sound takes in sound editing.
Or inconsistent track layout over 8 tracks, making you loose a lot of time finding out whether track x is now a mono boom mike or the M of an MS, run in parallel to the mono boom.

The fact that each manufacturer uses a different file naming convention. Using alpha-numeric filenames, it should be possible within a 6-character name ( 6 because of EDL restrictions) to uniquely identify each machine/project/slate/take, as (27 characters + 10 numbers)^6 makes such a large number that there should be place for everyone.

If there was a 'standard' software, reading any brands location files, standardizing their Metadata with iXML, perhaps batch renaming them to fit into the EDL spec, automatically creating 2-track mixdowns with correct metadata for editing, and a sound report on top - wouldn't that go a long way to standardize a workflow?

Rail?? I'm sure more than half the code needed is sitting somewhere in a drawer...

Cheers
Florian

[Frank Kruse]

Of course they can coexist, as long as all applications can understand all of them. You said yourself that OMF keeps its metadata in special areas of the OMF file, but there was no way to export that metadata to other applications; besides, most or all composition and description metadata end up in the bins, not on the file.

So, what we need is both standards for metadata carried with file (AES31-level2), e.g. BWAV with BEXT and MXF; description information; and compositional information (AAF, OMF, AES31-3 and 4).

So, right now iXML is supplementing the capabilities of BWAV+BEXT with things the standard was not designed for, because standards are almost always a most common denominator affair to get everyone to agree to them. The rest of the information would be on the private chunks of each manufacturer. How does MXF handle this? Can you have more, or less information than BWAV+BEXT+iXML?

Would you call iXML description information? Is it something like a crossplatform "Master clip", in Avid language, but carried inside the file ? But I'm not sure if ProTools will ever be aware of that, because the concept of a "master clip" is alien to ProTools. (regardless of .L .R .C suffixes - yes, it's a start).

At least ProTools7 has folders and "region groups", but that's still 10 years behind the technology curve. Let's see how (if) Digidesign implements MXF.

I said coexist because someone asked if files carry either iXML, MXF oder other meta-data. I meant that all of them can live in the same BWAV file at the same time. That´s all.

iXML is *not* an EDL type of thing. I was talking about meta-data not transfer formats like OMF sequences and AAF. That´s something defferent. Even in a referenced AAF or OMF composition the assets (the media files) need to contain meta-data so each one can be identified even when it´s renamed. This is something PT doesn´t support but the AVID does (as did the AudioVision).

Frank.

[Tiago Silva]

The workflow problems I have had to deal with are on a much more human level. For instance:
sound reports (disk or paper) which do not link filenames to slate numbers, making it very tedious say to find alternative sound takes in sound editing.
Or inconsistent track layout over 8 tracks, making you loose a lot of time finding out whether track x is now a mono boom mike or the M of an MS, run in parallel to the mono boom.

The real problem is that everybody is forgetting the way things are done: there never was a need to quote a timecode on a sound report. You filled the sound report, each line was a take, it had a channel description, it had an index number or a bit of 1Khz to mark the beginning (remember Nagra?) and it had a spoken slate. Filling the sound report and calling the slate is still a good working practice that shouldn't disappear because some newfangled machines became computers. Hey, now jumping from clip to clip is instant! Just sort the clips by date/time.

The fact that each manufacturer uses a different file naming convention. Using alpha-numeric filenames, it should be possible within a 6-character name ( 6 because of EDL restrictions) to uniquely identify each machine/project/slate/take, as (27 characters + 10 numbers)^6 makes such a large number that there should be place for everyone.

I don't understand this. Those names wouldn't be human-readable. A01rdshjlsuefhs2314asa.omf is also not human-readable. The only thing readable from that filename is that it's an Avid OMF media file, it's audio and was recorded from the input 1 of the Avid's audio interface.

If there was a 'standard' software, reading any brands location files, standardizing their Metadata with iXML, perhaps batch renaming them to fit into the EDL spec, automatically creating 2-track mixdowns with correct metadata for editing, and a sound report on top - wouldn't that go a long way to standardize a workflow?

Rail?? I'm sure more than half the code needed is sitting somewhere in a drawer...

Cheers
Florian

Aaton majax is almost there already, and yes, it modifies the file names and last time I heard, it offered a kind of workflow when integrated with Titan. What I don't think is that we should be looking for solutions outside of MXF because Avid/Digidesign will not support them. However, this could be the event that topples the juggernaut.

[Tiago Silva]

I said coexist because someone asked if files carry either iXML, MXF oder other meta-data. I meant that all of them can live in the same BWAV file at the same time. That´s all.

iXML is *not* an EDL type of thing. I was talking about meta-data not transfer formats like OMF sequences and AAF. That´s something defferent. Even in a referenced AAF or OMF composition the assets (the media files) need to contain meta-data so each one can be identified even when it´s renamed. This is something PT doesn´t support but the AVID does (as did the AudioVision).

Frank.

So you could have an MXF wrapping around a BWAV that had iXML inside? You just need to fill the MXF fields with the iXML information (if they are compatible), or unwrap the MXF to get to the iXML. That doesn't seem too difficult if the manufacturers are up to it, but if it hasn't become clear from my previous posts, I don't know a lot about MXF. We're still on Avid Media Composer 7 (at least in terms of thinking) around here.

We do have an MC Adrenaline, but recently one of my colleagues exported an embedded OMF from it for import into PT6.4.1, even though we have an SNS SAN and she could've had exported it referenced, and only copying the relevant files when sending the project to mix. It would've made the several picture changes the project suffered a lot easier to manage.