Matching 0dB references

[jcupp]

I'm not sure this is a feature request or a bug report. This was all fixed in VT4 and it's all screwed up again.

0dB in SpeedEdit VT = some unknown value greater than 0, +6? maybe? in audio mixer with gain at unity.

0dB on audio mixer = -6dB files after capture (If I wanted -6 I would have set it that way when I recorded it!)

0dB on an individual input = what looks to be +8 on the master VT Input meter

It's all messed up and it drives me nuts.

Are the meters simulations of an analog VU meter where 0 is 'nominal' and there is some amount of headroom above?

Are they peak reading digital meters where everything above 0 is clipped? If this is the case can we get a mark at -20dB.

What should be obvious in the meters is where the digital clipping starts and where the nominal level should be and it should be the same on all the meters.

 

[SBowie]

I'm going out on a limb and re-posting some info from the VTNT list, originally posted by the good Dr. Cross (I have made one small edit to comply with a subsequent correction he made). Hope this info helps:

This issue has always been a confusing one, and although what we were
doing was "technically" correct, I think that we had made things
probably far harder to understand than was necessary.

What we have done in VT[5], and SpeedEDIT is simply move to a
"convention" that is followed by almost every other windows sound and
video application and is the way that DV normally handles it. This would
be that while you remain in the digital world 0dB is the loudest signal
that you can get without clipping. So, the VU meters on the editor for
instance turn "red" at 0dB meaning that any signal above that will be
clipped. When converted to analog (or DV) the 0dB level is boosted to
+12dB which is the industry norm. This process is reversed exactly on
record so you consistently get 12dB of headroom through the system and
levels should always be preserved exactly.

The history of our audio levels was that when we did a survey to find
out what people needed in VT[4], we found that almost all users of VT[3]
had very serious issues with our audio. Needless to say we dedicated a
lot of effort into better audio support for VT[4] ... what we did is
really try to do things "right" by allowing user selectable reference
levels on all files both in reading and writing, etc...

No while this really did give us technically completely correct audio
support it confused everyone beyond reasonable belief. The number of
times I had to try to explain it both inside and outside the company was
quite scary.

As a result, with SpeedEDIT (and VT[5]), we did what we probably should
have done in VT[4], we looked at how all other apps are handling it and
made things work in the same way. Other vendors have clearly put time
and money into working this problem out and it makes more sense to go
with the flow instead of re-inventing the wheel.

 

[jcupp]

Well what Dr. Cross said would be OK I guess, screwy but OK but that doesn't seem to be how it's working.

I'd prefer not to be lied to by the meters even if it is for my own good

Let me record as close to digital 0 as I dare for maximum signal to noise and then before I go out to some media (where there is a standard) let me easily match that standard e.g. -12dB for DV and DVDs, -20dB for DVC Pro etc. 0dB for audio CDs.

So my work flow now is like this:
Bring up analog input adjust so peaks are close to 0, the meter is hard to read so one must be careful.

Adjust the main VT input to just about reach 0, I guess. I'll have to do some experiments to see what 0 actually means on this meter.

Adjust the main output, from where I will be recording, to +6 for maximum S/N ratio.

Record

Edit

Before doing a final render add a basic audio filter to the entire project and adjust it so that my audio is at -6 so that when the software knocks down my recorded file, my audio will be at -12dB where I want it for my DVD.

It's a pain.

I'd rather as I did in VT4:

Feed my properly recorded (my peaks are just at 0) audio into the analog input input. And since my playback deck's analog output is at -10dBv (or +4dBm I just need to adjust my Mackie mixer to compensate for the non-standard input levels that the VT hardware wants) I leave both the channel input slider and the main input slider and the main output sliders at unity and my file records with the peaks at 0dB just like they should.

I then record and edit my project and the only adjustments are made for aesthetic purposes or to keep multiple audio tracks recorded at 0dB from pushing my level too high.

At the end I select a reference level of -12dB for my final render to an MPEG file for my DVD. It will now be a similar loudness to a properly recorded commercial DVD as rare as those are.

This way at any point in the process I know exactly where my level stands, and I can manage my headroom as necessary for the program material and I don't have to make many adjustments.

 

[CreatvGnius]

Would you be merrier, Jeff, if we could revert SpeedEDIT-VT to the previous, NewTek-designed gain structure (what should we call *that*, pray tell?), via pa PREFERENCES setting?

I was happy with VT[4] gain methodology since it seemed to result in less guesstimating throughout the project.

Again, if the "VT[4] GAIN STRUCTURE" option were to be provided for within PREFERENCES, it would almost certainly have to have a name -- and a very well-written description of just what that setting does!

Well what Dr. Cross said would be OK I guess, screwy but OK but that doesn't seem to be how it's working.

I'd prefer not to be lied to by the meters even if it is for my own good

Jeff, that last line above struck me so funny, I nearly choked while spilling grape juice (no less) on the front of my faved casual shirt!
LOL
-PeterG

 

[ScorpioProd]

But the meters aren't lieing, it is simply a question of which standard they match up to, the legacy analog one or the modern digital one.

I think it's really more an issue of if you are looking for the VT to simulate the old analog world it connects with or if you are looking for it to more closely match the modern digital NLE world.

Because the VT always has been and still is linked to the old analog world, the analog simulation does make more sense for the live use world.

But for the pure NLE world, the current method makes a lot more sense and is a lot simpler.

So, the only answer is as Peter suggests, having a user preference. Problem is, this would have to actually change ALL the meters, so I'm thinking it's probably too complicated a solution and it won't happen.

 

[jcupp]

But the three different meters in the audio mixer read three different levels with the same signal and everything set at unity.

If we are going for the analog simulation I want true VU ballistics and at least 12dB between 0 on the meter and digital clipping. If we go for peak 0 should be 0 on all the meters.

In either case they should be calibrated to the same thing.

A pretty good primer on this and related subjects

How about switches below all meters so we can switch them between honest to God VU (0 = -12dB with averaging ballistics )and Peak Reading (0 = 0dB with fast peak ballistics) and a chapter in the manual explaining the difference.

 

[CreatvGnius]

VT[5] Audio Mixer Readings: Reliable?

:bump:

But the three different meters in the audio mixer read three different levels with the same signal and everything set at unity.

That phenomenon described by Jeff -- anybody else seeing this in VT[5]? Things look good from that standpoint in VT[4] version 4.6 here. All meters are behaving as I believe they should, throughout the gain structure, from input through to VT Audio Mixer output.

So, is this an issue unique to Jeff, or are other VT[5] users with the SX8/84 BOB experiencing the same variation among meters, for the same analogue stereo input signal going throughout the Audio Mixer?
-PeterG

 

[jcupp]

Things worked fine in VT4 but it's screwed up in VT5. Here is a 0dB (16 bit) sin wave file playing in SpeedEDIT. Notice that it is indeed 0dB on the SpeedEDIT meter but at +6 on the meters in the mixer.

The problem with the mis-match between the input and master meters seems to have gone away since I looked at this a couple of months ago. I think this is a different VT card so maybe that was a hardware calibration issue.

 

[CreatvGnius]

"Sinful Obomination?"

That's the metering result of a 0dB (16 bit) sin wave file playing [from] SpeedEDIT-VT, into VT[5] Audio Mixer?
A smooth sine wave?

Wh-- w-wh-why, that's just abominable.
:devil:
PeterG

 

[CreatvGnius]

Seriously, though -- glad it's behind you, and thanks for updating us, Jeff. 'Wonder what caused that.
:hey:
-PeterG

 

[jcupp]

I have four VT cards so it's easy to forget with which card I was doing a particular test. Especially as I'm always swapping hardware around into different configurations. I'm also up to my neck in several different programming and scripting languages right now so I have sin() on my mind I guess