Another Premiere pro question...

[Jimmy McKenzie]

The track is the track. Bring in a wav file file that has been normalized in your audio editor to 100% and there is no headroom but there is also no peaking. If the audio track from your firewire captured dv25 tape has no peaks and hovers around -2 or -3, then the export to your authoring program in the form of an avi will have no peak either.
To back the master track at export down to -12db as a absolute necessity is not a SOP.
Perhaps some of the anecdotal evidence in the Adobe forum points to this, but I have never seen this as the de facto method to go from edit to author when creating the finished work on a dvd.
If this can be substantiated, then I have been creating with the wrong set of standards for 6 years. How could we all be so horribly misguided in the way we manage our audio by stepping above the suggested -12db?
I think you might have a hardware issue that is shared by others in search of a technical resolution.

[Christopher Lefchik]

From Television Sound: The Basics (http://www.cybercollege.com/diglevels.htm)

"The optimum audio levels for digital audio signals are different than those for analog signals.

Whereas the 0dB peak setting is the standard operating level (SOL) for analog systems, for digital equipment the maximum level (in North America) is typically -20dB.

With both analog and digital signals it comes down to headroom.

Headroom is defined as the safe area beyond the SOL (standard operating level) point. With a SOL of -20dB, (which is typically the standard in North America) this leaves another 20dB for headroom. European countries tend to allow for -18dB of headroom. "
____________________________________

From Broadcast Engineering magazine, "Audio signal distribution and level measurements" (http://advertisers.broadcastengineering.com/ar/broadcasting_audio_signal_distribution)

“...all audio levels have a negative-dBFS value, with the SOL set normally to -20 dBFS indicating that the equipment has a 20 dB headroom.”

Originally Posted by Jimmy McKenzie

I think you might have a hardware issue that is shared by others in search of a technical resolution.

Actually, I’m not sure if my problem is the same as the others I read about, though it has been awhile since it occurred, so I don't remember the specifics well. I may master a DVD at close to 0 db again, though, just to see if it recurs.

Originally Posted by Jimmy McKenzie

To back the master track at export down to -12db as a absolute necessity is not a SOP.
Perhaps some of the anecdotal evidence in the Adobe forum points to this, but I have never seen this as the de facto method to go from edit to author when creating the finished work on a dvd.
If this can be substantiated, then I have been creating with the wrong set of standards for 6 years.

I performed an experiment. I extracted a portion of the audio sound track from The Patriot movie DVD. I used BeSweet to convert the Dolby Digital file to a WAV. It was a straight conversion with none of the compression or gain filtering that BeSweet can perform. I imported that audio into Premiere Pro, and checked it on the VU meters. None of it that I played went over -12 db. In fact, most of the audio lies well beneath -12 db. The portions that look to be the loudest (the section I was examining was the final battle) are two canon blasts, and these were the sounds that approached the closest to -12db. It looks safe to say from this that none of the rest of the sound track would probably go above -12 db.

Next I extracted the Dolby Digital audio from my latest DVD (mastered to -12db). I used the same procedure with this as I did with the audio from The Patriot. I then imported it into Premiere Pro and checked it on the VU meters. Again, none of it goes over -12 db, as expected, though most of my audio is pretty close to -12 db, unlike The Patriot sound track.

Thus, the sound track for The Patriot and the sound track from my DVD don’t peak above -12 db. In fact, my sound track looks to be hotter on average than The Patriot sound track.

Originally Posted by Jimmy McKenzie

How could we all be so horribly misguided in the way we manage our audio by stepping above the suggested -12db?

Simply put, most of us aren’t audio engineers. And while I use -12 db, apparently (from what I’ve been reading), the SOL is usually around -20 to -18 db.

[Jimmy McKenzie]

What a peice of research Chris! Thanks for the tech data and the experimentation.

We'll have to get to the bottom of this.

Thanks again! Ty? D.S.E.? Your comments?

[Christopher Lefchik]

Originally Posted by Jimmy McKenzie

What a peice of research Chris! Thanks for the tech data and the experimentation.

You're welcome! Thank you for the compliment. I have to say, it has been quite an educational experience for me as well.

Originally Posted by Jimmy McKenzie

We'll have to get to the bottom of this.

I agree. If it is true that the SOL should be about -20 db, as my research thus far has indicated, why isn't this more common knowledge? And if my research to date is wrong, well, then I want to know that, too.

[Dan Euritt]

chris, your testing methodology is flawed... if you want to accurately compare audio levels, import the files into an audio editor that gives you a visual representation of the peaks.

i'm certainly no audio engineer, but afaik, digital audio level meters as found on hardware like the mackie are not nearly as accurate as powered analog vu meters... and attempting to do it with software would be the worst case scenario... i believe that the inaccuracies with digital metering has something to do with the audio signal being averaged, instead of representing true peaks(??)... or maybe it's the other way around, lol :-)

the fact remains that premiere is apparently NOT handling audio correctly, you should never have to knock the entire track down like that... and this thread is even more confusing because paul has not weighed in with a definitive statement about whether or not his audio recording was clipped at the 0 db point.

it's easy to check for clipping when the audio track is opened up in adobe audition, which should be a mandatory step in troubleshooting an issue like this.

[Christopher Lefchik]

Dan,

At this point I'm not too concerned about any possible audio issues with Premiere Pro. What I want to know is what should the SOL be for recording, editing, and exporting digital audio. That was why I undertook that test. You claim "you should never have to knock the entire track down like that". Well, maybe not, though Premiere Pro mixes audio in 32-bits, which evidently means there is more headroom for audio inside the app than when exporting to a DV file. This is not Premiere's fault.

Further, you did not even try to address the articles I found, one from a broadcast engineer magazine, which stated the SOL for digital audio should be -20 db. In that case one should knock the entire audio track down, at least before exporting.

Originally Posted by Dan Euritt

chris, your testing methodology is flawed... if you want to accurately compare audio levels, import the files into an audio editor that gives you a visual representation of the peaks.

Premiere Pro does give one a "visual representation of the peaks", i.e., an audio waveform. I used that along with the audio meters to compare the two sources.

Even if Premiere Pro's audio meters and audio waveforms are not entirely accurate, they should be at least good enough for a rough comparison between two audio sources. As stated earlier, I had knocked down my DVD audio track to -12 db. It was still at that level when I imported it again after converting it from the Dolby Digital file; comparing my soundtrack to The Patriot soundtrack, it was very clear that the average level of The Patriot soundtrack was definitely below the average level of my soundtrack.

If Premiere Pro's audio meters and audio waveforms are so inaccurate that even a relative comparison like this cannot be trusted, then they are just about worthless.

[Dan Euritt]

what we are trying to tell you is that if the camera recorded digital audio at the correct level, you should not have to knock it down, o.k.? it should not be distorting, because it's not over the 0 db mark... and in fact, it most likely was recorded at a level well under the 0 db mark, because that's where a good agc would keep it.

i'm not sure why you'd post comparisons between digital and analog, since we are recording and editing in digital, but your point about the -20 db standard is a good one... i did some digging, you can download a -20 db wave reference file here: http://www.digido.com/portal/pmodule...der_page_id=59 ...bob katz has also written a ton of good info about digital loudness levels on that website.

please let us know what happens with that file when imported into premiere... the other thing that i would suggest is to first import that file, and some of these other files, into audition to compare the levels... also convert 16-bit files to 32-bit files within audition, and you'll be able to prove to yourself what happens to the db levels when you do that.

[Christopher Lefchik]

Originally Posted by Dan Euritt

what we are trying to tell you is that if the camera recorded digital audio at the correct level, you should not have to knock it down, o.k.? it should not be distorting, because it's not over the 0 db mark...

I completely understand what you are saying. However, the level the audio is recorded at may not be the optimal level that it should be mixed and exported at. This is the reason I am trying to figure out what that level, i.e., the SOL, should be for digital audio.

Originally Posted by Dan Euritt

i'm not sure why you'd post comparisons between digital and analog, since we are recording and editing in digital

Hmmm, I'm not sure what you are referring to. I am only interested in the levels for digital audio. I don't remember posting any comparisons between digital and analog. The Television Sound: The Basics article did mention analog audio, but I posted it because of it's mention of the correct levels for digital audio.

Originally Posted by Dan Euritt

i did some digging, you can download a -20 db wave reference file here: http://www.digido.com/portal/pmodule...der_page_id=59 ...bob katz has also written a ton of good info about digital loudness levels on that website.

Thanks for the link, I'll have to take a look at that site.

Originally Posted by Dan Euritt

please let us know what happens with that file when imported into premiere.

I'll be sure to do that.

Originally Posted by Dan Euritt

the other thing that i would suggest is to first import that file, and some of these other files, into audition to compare the levels... also convert 16-bit files to 32-bit files within audition, and you'll be able to prove to yourself what happens to the db levels when you do that.

Good ideas. Looks like I've got some experimenting to do. By the way, what other files were you referring to? I didn't see any other audio files on that site.

If we put our heads together hopefully we can figure this out.

[Dan Euritt]

i was referring to your dvd audio test files, vs. that -20db reference file... how they compare in various editors.

i think that there is actually an audio db standard with dolby stuff(??), which could explain why both of your dvd audio test files were at comparable levels... just guessing, i don't know what ac3 encoding standard you used... i typically shut off as much of the preset as possible.

[Christopher Lefchik]

Wow, I just finished reading the audio article you linked. It was incredibly informative. Absolutely great. This man knows his audio. I can't recommend it enough.

I didn't understand all of it, but here's the gist I got. Apparently, one should use a negative db point for the average sound level. Anything above that is considered headroom for peak sounds (explosions, etc.) up to 0 db. He says that with the new 24 bit systems one doesn't need to go all the way up to 0 db.

For film, the SOL was apparently derived from the Star Wars films, and is approximately -20 db.

The further up the scale the SOL is, the more compressed dynamically the final mix will sound.

His proposed standard for sound mixes looks good. From what he writes it looks like one should be able to set up monitor speakers to use his standard for mixing even without a sound level meter.

After reading his article and getting a much better understanding of the correct way to mix audio, I realize my most recent program mix (and probably all the ones I have done using audio meters) must sound pretty compressed dynamically, even when reduced by -12 db when mastered for DVD. The waveforms I looked at previously (The Patriot and my program) confirmed this. Most of my audio was in the same general region, while The Patriot soundtrack varied much more.

I hope to be able to give some audio test results from Premiere Pro and Audition soon.

Whew! There is so much to learn.

[Christopher Lefchik]

Dan,

Okay, finally got the tests done. Here is what I did.

I imported the -20 db file you found on that Web site into both Audition and Premiere Pro. In both programs the levels hit about -11 db. Rather odd.

I converted the audio tracks of three movies to wave files using BeSweet v1.4 via the BeLight 0.21 GUI. I used the default setting of -3db for "LFE to LR channels", with "Output Mode" set to stereo (I had also tried an LFE to LR level of 0db, but that ended up lowering the resulting levels, so I didn’t use the 0db setting). Below are screen shots of the audio waveforms from Adobe Audition 1.5 (WARNING: Large image files. I don't recommend opening them unless you are on broadband). The first two are entire soundtracks; the last two are only portions (it wasn't intentional; that was just way I had access to them). The Adobe Audition window has the decibel levels marked in it, so you can see what the levels are. Your browser will probably not open the images full size, but you should be able to click on them to zoom in.

Link: Gods and Generals audio waveform
Link: The Passion of the Christ audio waveform
Link: The Patriot audio waveform
Link: My project audio waveform

I also tried opening a 16-bit file extracted from an audio CD. I examined it at its native 16-bits in Audition, and dropped it into a 32-bit Adobe Audition time line (Audition converted it to 32-bits). I didn't notice any significant differences between the two; both hit the same peak (-1 db).

[Dan Euritt]

nice job! i had a reply typed up, and the computer ate it, lol!!

the gist of it was that i thought that the dolby encoding is tweaking those readings heavily... the total spl of 5 dolby tracks vs. total spl of 2 stereo tracks may not be accurately represented on a timeline, because there aren't the same number of speakers.

wrt the -20db target for tv audio... the 10/9/05 issue of tv technology has a user report from the engineer of a headend cable facitility, who had to get a dolby broadcast loudness meter to normalize the subjective dialog of all incoming sources before sending the programming on to his subscribers... so people at home watching tv don't have to ride the audio when switching channels... which means that most of the people creating and transmitting tv audio are not adhering to the -20db target level.

[Christopher Lefchik]

Originally Posted by Dan Euritt

i had a reply typed up, and the computer ate it, lol!!

That's frustrating, isn't it! I guess the computer was hungry, eh?

Originally Posted by Dan Euritt

the gist of it was that i thought that the dolby encoding is tweaking those readings heavily... the total spl of 5 dolby tracks vs. total spl of 2 stereo tracks may not be accurately represented on a timeline, because there aren't the same number of speakers.

Always, possible, of course. I can transcode the 5.1 Dolby Digital to a 5.1 .wav file, but I don't know if Audition can open it. I'll give it a shot, anyway.

Originally Posted by Dan Euritt

the 10/9/05 issue of tv technology has a user report from the engineer of a headend cable facitility, who had to get a dolby broadcast loudness meter to normalize the subjective dialog of all incoming sources before sending the programming on to his subscribers

I can't find the article on the TV Technology Web site, but was he by chance implementing the AutoNorm Dolby Digital metadata system (described in this article from Broadcast Engineer)?

Originally Posted by Dan Euritt

which means that most of the people creating and transmitting tv audio are not adhering to the -20db target level.

Good point.

[Dan Euritt]

the tvtechnology website did not have the article, probably because it was a user report, but the lm100 referred to in the link you posted was indeed the equipment in question... unfortunately it appears to be relevant to dolby audio only(??).

wrt premiere and what we have seen so far... if you encode the dvd audio to dolby standards, it will apparently limit the audio spl... it is adhering to the one common standard audio level that we were looking for, depending on how it's set up in the ac3 encoder.

so why are these people having problems with audio in premiere? i think that you showed that the audio levels are not changed when the file is brought into premiere for editing, which is just how it should be... and the vegas ac3 encoder should create the same spl as the premiere ac3 encoder, if the settings are the same.

[Christopher Lefchik]

Originally Posted by Dan Euritt

the gist of it was that i thought that the dolby encoding is tweaking those readings heavily... the total spl of 5 dolby tracks vs. total spl of 2 stereo tracks may not be accurately represented on a timeline, because there aren't the same number of speakers.

Okay, I finally checked out the sound level of all five surround tracks at one time. Audition can open a 5.1 file, but it splits the channels into 6 separate files. I dropped them all into the track window and played them back. I'm getting peaks close to 0db now. Of course the average level is still lower - dialog is -12db to -9db. The sound track I'm using is the one from The Patriot.

Originally Posted by Dan Euritt

i think that there is actually an audio db standard with dolby stuff(??), which could explain why both of your dvd audio test files were at comparable levels... just guessing, i don't know what ac3 encoding standard you used... i typically shut off as much of the preset as possible.

I don't know what kind of manual control you get over Dolby Digital audio encoding in your software, but the only manual control Encore DVD gives over Dolby Digital encoding is the bit rate (the default is 192Kbps, though I've recently tried setting it a bit higher manually).