MPEG2 audio and VUmeter levels

[Sciascia]

I have a device called a 'telestream' at work that encodes to MPEG2 in realtime and then I can send the video to an FTP server or another telestream device. It works great. The question that I have is about audio levels. I have always been taught that 0VU on an analog meter is -18dB on a digital meter. Well I am having problems getting a good solid clip to compare levels with and have clipping issues. What makes it worse, I am coming from a digital editor (old Mac based AVID Media Composer) into an analog mixer at 600 Ohms balanced and then out of the mixer using the tape bus using unbalanced outputs. The telestream has only unbalanced inputs so I figured that the levels should be right. I recorded bars and tone from the internal test signal generator on the Avid, levels looked perfect on the mixer and when I demuxed the stream and analyzed it, it played back at -17.8dB. Not too shabby. The problem is that when I use this as my reference, anything that peaks over 0dB seems to clip instantly. There are 3 settings for input gain, +10, -10 and 0. It really does not seem to matter what the level is in the encoded clip, even at +10, it sounds fine until the level is over 0 on the analog mixer. Of course the overall levels are 10dB apart, but they only sound bad at the same parts. They just have different levels of loudness.

So to summarize my lengthy question, what level should I be using for MPEG2 audio? And if I am using 0VU on the analog mixer, shouldn't I be able to have active audio go as high as +4VU without overmodulating or clipping?

If it was strictly a level problem that I would assume that the +10dB clip would be clipped all the way through but that isn't the case at all. And the audio played straight to the mixer sounds fine (just spoken voice from a standup intro).

[smitty]

Not sure if I read your question properly, however thereis no + db in digital. anything over 0 db will clip/overmodulate.
Personally , I use -6db as max. -10db as my target
unlike analog there isn't as much advantage to hitting 0db
the noise floor is typically much lower with digital
smitty

[Sciascia]

Actually I meant +4VU on the analog meter, sorry it was late when I wrote it the post, sorry. The problem is that there is no metering tools on the telestream so I have to use the analog mixers VU meter to ride my levels.

That is actually about the same levels that I have always shot for, but even at much higher levels or lower levels it overmodulates at the same point, regardless if it is at -23dB or -3dB. It is the strangest thing. The rest of the audio is fine, it is just this one spot.

[landrover]

Digital -18dB equals analog -9dB, not 0dB. Seems like you calibrated the system at too high a level, resulting in clipping at peak levels.

[Sciascia]

Originally posted by landrover
Digital -18dB equals analog -9dB, not 0dB. Seems like you calibrated the system at too high a level, resulting in clipping at peak levels.

That is what it was looking like this afternoon before I had to run out to do something else. I ran several tests at different levels and coming back into the analog mixer after recording I do not have what looks like unity gain. I am not real worried about that at this point as we are not using it as a source on our end, but I really need to get with someone on the receiving end and see how they are editing and at what levels they need them to be at.

This whole balanced to unbalanced from digital to analog to digital thing is getting on my nerves. It really isn't that complicated, it is just a matter of everyone getting on the same page and using the same levels. Avid apparently uses -15dB as 0VU and while that technically is within digital specs, a 3dB difference can be a real issue going back in.

[Brian Ellis]

It's a pity broadcasters can't standardise on sound levels on digital broadcasts. I watch regularly four digital satellite news programmes and there is a range of over 10 dB difference between them. Ditto for commercial DVDs, although I think the biggest discrepancy is possibly less than 10 dB.

Sorry to hi-jack the thread, but it is related.

[Sciascia]

Originally posted by Brian Ellis
It's a pity broadcasters can't standardise on sound levels on digital broadcasts. I watch regularly four digital satellite news programmes and there is a range of over 10 dB difference between them. Ditto for commercial DVDs, although I think the biggest discrepancy is possibly less than 10 dB.

Sorry to hi-jack the thread, but it is related.

No it is related very much so. Most of our stuff is news and small commercials and I happened to be with one of our operations directors last night (we were working the Landstuhl satellite feed for the SuperBowl) and I got to make sure that she 'overheard' me talking to her chief engineer about the level problem and then pointed them out on a series of spot breaks. She had a horrified look on her face so I think that this is going to become a priority.

I'm tired of it, I spend tons of time aligning these systems to within ridiculous specs and it sounds like crap because people don't take the time or know how to adjust their levels properly. And as an engineer it always comes back to us and they ask 'what is wrong, you need to fix it.' I am going to fix it, I'll probably end up writing up another training package and sending it out. As if I don't have enough going on as it is. And I know for a fact that these people have been trained on this at some point, it is just a matter of following the standards. Part of the problem is that a lot of programming comes from the states and it is coming off the bird hot as hell so people are compensating to try to fix the levels. Problem with that is when someone does follow the standard than their audio is about 10 db lower.