Reference Level for Broadcast Audio
 

I apologize if I get any of the terminology wrong; I am not really an audio guy.

I work for a small production agency. When we output to a Beta Deck we use a -12 db reference tone to set the level on our mixer and then on the the Beta Deck. We make sure the reference is set to 0 db on the mixer and Beta Deck though. I never really thought about it until recently but should the mixer and beta deck levels be at -12 db as well when the tone is playing?

SMPTE specs call for 0VU tone to be set to -20dBFS, peaks not to exceed about -10 dBFS

OK, please correct me if I am wrong because I am a duffer when it comes to audio.

When I run a 1khz tone at the beginning of the tape, Final Cut Pro's audio meter should be at -20dBFS and the VU meter on the beta deck should be at 0 and nothing should register higher than -10dBFS on FCP's audio meters.

That's the gist of it. This evening I'll look up that exact numbers for you.

Thanks a lot for your help, Steve. I am trying to get a little better at the audio production side of things.

One of the problems in translation is the FCP meters are digital peak reading while the Beta deck (I'm assuming Beta and not DigiBeta) meters are analog averaging VU meters. They're really apples and oranges.

if you do that, you clipped the beta deck out. the problem is its all relative. what you do in your NLE and what you put to analog tape are TWO different things. In digital land, the spec is -18, not -20, with peaks hitting -12. thems the specs I've been handed from numerous TV stations. thats digital.

in ANALOG its all relative. this means what ever you set as your digital level = 0 analog, you don't want peaks going more than 3db above. techincally, betaSP will take +6 over zero, but since so many folks are using UVW series machines that indicate clip on the meters at +4, don't do that ! instead keep you peaks to not exceed 3db of your reference level. I preffer to mix at -12 = 0 to maximize S/N ratio and other stuff within the digital environment when stuck with 16 bit audio. therefore if I set -12 digital = 0 analog and don't let peaks past -9 digital, all is good. same as mixing to -18 digital with peaks not going past - 15. of course if you want to save a lot of time, get a real analog compressor and set it as a peak limiter. this is what I do to catch and fix the occasional peak when going out to analog so save a lot of time. sure some folks may say use a compressor in the NLE, but it just doesn't work the same unless your NLE ( like Prem Pro ) can apply the compressor to the final output channels rather then each clip or track. big difference between the two.

The specs I have been given by national broadcast networks differ from what you suggest. And are exactly what Steve House has posted.

Digital for 0VU tone to be set to -20dBFS

I might add, conventional analog broadcast and VHS tape has about 8db headroom. So setting peak limiting to -12 dBFS in your NLE might be a bit safer than the -10 dBFS I originally suggested. Also adding, "-20dBFS" holds true when reference level 0VU corresponds to an actual signal level of +4dBu.

well the problem is there is no real standard in digital audio. I once went thru 4 sets of specs before finally having a conferance call with a head engineer and two other guys before getting the real specs because none of them new. national network BTW. conversation was over in 2 minutes, but before I got to there everytime I requested specs, I got analog ones. the best one I got was for 1" and reference to how the SC wasn't allowed to vary by more then such and such. Another they replaced BetaSP with HDcam which was just as funny, I think ;\

There is so much irrelevant discussion of digital signal levels in this thread - Beta and BetaSP are analog. If you end up with meaningful program material below -20db you're going to have one noisy tape, the noise floor aka. tape hiss *will* be audible when the tv engineer (cursing) turns up the deck outputs.

Fuggetabout what the equivalences are in the digital world - doesn't matter. What matters is that tone goes on the tape at 0VU, and that program peaks hit at least 0VU. Some stations say peak to 0, some say peak at least to 0.

What Steve Oakley said!!!

The VU meter on the record deck is the *only* meaningful reference. What the customer/tv station wants is the only meaningful standard. Use any digital workflow you want to get there, but watch that VU meter!

and the peak meter.

Regards,

Ty Ford

To add insult to injury, what TV wants and what film wants, are two different things. But then, the title of this thread is "....Broadcast Audio"

You're right in that what gets to the tape is what counts. That's why I added the bit in my last post, that according to SMPTE, reference tone at -20dBFS digital should produce an output level of +4dBu which in turn should result in a reading of 0VU on the analog recorder's meters. If things aren't calibrated throughout the entire chain, all bets are off. And just to make it interesting, the EBU uses a 0dBu reference tone set to -18dBFS digital.

One problem is the VU meter is not a peak reading meter so peaks can go too hot while the VU meter sits there well behaved and apparently within limits.

'The Truth' is visually right here.

What's the source of that chart? Everything looks right to me except the SMPTE bar - it appears to be offset downward. According to all my info, -20dBS is supposed to correspond to +4dBu. (Holman, "Sound for Digital Video", pp144-145 among many others)

Steve, I got it here from SoundonSound.

http://www.soundonsound.com/sos/oct0...cles/qanda.htm

Now I need to cross ref this.

UK Style
 

In England we have always had the PPM as king and set 1K tone at 0db or -18 digital.

We then peak at +8db or -10 digital.

Gives you 10 db headroom in the digital domain.

The PPM meters are very fast but stil take about 10ms to register so the apparent peak of +8/-10 might in fact be +10/-8. Still gives one 8db of headroom in the digital domain.

I think digibeta records at 20 bit and no one seems to mind having 8 db of headroom. You could if you were recording to cheaper, (semi pro) equipment peak a bit higher if you feel secure with your levels ( say you have limiters).
Personally I would be happy with less of a safety margin but the standard came about to avoid mistakes and was set with a high safety factor.

The above chart is very helpful.

In the old days of 1/4 inch tape we set our tone, (100hz, 1khz, & 10 khz) at PPM 4 ( 0db) and at -4vu on the revox1/4", 1"vt, beta SP etc. And peaked at an apparent level of 8 db above tone. (+8 db or PPM 6)

Just read the Hugh Robjohns Sound on Sound article and although complicated it is the definative article.

He says that the PPM could be letting through as much as 4 to 6 db during that initial 10 ms before it can respond.

(In the old days the PPM was considered a very fast meter and VU was slow).

That would mean peaking at +8db / -10 digital could really be hitting +14db / -4 digital.

That would mean only 4 db of safety/headroom in the digital standard which makes sense.

So there we have it.

My question is....

Are the digital meters telling the truth then?

The SOS article states that they read the sample values and not the combined value at the DA which indicates that the need to leave the stated headroom in digital systems is neccessary to avoid clipping. I think we can asume that the lowest levels of clipping are not actually noticed by us but are none the less there and that what seems like excessive safety margins are in fact required for complete fidelity. ( ie 0% clipping).

Funny how this debate rages on these forums and never seems to go away. And we all thought digital would be straight forward compared to anologue!

I've read this thread three times and am still confused.

We're doing web video and trying to match volume levels to pre-roll ads, which I'm assuming are broadcast level at some point in their lives.

We're editing in Final Cut and can't get near the audio volume of the ads without clipping. We're not going to tape but straight to the web. I keep pushing the levels up higher and it ain't enough.

Can anyone point me to some sort of reference for this kind of work? We've experimented with Levelator and 'normalize' in Soundtrack but it sure slows things down for news work.

thanks much

Hello Chuck,

Here's the dirty little secret of audio production. You're right. You can't use gain to make other content sound as loud as spots. The reason is that most spots are compressed and then limited.

Next time you have your editing program open use the waveform view. Notice that the spots are more like thick rectangles, whereas the other audio is not as dense. That's what compression does.

Do all of us a favor and knock down the level of the spots a bit and compress and limit your other audio a bit until they match.


Regards,

Ty Ford

Thanks, Ty. We're feeding local video into a national product, so we can't take the ad levels down.

Is there some way to compress and limit our audio within Final Cut -- a plug-in or hardware add-on? I see compressors in the catalogs, but they seem to be designed for analog-to-tape workflow?

Yes FCP has that in audio filters and if you don't find satisfasction there, Sountrack Pro does as well. Just remember NOT to process the spot audio.

Regards,

Ty Ford