Why does my ME want me to keep 3db headroom?

[huejahfink]

Originally Posted by dcollins

I was doing fine until: "A little secret: -0.5 is usually a little bit grainy and digital-sounding, but -0.6 can be rapturously warm and fuzzy."

[Paul Gold]

Originally Posted by Laarsø

Indeed, the lacquer is destroyed during the mastering process.

For an ace pre-mastering clerk obsessed with semantics I thought you would know the difference between mastering and plating. The above is not correct. And with luck you could replate it.

[dcollins]

Originally Posted by Bob Olhsson

To me the main reason to print moderate level mixes is so that I won't risk compensating in the mix for the sound of overtaxing my analog monitoring chain. When it comes to digital audio. erring on the low side always seems to sound way better than erring on the high side.

Agreed. And using "pro" equipment is not always a panacea, either.

Here are some plots of a modern D/A chip that has no internal headroom for the overshoot of 0dBFS squarewaves. You know, the kind that come from loud masters. This is from the chip itself, no amount of headroom in subsequent circuits will help.

The first two are at -6dBFS. Sine and squarewave.
The third and fourth are at 0dBFS. Note the flat-top of the squarewave. You can also see the horizontal cursors at the same location on all plots to show the amount of overshoot.

Is it the end of the world? No, but it does show the phenomenon is real......


DC

[12ax7]

Originally Posted by dcollins

[...]

Is it the end of the world?

No, but it does show the phenomenon is real......

[...]

.

Thank you for that very level-headed response to all of this, Mr. Collins.
.

...And with respect to this particular issue:

Obviously, while this phenomenon is not "the end of the world", please consider that if it happens to various elements of a project several times during its production, the effects will be cumulative.

...And...

If the ONLY result of the ME's request for a certain peak-level is that the whole damn thing just gets scaled-down after the fact, it really won't matter much whether this is done before or after the ME gets hold of it.

...Hmmm...

Come to think of it, if this were the case, there might be less chance of somebody hitting a wrong button somewhere with the ME doing it.

...Just sayin'...

.

[Jerry Tubb]

As Mastering Technicians, we regularly receive mixes with a wide variety of peak and rms levels, everything from mixes with -20dBFS peaks, to mixes we need to attenuate by 10-12dB just to pass through our analog EQ path, from 16-bit to 32-bit float.

Regular clientele know we like to recieve nice dynamic mixes, but others deliver as they like it, even slammed on purpose.

So you just learn to deal with it as part of the Gig.

the OP has a valid question, the rest of the thread, another typical noisy GS debacle ;-)

Cheers, JT

[dcollins]

Originally Posted by Jerry Tubb

As Mastering Technicians,

How dare you insult Clerks everywhere with this gross distortion!

the OP has a valid question, the rest of the thread, another typical noisy GS debacle ;-)

Looks like I missed some more okidoki-isms. I should turn the email responder on so I can see the genius before it's deleted.

You know, GS actually has the best S/N ratio of any board. Because it also has 10X more users than any other. Maybe 100X.

So things could be worse.


DC

[impact87]

Originally Posted by Cheebs Goat

This does not mean mix at full scale and then lower the mix 3db before sending it off.

Originally Posted by StringBean

Pardon my ignorance but could someone elaborate on this for me?

Me too... Would having a mix that peaks at -.5, then lowering the master down to -3db still be an issue?

[Alécio Costa]

Hi,

I use Pro Tools HD for mastering. I know there are several softwares on the market that can handle 32 bit floating point word length.

1) Are there real advantages mastering at 32 bit floating point?

2) Also, at 32 , doing SRC, are we free from conversion errors like the one where ceilings at -0.2 may go up to -0.1 or 0 db Clipped?


BTW... I recommed to my clients mixes with Peaks < -3dB and RMS < -19 for mastering.

I also see lots of problems related to calibration playback.
Last week I received 12 songs dumped at my FTP server where I had pop/reggae songs ranging from 0dB clipped/RMS -14 (!!!) to -9dB FS/RMS -27.

Thanks

[huejahfink]

I'll answer 2) for you, Alécio.

SRC will very likely still raise your peak levels in the same way... however - nothing will actually clip. So then if you get peaks over 0dB you can either choose to employ a limiter post SRC to catch them, or just change the gain.

I suppose that's the advantage in a nutshell - you can use 32bit digital processes without worrying about clipping.

[Nordenstam]

Originally Posted by huejahfink

I suppose that's the advantage in a nutshell - you can use 32bit digital processes without worrying about clipping.

Now introducing the floatamaximizer; +1550dB gain to cater for all your clipping needs, even in floating point!

[Nordenstam]

Originally Posted by dcollins

Is it the end of the world? No, but it does show the phenomenon is real.....

Nice graphs.

Have had some occasions where I've walked into a new setting and instantly proclaimed that there's a macbook source with the level control at max. Bingo. There's a sound to it and it comes from the gear, not the waveform.

[huejahfink]

Originally Posted by Nordenstam

Now introducing the floatamaximizer; +1550dB gain to cater for all your clipping needs, even in floating point!

Hilarious. Have you considered a job as a comedy writer?

[Jerry Tubb]

Originally Posted by Mr. Sheen

Best S/N ratio of any board? Really? That's hysterical.

I think Dave was kidding, by intentionally saying the reverse of the truth.

JT

[MASSIVE Master]

Originally Posted by Jerry Tubb

I think Dave was kidding, by intentionally saying the reverse of the truth.

But that would be like "sarcasm" or something --

[Zep Dude]

Originally Posted by MattGray

...However I don't subscribe to the notion that a dynamic mix peaking -18dBFS is not going to be perceived as lesser quality than the same mix printed to a peak ceiling of -3dBFS (or higher) & here is why.. if you have a mix peaking to -18dBFS but the RMS is around -40 than the quieter parts of the mix are definitely going to be losing some significant resolution and be more likely to be heard as a slight loss in fidelity. The one or 2 peaks that reach -18dBFS will sound fine, pity that the rest of the mix which our ears are more in tune with (the RMS) will sound slightly lower quality.

This is a popular misconception. Each bit has equal resolution of 6db. The 1rst bit has just as much resolution as the 24th bit. When it comes to bits the only issue is noise floor. A 24 bit mix with RMS of -40 is still 104db above the noise floor (more than most converter specs). It's RMS is 8db higher than a 16 bit CD.

[12ax7]

Originally Posted by Zep Dude

[...]

Each bit has equal resolution of 6db.

The 1st bit has just as much resolution as the 24th bit.

[...]

.

Yep.

...However, I must admit to being hard-headed enough that it took me quite a while for the reality of this to dawn upon me.

.

I finally grasped it when I realized that when you use math to create a chart, the accuracy of that chart is every bit as good at the bottom of the chart as it is at the top!

(This is basically what I was trying to get at back at Post #17 of this thread.)

.

[dcollins]

Originally Posted by Jerry Tubb

I think Dave was kidding, by intentionally saying the reverse of the truth.

When you measure noise, you have to ask "over what bandwidth?" Frequently I see 200-500 people viewing this forum at one time. Maybe even more.

This is more at one time, on one forum, than the total membership of other boards.

So, when you include the large number of users (wide bandwidth), GS actually has a good S/N ratio. If you have 10 users and 4 of them are puds, that's a bad S/N.


DC

[Bookerv12]

Hi,

We sat down the other day and made some notes for the projects that have come through here in the last year.

Three mastering journeymen.

I'm not going to get into the theoretical stuff, complete with graphs and all...

About 70% of songs come in really hot.
I mean really hot.
The pro's achieve it without limiting and tricks.
The not so pro's insist that there is nothing on the 2 bus, but you can hear the grainy crush of L3's and such. Complete with the brick shaped waveforms.
Whenever I have had the opportunity to go to their studios and see, there are always L2's, L3's, or whatever flavor of the month loudness slammer on the session. (Grayed out of course)

About 5% are pretty loud and sound a lot better.

About 25% are around -20RMS or less. The peaks can be as high as -5.
These mixes always sound much better. No comparison at all.
If the clients ask for stupid loud, we can do that and they still sound great.

Out of these 25% of mixes, the mix technicians are usually over 40 years old and have been doing this for years.
There are rare exceptions, but those tend to have been involved with sound for film or high end audio on a pro basis.

It is almost impossible to get real young guys to take the expensive hardware comps that they don't quite understand off the mixes.

Out of the 70%, we end up getting about 4 revisions of the tunes before they are worth working on.

Sucks!

[Zep Dude]

I've can't say this enough to mix engineers but the trick to having your mix sound great in mastering is to send us your mix without the brick wall limiter.

Send whatever you want to your client, let them hear it the way they want, but print your mastering version without the L2 etc.

[Jerry Tubb]

Originally Posted by dcollins

When you measure noise, you have to ask "over what bandwidth?" Frequently I see 200-500 people viewing this forum at one time. Maybe even more.

This is more at one time, on one forum, than the total membership of other boards.

So, when you include the large number of users (wide bandwidth), GS actually has a good S/N ratio. If you have 10 users and 4 of them are puds, that's a bad S/N.

DC

Har! I should've known you'd do it by statistical analysis!

How do you rectify the "register members" viewing -vs- "guests" -vs- trolls?

Best, jt

[Bob Olhsson]

Originally Posted by dcollins

...So, when you include the large number of users (wide bandwidth), GS actually has a good S/N ratio. If you have 10 users and 4 of them are puds, that's a bad S/N.

This is exactly why big stars often need to hire bodyguards.

[doctormillice]

Originally Posted by Jerry Tubb

I often receive mixes that I have to turn down 6-12dB before going through my analog chain.

I tell clients that at the mastering studio we have all the level you need, at no extra cost.

So there's no reason to clip the mixes.

With boatloads of dynamic range at 24-bit there's just no need to push it,

just go for a great sounding mix.

Cheers, JT

Jerry For the Win!

[mdoelger]

Originally Posted by dcollins

What's the Y axis on those plots? Is it listed? Are these harmonics ever >90dB below the fundamental?

Am I really engaging Okidoki?


DC

What makes me wonder is why does the frequency spectum in the fourier transform graph only start at 1khz. And how can that always be below the fundamental of a piano note?

Very strange article

[Unclenny]

Originally Posted by dcollins

So, when you include the large number of users (wide bandwidth), GS actually has a good S/N ratio. If you have 10 users and 4 of them are puds, that's a bad S/N.

See?

I knew you were serious.

Oh, wait! Did I just skew the ratio?

[da goose]

Do you think this loud enough?

and yes.. this was how i got it in.

[firubbi]

Originally Posted by Jerry Tubb

I often receive mixes that I have to turn down 6-12dB before going through my analog chain.

what should be the mix level before send for mastering?
thanks

[MASSIVE Master]

SOME (as in ANY -- A half dB, two dB, 6, 10, 12dB, doesn't matter an awful lot) headroom with no limiting or excessive compression on the 2-buss.

Most of those will almost take care of themselves. Figure they'll be riding anywhere from -25 to -18dBRMS or so -- A fine and reasonable place to be.

[Jerry Tubb]

Originally Posted by firubbi

what should be the mix level before send for mastering?
thanks

Yes, 3 - 6dB of headroom is nice, with No limiter on the master bus.

Mixes fully dynamic, or perhaps with light to medium 2-bus compression.

But anything is fine, as long as it's not crushed like a brick before it arrives.

We can "restore" some peak information with RX2 DeClipper, although it's an educated guess.

Falls into the area of pre-mastering processing, making corrective moves (usually ITB) before the mixes are passed through the analog EQ path.

Gain adjustments, level automation, surgical EQ, spot DS'ing, plosive repair, HPF, deClicking, deCrackling, tape simulation, M/S adjustments, etc.

Maybe even a touch of reverb in rare instances.

Sibilance is another big issue, vocal brightness anywhere from 2.5 - 10k to be controlled.

It can be processed offline, or in real time on the front end of the pass through the mastering EQ path.

Best, JT