Format for Mastering.

PaulBrewer

Does everyone use 24bit to send to Mastering? At what frequency?

DamagedTrax

24/44.1

PaulBrewer

Recorded at 44.1?

woodsdenis

PaulBrewer wrote: »

Recorded at 44.1?

You should send them whatever the final master is recorded at.

16/24 bit , 44.1/48/88.2/96 whatever, and let them dither etc. Proper mastering houses should have much better conversion hardware/software than most of us do in our systems. However it is always worth checking what their requirements are. Mastering used to be a very specialized and expensive business with very expensive hardware/software. Nowadays there are many Mastering services who use the same setup you have at home.

I have had a track re mastered for a compilation which was done by an inhouse record label Mastering studio. The clocking was wrong and the Master came back with a digital clocking error which sounded like a single shaker at 1 sec intervals on the right channel. Nobody checked, or sent a reference to the artist.
50,000 cds were duped and had to be recalled. This was a major American
record company. When the mastering engineer was asked what happened he said he thought it was part of the track!!!!!!!!!!!!

Always get a reference Master

The only exception I have come accross is for Film/TV where the delivery spec has to be 24/48 or more often than not 16/48.

Denis

trackmixstudio

24/44.1 here too.
As Denis says, I would never convert SR or dither before sending.

Neurojazz

48k 32bit for the pre-master (not a quality thing, it's a plugin thing) / then dithered down too 44k 16bit master

studiorat

Usually 24 bit, and whatever sampling frequency I am using. Usually 88.2kHz or 96 kHz.

PaulBrewer

Neurojazz wrote: »

48k 32bit for the pre-master (not a quality thing, it's a plugin thing) / then dithered down too 44k 16bit master

32bit? What does that? Actually makes a 32bit file or processed at 32bits?

ogy

cubase doest 32bit float. obviously what your recording is still 20-bit at best, but for in the box sound sources and effects it gives the plugins lots more decimal places to work with before truncation

Neurojazz

Yep, like the hardware UAD reverbs processing at 32bit... helps create a smoother file to dither with - also the CPU overhead is reduced... pity the powercore is 24bit (i use the md3) - so work around this where possible until i have too use it.

PaulBrewer

But what does one use to Play Back a 32bit file? I don't know of anything.

johnnylakes

Had our first album recorded at 16 bit (as we didn't know any better at the time! / and didn't realise we did !)...our mastering engineer told us that 24 bit was best. All part of the huge learning curve. The difference is (obviously) quite noticable!!
Rookies......

madtheory

PaulBrewer wrote: »

But what does one use to Play Back a 32bit file? I don't know of anything.

Pretty much anything, except Pro Tools! In reality, 32 bit simply allows you to lower the peak level afterwards, if the client has overloaded the master outs. As a delivery format it's a bit pointless.

Even if you track at 16 bit, it still sounds a whole lot better to mixdown to 24 bit.

As has already been said, don't change the sample rate of the mixdown. Leave it at whatever you tracked at.

PaulBrewer

madtheory wrote: »

Pretty much anything, except Pro Tools! In reality, 32 bit simply allows you to lower the peak level afterwards, if the client has overloaded the master outs. As a delivery format it's a bit pointless.

Even if you track at 16 bit, it still sounds a whole lot better to mixdown to 24 bit.

As has already been said, don't change the sample rate of the mixdown. Leave it at whatever you tracked at.

But hardware converters are generally only 24 bit, No? So 8 bits get truncated?

DamagedTrax

PaulBrewer wrote: »

But what does one use to Play Back a 32bit file? I don't know of anything.

as far as im aware there is no 32 bit file. its just the resolution the engine works at. the rendered file is usually 24bit (correct me if im wrong)

as far as 16 vs 24bit. it only matters if you have highly dynamic material. i usually record my kicks and bass to 16bit as they're levelled on the way in anyway so the dynamic range isnt to important anyway.

but yeah, final mix always at 24bit.

woodsdenis

http://duc.digidesign.com/showthread.php?t=180818

I have never heard of a 32 bit file used for delivery for a Master or anything else.
I think what people are getting confused about is the internal summing and processing of the mixer in each DAW. In Protools the internal mixer uses 48 bit summing ,if you really want to know what all of that means follow the link above. The primary difference between 24 and 16 bit is the amount of headroom available. It is virtually impossible to achieve 16 bit resolution
when recording/tracking at 16 bit unless you are brickwall limiting everything. Levels become an important factor.

However when recording /tracking at 24 bit it is easily possible to obtain
16 to 20 bit resolution without recording everything as close to max as possible. It has become accepted in many circles, that infact when recording at 24 bit, levels should on the conservative side. The reasoning is that when
summing 50 to 100 tracks through a DAW mixer it sounds better/cleaner, even if according to Digidesign it is impossible to clip its internal mixer.

http://recforums.prosoundweb.com/index.php/t/15038/0/
This above link is a good well informed discussion about this very subject.

Denis:p

PaulBrewer

woodsdenis wrote: »

I have never heard of a 32 bit file used for delivery for a Master anything else.
I think what people are getting confused about is the internal summing and processing of the mixer in each DAW.

That's what I was thinking .... but was much to polite to say so

DamagedTrax

DamagedTrax wrote: »

as far as im aware there is no 32 bit file. its just the resolution the engine works at.

wow all this books smarts isnt going to waste after all :pac:

ogy

as far as im aware there is no 32 bit file. its just the resolution the engine works at. the rendered file is usually 24bit (correct me if im wrong)

cubase lets you mixdown to 32bit float file resolution, would imagine its just a waste of megabytes but it is an option

Neurojazz

it's for the CPUs' (using 24bit makes for extra cpu cycles to convert that 24bit word into a 32bit word for the cpu to operate on) and some plugins (UAD being the main ones... not sure about waves etc...) and the peaks mentioned earlier... even UADs own guru engineer takes 32bit files for preference... (and he knows his sh1t) if i could be arsed i'd quote from their videos i would, but way too busy at the moment...., i can understand that 24bit sounds fine, it's just optimal for the box with certain plugin chains.

PaulBrewer

So to conclude -

There are only 24 bit files processed by a 32 bit process.

The delivery file remains as a 24 bit?

madtheory

Ya, 32 bit as a delivery format is pointless. You're not going to hear the difference between 32 and 24, even without dither, unless you're overloading the main outs. That can be corrected in a 32 bit FP file because the headroom is huge, but not in 24 bit file.

The output of PT HD is 48 bit fixed, and when you bounce internally, it gets dithered to 24 bit.

PT LE has a 32 bit float output, again, dithered to 24 bit when you bounce internally.

Some plugins are double precision, so in PT HD they'd be 96 bit and would dither internally to supply 48 bit to the PT bus.

If you're mixing on an analogue desk or using analogue summing, then the converter is 24 bit, so making a 32 bit FP file from that is pointless. Very few converters give you the full 24 bits anyway.

Similarly, tracking (from the AD) in 32 bit is a waste of space.

Bear in mind that the A to D does a whole lot of DSP onboard- the sampling rate at the input is a large multiple of the output, so that the filter can stay out of the audible range. And because of this sample rate conversion, there's dither too. But that's all too much engineering for us, so they just give us a 24 bit stream at a standard sampling rate.

Neurojazz

yep, same here - got sidetracked, all my renders from wavelab end up 24bit then dithered down to 16bit - just wanted to highlight the 32 chaining issue for those ITB.

johnnylakes

Well there's a coincidence! Was just debating the 'correct' approach (with myself!) last night..
Also curious as to how much headroom to leave...using Cubase
Would I be right in thinking keep the individual levels of the tracks down as opposed to keep them 'up' and dramatically lowering master level??
Does that make sense?
...the question I mean.......
my head hurts

Neurojazz

set the project to 32bit float and you don't need to really worry out levels - then just bring the master buss down towards the end of the mix and then start to insert compressors/limiters then - don't use FX on the master buss to reduce the volume.