Is rendering limited by sound card bit depth?

[jtxx000]

If I have my sound card set to 16 bit and render to 24 bits, will it work?

[jones-y]

If I have my sound card set to 16 bit and render to 24 bits, will it work?

Yes.

[jtxx000]

so you will still get full range of all 24 bits?

[rpc9943]

lol

do you think it uses the sound card to render it or do you think the computer renders the numbers all together? think about it

RonC

[mauseoleum]

Lol, you've got it all WRONG ! Computer renders, and soundcard watches and listens

[nuisance sonore]

Internal processing is done with 32 bits, so you can render at anything you want. But if you started with 16 bits recordings, you will not gain anything from it except a bigger file size.
You might have mixed up recording and rendering; rendering doesn't involve the soundcard AFAIK. And for those who seem to find that funny, I don't find it funny at all as a lot of this stuff can get confusing at time, especially for musicians who are not necessarily engineers.


[edit]you might be interested in this:http://www.tweakheadz.com/16_vs_24_bit_audio.htm

[Mr. Tingle]

That is a question that had been on my mind. And now I know the answer! (though I'd like to hear Jules or Valley's opinion as well, 'cause there are bound to be other factors)

[Ixox]

Internal processing is done with 32 bits, so you can render at anything you want. But if you started with 16 bits recordings, you will not gain anything from it except a bigger file size.

Adding 2 "16 bits" tracks (may) need 17 bits not to loose any informations.
Adding 4 "16 bits" tracks (may) need 18 bits not to loose any informations.
Adding 8 "16 bits" tracks (may) need 19 bits not to loose any informations.

Moreover, a 16 bits audio track going through a 32 bits audio effects (all VST effects are 32 bits if i'm not wrong) go out from this effect with 32 significant bits....

So rendering 24 or 32 bits can only be good before entering the final (pre)mastering stage....

[nuisance sonore]

Internal processing is done with 32 bits, so you can render at anything you want. But if you started with 16 bits recordings, you will not gain anything from it except a bigger file size.

Adding 2 "16 bits" tracks (may) need 17 bits not to loose any informations.
Adding 4 "16 bits" tracks (may) need 18 bits not to loose any informations.
Adding 8 "16 bits" tracks (may) need 19 bits not to loose any informations.

Moreover, a 16 bits audio track going through a 32 bits audio effects (all VST effects are 32 bits if i'm not wrong) go out from this effect with 32 significant bits....

So rendering 24 or 32 bits can only be good before entering the final (pre)mastering stage....

Care to expand a little bit? I always figured that the bit depth only changed the decimal place, which going from 16 to 24 bits only added... zeroes, and adding zeroes can only produce more zeroes. Of course, any more light thrown on this obscur subject can only be good.


[edit]Ah, I see now that some math calculations will produce result that will get trunkated in 16 bits (well, in 24 bits also, but with more precision). So yes it's probably good to mix down to 24 bits if more effect are to be applied to the result later on.

[Ixox]

Care to expand a little bit? I always figured that the bit depth only changed the decimal place, which going from 16 to 24 bits only added... zeroes, and adding zeroes can only produce more zeroes.

That's true for a lonely track without any effect....

Now add two 16 bits audio tracks wich contains at one moment :
1001 0110 0000 1111
+ 1100 0000 1001 0000
=1 0101 0110 1001 1111 <= 17 bits....
If you render to 16 bits you'll lose the last zero... That's the idea why you need more bits while mixing even if all the tracks are 16 bits audio.

There is also the 32 bits VST stuff....
Your 16 bits audio track will go through a compressor, an EQ and a reverb...
Each of these 3 plugs will make calcul in 32 bits.... Even if the entering track is 16 bits... Everything is done in 32 bits floating point (even 64 bits with some VST). So the 32 bits of the outgoing track will be all used !
A simple stupid exemple :
A VST that divides by 3 what it receives.
1000/3=333.333333333
333.3333333 needs many bits not to be trunced.... much more than 1000.
So if you want to keep all the quality of the VST you need more than 16 bits.

Go down to 16 bits the later you can...
Usually in the final stage processing (premastering) just after adding some "dithering" noise... That should smooth the 32 bits=> 16bits conversion.

(I saw your edit after writing that.... )

[jtxx000]

lol

do you think it uses the sound card to render it or do you think the computer renders the numbers all together? think about it

RonC

i thought that the computer did it but tracktion says rendering for <insert your sound card name here> and then i thought i read something around here about tracktion working at the same bit depth as your soundcard. what bit rate do frozen tracks get rendered at?

[valley]

I think that's right, jtx. If memory serves, when you don't get prompted for a bit depth and sample rate Tracktion renders at the level defined by the ASIO drivers[1]. Otherwise you render at whatever you select.

[1] For sample rate this is intelligent behaviour, but I'd prefer if rendering defaulted to the Tracktion internal bit depth (IE 32bit).

[tibenmik]

I'm digging up this thread as I was searching for info on default bit depths on rendering. Put another way, I was surprised that when selecting the 'render track' option (the button next to 'freeze track', not the 'export' menu option) for a track containing audio, that the resultant render is in 24bit, even if the source audio file is 32. Is there any way to adjust this behaviour, other than doing an export instead (in which case you *can* select bit depth - perhaps I've just answered my own question...)?

TIA,

_mt

[IIRs]

No, rendering will be 24-bit at best. I agree its a little odd..

[aMUSEd]

I get confused about this too. My Creamware system has 16, 24 and 32 bit drivers (all in a choice of ASIO and ASIO2) but as I can still record from eXT in anything up to 32bit and can't see or hear any major differences in CPU hit or quality from each driver then what exactly is the difference practically between them? I can only see and hear a difference if I go to a higher samplerate (esp 96) which makes sense but what is the point of the different drivers if they don't do anything different?