Master buss order.

  • Thread starter Thread starter PDP
  • Start date Start date
Audio Mastering said:
Not correct, what if you bounce it in the same bit depth for mastering? You do not dither when bouncing down and keeping the same bit depths

You dither only when going down in bit depths.

LOl, see its very confusing
Click to expand...

haha AYYYY 2/2! I get the gist, I just need to read up on it. Thanks brother!
 
No problem and I completely understand. Im almost 42 years old, so ive had allot of time to wrap my head around this stuff.
 
Your welcome. Here is some info on floating point bit depths. This is kinda different form the internal 64bit engines that some plugins have, but its kinda the same outcome, if thats the right word for it..

32 bit or 64bit floating point processing means that after you record your audio at 24bit, you can convert it to a 32bit or 64bit floating point. This means that it will add the extra bits after it is recorded. These extra bits that get added onto the file after its recorded, will give you more headroom for processing (audio calculations).. I always convert all my files to 32bit floating point.

Having 32 bits, rather than 24bits is going to render a more accurate result. Some recording platforms do not have this option yet and its a shame :(
 
Cool,

My M-Audio 2496 has the 32bit float option, which I've used but switched back to 24bit because I wasnt really sure what the benefit was. So If I select to record 32(floating) 88, there is headroom being added? I can also export my mix with that too. Would you use it to record or only convert after, or is that kinda the same thing since the bits are being added anyway.
 
you won't record in 32bit floating point or any other floating point. You can convert your audio file to a floating point after its recorded. The highest you can record with your sound card or any other sound card is 24bits

Your card can only record in 24bit, not 32bit floating point. It can process it though
 
Racherik said:
I thought dithering was done to smooth jitter when converting from different sampling rates rather that sample sizes. i.e. Going from 96khz to 44.1khz. Am I mistaken?
Click to expand...
Yes

If so whats that called.
Click to expand...
Sample Rate Conversion

And in the real world is "truncation noise" an issue?
Click to expand...
Yes. When it comes to the best sound quality possible, everything matters.
 
Waltz, Did you ever trademark the word mixtering :)
 
How? I record an audio track at 24/88 then what?
Click to expand...
If your DAW supports it, you can bounce it to a 32bit floating point audio file.
Thats the way I do it in Sonar. Each DAW will have different rendering (bouncing) options.
Floating point is different than fixed point. You cant record to a floating point.
Cj
 
Audio Mastering said:
If your DAW supports it, you can bounce it to a 32bit floating point audio file.
Cj
Click to expand...

Yes, When I export I can select 32bit float as an option. I can also select Wave 64 file, that doesnt come out as playable file, gotta look in the manual on that one.

So you export every 24bit track and re-import it as a 32bit float file/track? Can you hear the difference?
 
So you export every 24bit track and re-import it as a 32bit float file/track? Can you hear the difference?
Click to expand...
No, no need to export. Thats like walking 2 miles to get across the street.
You can just boucne it right in your DAW program (if your DAW allows it) Most should!

Can your hear the difference?
Click to expand...
Its more about having more headroom for processing. Having 32 bits, rather than 24bits is going to render a more accurate result. Its all about the math and calculations.
 
Racherik said:
Glen,

I've slept a few night since I read Bob Katz Master book, but I thought dithering was done to smooth jitter when converting from different sampling rates rather that sample sizes. i.e. Going from 96khz to 44.1khz. Am I mistaken? If so whats that called. And in the real world is "truncation noise" an issue?

Thanks
Click to expand...
In my book, frankly, 98% of the time, none of what we're talking about is an issue, as 98% of the time it's all inaudible.

One has to take a lot of Bob Katz's stuff with a grain of salt. He's a very smart guy, but he tends to get purposely lost in the intricate detailed trees of the math and can't see the real woods of the situation. There's a whole lot of his stuff that while technically true, is largely irrelevant in the real world, IMHO.

As far as using dither to attack jitter from sample rate conversion, I just don't see how, myself. Maybe dither noise (for all dithering really is is is a form of noise) may help confuse the ears or mask other extreme low noise low imperfections, but it's not really designed to do anything other than to introduce noise at the last bit of a 16-bit word in order to average out just how inaccurate it winds up being.

Think of it like this. Take the value of pi, which we never can get quite exactly, because the decimal places just keep going. But let's say we "sample" it's value at 3.14159. But the spreadsheet cell we're sticking it in only has room for four decimal places, 3.1415. We might "dither" the last digit to say 3.1416 because it's actually closer to the "real" value for pi than 3.1415 actually is, even though the 5 is accurate for that decimal place.

Well, while not really exactly, that's kind of like what dithering is doing to the 16th bit of the binary sample value after chopping off the rest. But instead of rounding the value up or down specifically, dither is choosing a 1 or a 0 for that last bit based upon it's own pattern of selecting 1s and 0s that's not random, but is random in relation to the actual signal value. The argument is that this tends to "average out" the inaccuracy of that last bit in a way that is less audible or more pleasant to the human ear. The fact that this is happening at -96 dBFS on signals that rarely have any useful information at all below -75dbFS notwithstanding.

What that has to do with jitter, I don't know. Maybe it might help in smoothing out the last bit in the calculated values at the new sample rate. But I'd say that the best way to handle that is not to deal with sample rate conversion at all. Again Katz will probably say something different, and this is one of those never-ending forum arguments, but unless one is forced to convert from 48k audio for video, nobody here IMHO should ever be working in anything other than 44.1k.

Let the flames begin :).

G.
 
SouthSIDE Glen said:
nobody here IMHO should ever be working in anything other than 44.1k.
G.
Click to expand...

Thanks for the detailed info. Even though I dont think I can honestly hear much above 12k, 88 seems to have better highs, it could be my imagination, technically 44 is going up to 22k from what I understand (which is plenty for human hearing) The only other benefit is latency, the higher rate does lower monitor latency.

It'll be interesting to see if someone finds a benefit for 192.
 
It'll be interesting to see if someone finds a benefit for 192.
Click to expand...
I think its overkill and a waste of disk space and resources for the average home recordset.
If you dont have an expensive high-end signal chain, you wont hear the difference or benefit form it. 44.1Khz or 48Khz is good enough for the home recordset. If your playing acoustic or classical, Then 88 Khz or 96kHz is good, but 192kHz is pretty much overkill for the average home record set
Cj
 
PDP said:
Thanks for the detailed info. Even though I dont think I can honestly hear much above 12k, 88 seems to have better highs
Click to expand...
Perhaps through your converters that may be true, which is most likely an indication that your converters do not work as well and they could at 44.1, not that the sample rate itself is better.

And, honestly, if you hear a difference anywhere near 12k between the two, either you can indeed hear way above 12k, or your hearing above 12k is indeed shot and your converters (in one direction or the other, IDK) are pretty severely malfunctioning or very poorly designed at the 44.1k setting, because there should most definitely be no difference whatsoever as far down as 12k.

Any difference in quality between the sample rates is not because the sample rates themselves matter, they don't. A theoretically perfectly good converter will sound exactly the same at every sample rate from 44.1 up to 192 and beyond. For humans, that is. Dogs and cats may disagree. And even then, they should only start barking or meowing about the difference on recordings make from an analog input chain capable of grabbing stuff accurately above 20k. Not a whole lot of mics and pres that are made to do that.

G.
 
SouthSIDE Glen said:
Let the flames begin :).
One has to take a lot of Bob Katz's stuff with a grain of salt. He's a very smart guy, but he tends to get purposely lost
G.
Click to expand...
One has to be careful digesting this type of salt, it can cause heart attacks. Projecting can be hazardous to your health :spank:
 
You must log in or register to reply here.
Back
Top