should I use 16 bit or 32 bit float?
This has been bugging me for a while, and I know you guys will be able to answer and explain…
Should I be using 16 bit or 32 bit float? I record outboard with an MR8 (16 bit), but someone over at HR.com said I should have my settings in N @ 32 bit float?
Is this correct?
Where do I change it?
What will this do for me?
Edit: Deleted, 'cos I’m not sure of my answer.
cellardwellar, it depends on your PC soundcard… - I am not sure it is capable of handling 32 bit float. It isn’t necessary if you don’t do a lot of processing.
I myself record, edit and mix in 16 bit, 44.1 kHz, and in this case I can’t tell the difference. If I have to do a lot of external post-processing e.g. in Cool Edit 2000 or Audacity, noise reduction, for instance, I convert the wave file in question to 32 bit before the transformations, because after just two or three digital transformations/effects/filterings it gets really audible. I then convert the file back to 16 bit after I’m done processing.
n-Track processes all audio as 32 bit float internally. Don’t worry about it any more. Spend your time making music instead…
I’m not sure what you are saying exactly, as I am not familiar w/ your equipment. But if your original recording (either recorded earlier or sent live as digital data) is 16 bit, then mixing down with anything higher will just add extra file size for no reason. Where “bits” come from is how information is stored. You know bits vs. bytes, right? Well, in 8 bit, 8 bits make up one what I am going to call “byte” (or, more correctly, “word”) of information. (Ok, I am not actuall talking about actual “bytes” here, which are something different. I am talking about words of information. However, that is irrelevent. I have defined the term as I am using it, so hopefully it will make sence.) In 16 bit, there are 16 bits in every “byte”. the larger the number of bits in a byte, the more detailed the information, as there is “more bits in every byte” (not to be confused with the jingles you might have heard in any cereal comercials). But if the original information is recorded in 16 bit, then the information can’t be added in later (this is one of the limitations of digital information as opposed to analoge). Programs simply fill in the extra bits to make it a higher bitrare. That way the number of bits in that file will be compatable with other files at the higher bit rate that you may be mixing together. But if all of your files are at the higher bit rate, then it does you no good.
I will illustrate. Now, be warned. these are not really binary.
1011 0110 1010 0101
1011 0110 1010 0101 1100 0011
a 16 bit file saved at a 24 bit file:
0000 0000 1011 0110 1010 0101
Make sence? Or think of it as a picture file. If it is saved at a really low resolution, the picture will become pixelated. If you later try to resave it as a higher resolution, it won’t help, because its source file is the low resolution file. Now 32 bit float is slightly different, I think, as there is the “float” that comes in. So maybe this is where that advice comes in?? But from what I understood, the “float” had to do with where the divisions of the individual words (or “bytes”, as I was calling them) were divided. But I could very easily be wrong.
|Quote (Nils K @ Sep. 22 2004,18:02)|
|cellardwellar, it depends on your PC soundcard... - I am not sure it is capable of handling 32 bit float. It isn't necessary if you don't do a lot of processing.|
I myself record, edit and mix in 16 bit, 44.1 kHz, and in this case I can't tell the difference. If I have to do a lot of external post-processing e.g. in Cool Edit 2000 or Audacity, noise reduction, for instance, I convert the wave file in question to 32 bit before the transformations, because after just two or three digital transformations/effects/filterings it gets really audible. I then convert the file back to 16 bit after I'm done processing.
n-Track processes all audio as 32 bit float internally. Don't worry about it any more. Spend your time making music instead...
that is a very good point. adding the extra bits would added a "buffer" of extra bits. Since you will lose a few bits here and there in processing, then the buffer would prevent data loss. hmmm.....
I always use 32 bit for any mixdowns and submixes that I’m going to be re-importing into a song or dowing further processing on (eg. mastering).
I then mixdown and dither to 16 bit as a final step before burning to CD.
Keeping things at 32 bit as long as possible just gives you a higher resolution for any maths that has to be done in processing. This results in more accurate calculations and sound.
As a (very) simple analogy using decimal points instead of bits…
10 / 3 X 1000
If you only have accuracy to 1 decimal place you get
10/3 = 3.3
3.3 X 1000 = 3300
If you have 3 decimal places you get
10/3 = 3.333
3.333 X 1000 = 3333
Every time you process something at 16 bits compared to 32 bits you are losing bits of information as it has to round off a lot more.
Do I process and you might not notice much but do a number of things to a wav file and you probably will start hearing a difference.
Hard disks are so big and cheap these days the extra space required for 32 bits is not a big deal
Yeah. That is a lot clearer than I way saying. And more accurate as well.
I’m not sure what the question is. What bit depth you use depends on what you’re doing. Regardless of what you choose, n-Track uses 32-bit float internally. I think you got advice from someone who didn’t know n-Track very well.
For listening, you set N to use the playback mode that’s the best mode on your soundcard. (You do this by clicking on the hammer icon under the meters.) 32-bit float is not one of the options. It’s either 24-bit (in various formats; generally only one works with your soundcard) or 16-bit. Use 24 if you can, and 16 if you can’t. This only affects listening while working, and has nothing to do with the format of the wave files you’ve imported.
The same would apply if you were recording in n-Track: use the highest setting your soundcard allows.
For both of the above, you need to have paid for the 24-bit version of N to use 24 bits. If you haven’t paid, just use 16 bits and don’t worry too much. (We won’t get into the 16 bit vs. 24-bit discussion here – 24 bits is better, but use what ya got and upgrade if & when you deem appropriate!)
The other place where bit-depth gets selected is for doing a mixdown (the “More …” button). It’s not a “setting” that sticks, you set it each time you do a mixdown. (It defaults to “your soundcard’s settings”, which is usually the wrong thing for me.)
When doing a mixdown, the bit depth you want depends on what you’re going to DO with the mixdown. If you’re going to burn it straight to CD, you have to use 16 bits – because that’s what they are.
But if you’re going to do any further processing of the resulting mix (i.e., mastering), whether in n-Track or in another program, mix down to 32 bits to avoid introducing more quantization noise.
If you’re working with a mastering house and they ask for 16 bit files, ask if they can handle 32-bit float files instead. Unless they’re idiots, they’ll prefer the 32-bit float files – they make it much easier for them to get good results. If they show the slightest resistance to 32-bits, find another mastering house!
You don’t need the 24-bit version of n-Track to mixdown to 24 or 32 bits.
I’m basically saying the same thing RichLum said, only using more words.
From the V4.0 change log page…
Build 1659 - Monday, August 9, 2004
Playback of 64 bit (double precision floating point) wave files.
Does this mean V4.0 of n-Track now does internal processing at 64 bits?? Just wondering…