16/44100 vs 24/48000?

  I've been recording the last few months using 24/48000 but I've noticed several people who have told me that they record using 16/44100hz and their mixes sound great to me. I know CD's are 16/44100hz  so in the end it's going to end up that way if you want your music on CD.

 This is probably a question I can answer for myself by just trying to record that way to start with, but I wanted to ask because I don't wanna put a lot of time into something only to find out that I could have done better.

I was exporting some files today at 16/441000 where I had previously exported them 24/48000, Ofcourse there was a huge difference in the size of the files. But honestly listening to them by themselves I could tell a difference. This makes me think that resource uses recording would be a lot better served recording 16/44100, not that it's a problem right now but it's a thought either way.

What do you record at? I'd like to hear your opinions on this. I use Music Creator 4 but I also have Sony's Acid Music Studio 7 but it only records at 16/44100.

How big of a difference is it? I wonder if recording at a 16 instead of 24 might be easier to mix? Or that is a bit more forgiving on picking up sounds.

I'm just looking for thoughts and advice on this, I haven't recorded long enough to know better. I've just always figured 24/48000 was better to record at, or I have been told that.

What do you think?

I wrote a tutorial about this about a year ago.  I can't seem to find it right now, but I'm sure when I get home I can find it.

basically I think it's best to record in 24 bit resolution before dithering down to 16 for export.  you have more resolution to record at lower levels giving you more headroom for mixing.

With 16 bit the theoretical noise floor is at - 96 db. Assuming we don't get the last 6 db then we have a noise floor of -90 db in real terms, which is not bad anyway.

24 Bit recording puts the noise floor down at -144 db! Now even if we loose 6db down there as well it means the noise floor is now down at -138 db ! So with a 24 bit recording you can record as much as 48 db lower level and still maintain the same Signal to Noise ratio as 16 Bit!

It just means you can safely make -20 db FS your reference level during recording and mixing for that matter and still be sure you are well clear of 0db FS but the noise is also still way down from -20db (a whopping -118db below)

Its good to master at this bit depth and then dither down to 16 bit at the last minute.

The difference between 44.1 Khz and 48 Khz are minimal but 48Khz does have certain advantages. If you are going to record and mix in a higher resolution than CD, then you might as well use the higher sampling rate as well. 48 Khz is closer to 50Khz which people like Bob Katz say is closer the ideal sampling rate. (You do NOT need a higher sampling rate though!) Also if your music ends up on any vision related DVD's etc you are already in the correct sampling rate for that.

It largely depends on what you are going to do. If you are recording a voice over that is going to end up as an mp 3 file then the lower resolution will suffice as it would in a loud rock track.

But if its a beautiful detailed and dynamic mix that you are after at the end of the day, even if the final result is 16 Bit 44.1K then there is no comparison recording at 24 bit. Especially if you are using quite a bit of processing and also putting Sonar into 64 bit mode on the mix buss and this where a lot of the action happens. I can hear it bigtime! By keeping so well clear of the dreaded 0dbFS then that is worth it alone. Many problems occur when the signal starts approaching 0db FS. This is where many converters do not perform well. 24 Bit recording and mixing allows you to keep well clear of that limit.

One of the big things I really got from Bob's book is 24 bit recording and how great it is. And as he says if you can do it then do it!

24/44.1

I keep everything at this until I'm ready to burn a CD, then I quantise down to 16 bit.

Any internal bounces within SONAR are done at 32 bit recolution (with the 64 bit engine engaged)

I'm using 24/96. Is this useless and inefficient?

Murray, what you say is definitely correct. I was referring to the noise floor of the digital system and any Signal to Noise ratios that are different to that and come before will set the overall Signal to Noise ratio. (The analog noise sounds nicer if you can hear it)

With digital you need to think about what the reference level is for your recordings on tracks and busses and with 24 Bit you can think lower with that level with not too much fear that the digital system itself is going to become audible.

And Midiminds puts it well too. We can be concerned for frequencies that are near the Nyquist frequency and talk about sample rate, but I am also interested in what happens when signals get too close to 0 db FS and the attack part of a sound and how that may be effected. It is important to keep it accurate. This might mean close to 90 db of dynamic range being needed to capture sounds really well. I just like the sound of digital more when a higher bit depth is used and more of the dynamic range of the digital medium is being used.  eg a sound with a very fast and short attack and complex envelope shape too. This sound ends up being a certain level in the final mix, but there can be two ways it can be recorded. One at higher average level with the attack portion going very close (or over) to 0 db FS / 16 Bit or much lower in level on the track but in 24 bit instead.

I may be referring to percussion instruments for example as well when I talk about attack transients, and things like ride cymbals are good contenders. This is what I like about digital and to my ear it preforms better than analog. Its ability to capture and replay very fast attack portions of envelopes. Analog as we know can soften these edges and distort this part of the sound.

To capture the detail of the mix means you have to think about how to capture these parts of any sound well. But as to sampling rates higher than 48 Khz well there is plenty of discussion on that but for me if we use the resolution in the digital system for attack transients then I think a big part of the sound has been accurately recorded and that is great.