Help with "Upper Frequency Mixdowns"...

[Riverdog]

I just obtained the ability to record 48 to 96k audio in Logic Audio 5 on OSX...
(Obtained new Audio Device)

- If I record my tracks at 48k - I then will mix down to 48k in Logic...

My question is:

- When I go to CD how do I change my final tracks to 44.1?
- Is it necessary?
- Does it happen automatically using Toast 5 Titanium?

Just need to know if there's some "magic" I'm missing in here - And thanks in advance...

 

[littledog]

Yes, it is necessary to do a sample rate conversion from 48 to 44.1 before burning your CD. If you burn 48k files onto a CD (assuming your software would let you) your music will be slower and transposed down about a step and a half.

I don't know anything about Logic, but many software programs let you do a sample rate conversion at the time when you make a stereo file (or bounce). Often they will even give you a range of choices ranging from lowest quality/highest speed to highest quality/lowest speed.

 

[Riverdog]

AH HA!!!

So THAT's why my music sounded so funny!!!
I thought there was something wrong with the program - But I figured what you stated was the case...

When I switched my input to 48k I sound like me on Helium... ;o )

So, essentially when you record in that frequency it will add more "samples" which speeds your interface up... hmmm.... Interesting...

My Final question will then be:

Is there any reason to go to 48k or 88 or 96k?
I only hear a difference when I'm recording in 16 or 24 bit... The 48k or above doesn't seem to do much in the way of fiedelity or clarity for me... (I know there are many out there who say they can hear the difference)...

But, it seems that my gear gives me more "headroom" in 24bit mode at any frequency... My mics can handle a little hotter signal and they can capture transients more efficiently...

What do you think...?
And thanks for your reply!
I'm sure I will need to experiment a bit to get the best performance but understanding some of the "causailty" helps a bit...

 

[TexRoadkill]

The reason for higher sample rates is that the highest frequency you can record is half of the sample rate. 44.1 = 22.5khz compared to 96 = 48khz. The advantages depend on your entire signal path and if you have the hardware power to process those larger samples. Whether you can even hear the difference is a matter of much debate and obviously you have to convert to 44.1 for the final CD master anyway so that brings up the issue of the sample rate conversion. Some say that any sample rate conversions leave artifacts and nullify the advantage of recording at the higher rate in the first place.

If your entire project is staying in the digital realm it is probably wise to just stay at 44.1 for the entire process. If you are recording digitally but doing an analog mixdown it may be better to track at a higher sample rate but it depends on the bandwidth of your analog gear.

 

[Track Rat]

There's a difference but in my opinion it's a point of deminishing return. I track at 24/44.1 and feel there's more advantage in 24 bits than the higher sampling rate.

 

[volltreffer]

It might be an idea to record in 88.2kHz...

The conversion is much easier from there and you have the advantage of EQs and everything sounding much nicer (if you have the computing power!). The thing is that EVERY EQ algorithm has to deal with the upper end of the audible frequencies. There you have differences between the anlogue and digital EQing (cause every frequency processed as higher than your half sampling rate would appear at the lower end of the freq spectrum...)

Let us assume that good EQ algorithms have noticeable differences to 'real' EQs just one octave down... Then you'd have an influence down to 11kHz for the 44.1kHz recording and to 22k for the 88.2kHz recording...

That'S why there are double sampling effects especially EQs



aXel

 

[scottboyher]

I still don't understand this shit.. Why would you record at a higher bit rate just to bring it back down later?

I wouldn't record on a CD so I could record it to tape... Or would I?

 

[Track Rat]

It's a very good idea to record at the highest resolution as possible. 30 tracks of 24 bit resolution when mixed down and dithered to 16 bits will sound better. And any processing you do to individual tracks (compression, EQ, reverb, etc) will also sound better at the higher resolution.

 

[Riverdog]

mmm...

I see I've sparked "the massive debate" again...

- The idea is that even though you dither your sampling rate back to 44.1k, The signal input at the front end has recorded a "cleaner" / "Higher fidelity" Track at a higher sample rate...

It's kind of like putting a 60 watt amp with a 60 watt speaker...

If you crank that amp so it puts out the max that the speaker can handle, the amp is straining itself to get to the output which creates a "noisier" sound and may damage your speakers...

Put a 120 WATT amp with a 60 watt speaker, however, and you can move the amp to the maximum output of the speakers handling capacity and have a better "cleaner" sound...

Same Idea with sampling rates...
Higher rate / even though dithered back = better fidelity 44.1k mix...

??? Make Sense ???

 

[scottboyher]

Re: mmm...

??? Make Sense ???

yessum...

 

[littledog]

i don't think the above amp analogy is relevent, even though the final conclusions are correct.

in digital audio any time you manipulate data you are performing mathematical calculations. If you wait until the "final" result to round off the answer, it will be more accurate than if you round off at every stage of the process.

Example:

2.4 x 6.6 = 15.84

now, let's say your final answer must be in whole numbers. If you round off AFTER doing your calsuclation you get:

2.4 x 6.6 = 15.84 which rounds to 16.

but if you round off BEFORE the final calculation you get:

(2.4 rounds to) 2 x (6.6 rounds to) 7 = 14.

In the first case the answer is innacurate by 0.16, but in the second case the answer is inaccurate by 1.84!

That's why you maintain a high resolution (24 bits) until the very end and only then "round off" to 16 bits. Dithering is introduced to insure that the extra bits are rounded as accurately as possible, rather than just truncated off the end.

What is the practical result of recording 24 bits and dithering to 16, as opposed to staying at 16 the whole time?

•smoother fades
•less gritty reverb tails
•more precise stereo imaging
•better signal to noise ratio
•more pleasing sound

 

[scottboyher]

i don't think the above amp analogy is relevent, even though the final conclusions are correct.

in digital audio any time you manipulate data you are performing mathematical calculations. If you wait until the "final" result to round off the answer, it will be more accurate than if you round off at every stage of the process.

Example:

2.4 x 6.6 = 15.84

now, let's say your final answer must be in whole numbers. If you round off AFTER doing your calsuclation you get:

2.4 x 6.6 = 15.84 which rounds to 16.

but if you round off BEFORE the final calculation you get:

(2.4 rounds to) 2 x (6.6 rounds to) 7 = 14.

In the first case the answer is innacurate by 0.16, but in the second case the answer is inaccurate by 1.84!

That's why you maintain a high resolution (24 bits) until the very end and only then "round off" to 16 bits. Dithering is introduced to insure that the extra bits are rounded as accurately as possible, rather than just truncated off the end.

What is the practical result of recording 24 bits and dithering to 16, as opposed to staying at 16 the whole time?

•smoother fades
•less gritty reverb tails
•more precise stereo imaging
•better signal to noise ratio
•more pleasing sound

Kick ass that totally made sense.. thanks!

 

[Riverdog]

Understood...

Hmmm...
I knew I was "doin the right thang!"...

Just didn't know why!

I've seen this math before and it does make sense...

Good refresher course...

Thanks,,,

And thanks ALL for your comments...

Somehow I figured buying this little interface would help somehow!!!

Cheers..

(PS - I'll remember the part about waiting till the very end... The way I have things configured this shouldn't be an issue...)

 

[TexRoadkill]

Bitrate and sample rate may be getting a blurred here. Just to clarify - Bitrate, measured in bits, determines the dynamic range and Sample Rate, measured in KHZ, determines the highest frequencies recorded.