exporting 44.1 mixes when recordings were at double or quad rates

I notice that the Freq Analysts show frequencies up to 48 kHz when my tracks recorded at 96 kHz.
 
Should I low pass filter the final mix before exporting to 44.1. I think I don't have an EQ that allows me to filter out above 20 kHz although I'm not positive about that.
 
When I export to 44.1 from a 96 kHz project, is there a risk of artifacts from reduction in available frequencies?
 
I feel like it is a silly question, but mixing at the double and quad rates is still sort of new to me. I was mixing on an old laptop before Fall 2013, so I always recorded at 44.1 or 48 due to processing constraints. Now the ceiling has been raised sufficiently that I can record at 96 and mix a many track project bumping against ceilings for Hard Drive read/write or processor consumption.
 
I'm guessing that there is no extra step required, but I thought I'd ask. I know that a lot of convertors apply low pass filters above the audible range when converting.

Yes, I can see the sound up to 48 kHz and yes way outside of the audible range.
 
I think the maximum frequency that can be reproduced is about 1/2 the sample rate minus a little because at the top of the range there needs to be a low-pass filter. This is why they use 44.1 rather than 40 kHz. The extra 2K of frequency inclusion is for the filter. That's why 48 is a little preferred because it puts the filter a littler farther outside of the audible range. The filter usually causes some minor artifacts that hopefully never fall within the audible range. Having more frequency outside of the audible range to work with makes it safer.
 
The bottom line is I'm not sure how it works when you go from 88.2 to 44.1 or 96 to 44.1. Is a filter applied? Should I apply the filter manually? Is there no need for a filter? Does the filter only matter at the time of digitizing?
 

 
http://stabilitynetwork.blob.core.windows.net/g-tunes/Screenshot_96.png
 
From Digital Audio Explained for the Audio Engineer by Nika Aldrich (Sweetwater Press)
P124

Aldrich
Aliasing
If the material to be sampled contains frequency content above the Nyquist Frequency then a type of distortion to the signal called aliasing occurs. If frequencies above the Nyquist Frequency are present then the samples will not be taken twice per highest frequency, but will be 'assumed' to have been taken twice per highest frequency and the signal will be reconstructed differently than it originally existed.
 
Since there is only one possible way to reconstruct a "legal" signal through the sampling points, and since we assume that the waveform is indeed a legal waveform and will be reconstructed as a legal waveform, any frequencies above the Nyquist Frequency will be re-created as frequencies below the Nyquist Frequency. The frequencies created are actually very predictable and mathematically determinable. The Nyquist Frequency acts as a sort of mirror in that any frequency content above it gets mirrored around it and creates frequency content below it.

Sorry for the delay, I saw that post on my cell and wasn't sure if I was reading it correctly. I have no answer to this, and have never seen an audio spectrum analyzer go that high before. The best way to find out is to actually op test it and listen to the results in a good environment.
 
My gut reaction is to low pass that scenario (only because I can see it), but I am also very curious if anything audible comes out in the mix-down from not low-passing it.

Sampling audio at 96khz is not the same as producing an audio wave of a 96khz frequency.

So I emailed Nika Aldrich who wrote the book I quoted, and he answered with this...
 

Nika
The proper way of doing sample rate conversion involves filtering the material below the new Nyquist frequency prior to the sample rate conversion.  This would eliminate any aliasing due to the conversion. Assuming you used a good sample rate converter, there should be no aliasing in the result.

 
So, I guess this is really a question for the bakers... Is material above the Nyquist frequency being removed prior to the sample rate conversion? Is Nika Aldrich correct that this material needs to be filtered before doing the conversion? And if so, I don't have any filters that filter above 20kHz so how would I go about it?
 
Changing the sample rate from 96 to 44.1 changes the Nyquist frequency from 48 to 22.5. Does this sound right to you guys? Am I correctly understanding this?
 
I think this is just handled for us by Cakewalk, but I'm checking to be sure.

I suppose another question is does the sample rate conversion happen through the Interface and interface drivers or does it happen independently within Sonar? If Sonar depends on the interface, we may all get different answers as to whether a filter is applied and how the filter is applied.

The filtering you are asking about is already an intrinsic part of any competent SRC (Sample Rate Conversion) process, and Sonar's is quite competent. So there is no need to do it manually as it is already being done.
 
 
And dither has absolutely nothing to do with sample rate reduction - it is used when reducing bit depth.

mudgel
Whenever you reduce sampling rate you should be dithering

I'm sorry to say I wrote this in error. I typed up some info then copied n pasted the wrong part. Didn't realise till later when other comments made. Don't want to correct the post because others have since commented on it.
Dither when reducing bit depth not sample rate is what I should have said.

drewfx1
The filtering you are asking about is already an intrinsic part of any competent SRC (Sample Rate Conversion) process, and Sonar's is quite competent. So there is no need to do it manually as it is already being done.
 
 
And dither has absolutely nothing to do with sample rate reduction - it is used when reducing bit depth.

Thanks, DrewFX1. This again was my assumption. I was just asking for confirmation - thanks for providing it.
 
Nika Aldrich says dither helps when the division of the sample rate is not even. So for example, that it should be used when going from 96 to 44.1 but it should not be used for 88.2 to 44.1. For this reason, up sampling and down sampling done by a factor of 2 does not require dither.
 
I can find the page number and quote the book if you would like. I know that many of the forum members feel strongly that going from 96 to 44.1 doesn't require dither, but I don't know their source of information on this and I have trouble accepting the assertion without knowing the source and that the source is more respectable than Nika's book published by Sweetwater.
 
 

gswitzNika Aldrich says dither helps when the division of the sample rate is not even. So for example, that it should be used when going from 96 to 44.1 but it should not be used for 88.2 to 44.1. For this reason, up sampling and down sampling done by a factor of 2 does not require dither.
 
I can find the page number and quote the book if you would like. I know that many of the forum members feel strongly that going from 96 to 44.1 doesn't require dither, but I don't know their source of information on this and I have trouble accepting the assertion without knowing the source and that the source is more respectable than Nika's book published by Sweetwater.

 
I would indeed be curious what his argument for dithering is. Perhaps it refers to SRC algorithms that process at a higher bit depth?

Nika to me
The proper way of doing sample rate conversion involves filtering the material below the new Nyquist frequency prior to the sample rate conversion.  This would eliminate any aliasing due to the conversion.
 
Assuming you used a good sample rate converter, there should be no aliasing in the result.

Me to Nika
So let me restate just to be clear...
 
I record a multi-track project at 96 kHz and bounce it in the box down to one stereo pair at 32 bit 96 kHz (the bit depth increases from 24 at time of recording to 32 for bounced tracks).
 
Now, to export, I'm reducing back to 24 bit 44.1. I'm using Sonar Cakewalk to export the audio. I'm guessing that the filter would be applied by Cakewalk and my audio interface would not matter. Do you agree with this?
Also, you are saying that Cakewalk would need to apply the filter to remove material above the new Nyquist frequency prior to exporting at the new sample rate.
 
I'm guessing the responsibility lies with Cakewalk to convert the data to the new sample rate, not my audio interface. Therefore, they should be able to answer to the question of the filter.
 
Does it sound like I correctly understand you?

Nika to me
You have it down precisely.  When Cakewalk does an "export," it is necessarily doing a sample rate conversion.  In order to do that properly, it must filter any data above the new Nyquist frequency prior to eliminating the excess samples.  (The way it actually converts from 96kS/s to 44.1kS/s is likely by upsampling to the lowest common denominator of the two sample rates, then adding the new Nyquist filter, and then removing the excess samples).  That algorithm lies entirely within Cakewalk, and is not dependent upon your converters.

Me to Nika
Given that path, dither would not be required for a sample rate conversion where bit depth remains unchanged. Do you agree?

Nika to me
To the contrary, dither is required.  But I want to make sure I'm clear about what that means.
 
Dither is required any time bit depth is reduced.  When your 24 bit/96k file is upsampled and downsampled to 44.1k, a filter is used.  That filter requires processing.  That processing inherently invokes more bits (Cakewalk likely uses either 32 bit or 64 bit floating point processing to do its internal processing, including sample rate conversion).  The downsampling to 44.1k thus requires a reduction in bit depth from the processing bit depth (32 bit or 64 bit floating point) back to 24 bit (fixed point).  Thus, to get down to 24 bit/44.1kS/s, Cakewalk should be using dither as part of the process.
 
But that does not mean that you, the user, should be adding dither during that process.  Cakewalk's software should have the dither functionality built in for those internal processing operations.  You, the user, should only be adding dither when you reduce the operating bit depth of a file (as in, from 24 bit to 16 bit).
 
If Cakewalk gives you a dither "option" in its sample rate conversion utility, I'd be interested to know what that option is for.  My guess is that the option is there only for when you are also reducing bit depth, as in going from, say, 24 bit/96k to 16 bit/44.1k.

OK good, thanks.
 
The reason for the dither is because it's being assumed that you're doing a bit depth reduction alongside the SRC. This will generally be true unless you are exporting to a 64bit or 32bit floating point file. 
 
 
In terms of dither type, the "problem" with noise shaping is that if repeatedly applied, it could concentrate lots of energy in a narrow frequency band and that could potentially cause problems.
 
Personally I would argue that in reality when dithering down to 24bit (or above), the dither level is so low that it's not really going to matter what you do. But since the level is so low, there's certainly zero harm in using simple rectangular dither if someone wishes to argue that this is a best practice.

I use Sound Forge and other programs as well and export to 32bit floating point for this purpose. Also, when using an external mp3 encoder, you want to export to 32bit floating point.
 
IOW, 32bit floating point is an interim file format to be input into another program, not a final destination format.

If you select 32 or 64 bit, it's floating point.

http://src.infinitewave.ca/
 
That is a site that compares the output spectra of many different sample rate converters (SRC).
Use "Sweep" for the test result. The best SRC is the one that makes a curved white line and nothing else.
Any other lines besides the single curved white line would be aliasing.
Sonar's is pretty good according to the graph.
 
About the sweep mode from the help file:
Swept sine wave with -6 dbFS peak amplitude, spanning the frequency range from 0 to 48 kHz for 8 seconds. As a result, the spectrograms of converted signals can be drawn. They allow identification of non-linear distortions introduced into the signal and aliasing. The dynamic range of this spectrogram is 180dB.
Before the 5 second mark, the tone is in the audible frequency range, so the level of harmonics and distortions in the left part of the spectrogram shows how signals in the audible range are distorted at different frequencies. After 5.5 seconds, the input tone goes above 22 kHz and cannot be represented in a 44.1 kHz format. So, ideally, it should be suppressed by the low-pass filter.