• SONAR
  • The Foxboro Plug-In Upsampling Thread
2015/06/30 12:21:59
Anderton
I'm pretty sure there are going to be questions/comments about this, so thought I'd be pre-emptive and create a thread so related posts wouldn't get lost in a more general Foxboro thread.
 
First thing I'd recommend is reading the eZine description of what it does, and the limitations. It's a very new concept so there are some fine points.
2015/06/30 12:28:24
Keni
Anderton
I'm pretty sure there are going to be questions/comments about this, so thought I'd be pre-emptive and create a thread so related posts wouldn't get lost in a more general Foxboro thread.
 
First thing I'd recommend is reading the eZine description of what it does, and the limitations. It's a very new concept so there are some fine points.


Thanks Craig...

But the link is blocked?

Keni
2015/06/30 12:29:47
Fred Holmes
FYI
Web page (http://static.cakewalk.co...nar_foxboro_update.pdf/) reports as not available @ 12:28 EDT
2015/06/30 12:35:37
scook
try this http://static.cakewalk.co...NAR_Foxboro_Update.pdf
 
Edit: I see Craig has fix msg #1
2015/06/30 13:41:41
John
Why would there be a problem with the way its always been? Where is the oversampling meaningful, in what pert of the spectrum? I am a bit doubtful as to its necessity. 
  
2015/06/30 13:47:34
scook
here is some justification http://forum.cakewalk.com/FindPost/3046633
2015/06/30 13:55:18
lfm
John
Why would there be a problem with the way its always been? Where is the oversampling meaningful, in what pert of the spectrum? I am a bit doubtful as to its necessity. 
  


I love that feature of Metaplugin(2xoversampling), and always run Dimension strings programs inside that.
Quite noticable difference.
 
I guess it depends on quality of sample libraries also.
One that stretch samples a lot to resample to many pitches is especially great value.
One of quality sampled at every pitch in keyboard range, less so.
 
In general you could say that aliasing is reduced. Some rough edges of mid and high frequency content reduced.
 
Maybe start with some chords of strings.
2015/06/30 14:05:32
John
scook
here is some justification http://forum.cakewalk.com/FindPost/3046633


I do know about the arguments I was on that thread. 
2015/06/30 14:10:53
Anderton
John
Why would there be a problem with the way its always been? Where is the oversampling meaningful, in what pert of the spectrum? I am a bit doubtful as to its necessity. 
 

 
Excerpted from the eZine:
 
Some plug-ins, both processors and virtual instruments, can produce unwanted artifacts when running at lower sample rates (e.g., 44.1 and 48 kHz) if they don’t oversample internally and lots of high frequencies are present. Most modern plug-ins give the option to oversample at the expense of drawing more CPU power, but many older ones still in common use do not. Amp sims and synths are affected the most, but so are dynamics processors and some reverbs. The artifacts result from “foldover distortion,” which produces a sort of “wooly” noise when higher frequencies from the digital signal processing "fold back" to create noise in the audio range.
 
The typical workaround is running projects at higher sample rates, such as 88.2 or 96 kHz. However this uses more CPU power, which can limit the number of tracks and plug-ins you can run in real time.
 
[Upsampling] provides the benefits of processing or generating audio at a higher sample rate in projects using lower sample rates (e.g., 44.1 or 48 kHz). When selected, SONAR begins the bounce process by upsampling the incoming audio to the specified higher sample rate, processes the plug-in at the new rate, then downsamples the resulting output to the current project sample rate. This process happens automatically, behind the scenes; it works with VST, DX and/or virtual instruments.
 
LIMITATIONS OF UPSAMPLING
 
Please note that only some plug-ins, generally older ones, benefit from upsampling and only if significant high frequencies are present.If no harmonics exist that reach into the range of the clock, there will be no foldover distortion, hence no need for upsampling.
 
There can be a significant improvement in sound quality with some plug-ins, no improvement with others, and a few may actually sound worse. So, upsampling is enabled on a per-plug-in basis—in other words, enabling upsampling for one plug-in enables it for all instances of that one plug-in, in any project. Because this rendering process is CPU-intensive, do not enable upsampling for a particular plug-in unless you can hear an actual difference.
 
Note that SONAR’s high-end sample rate conversion requires considerable CPU power, so this process is available only when doing a fast (non-real-time) bounce. Also, the maximum upsample rate is 384 kHz, so upsampling is not available for projects that run above 192 kHz; and at present upsampling cannot be applied to plug-ins in surround buses, or to bit-bridged plug-ins or region effects. Finally, note that some plug-ins may not support operating at a higher sample rate. In this case, SONAR displays an error message toast notification, and performs the plug-in bounce at the original project sample rate.
 
It’s also important to remember that the sound designer probably built a sound based on what was heard. If you now process at high sample rates, the sound may be brighter because the high frequencies are no longer being folded back, and there could be less perceived low end because the foldover distortion is no longer there. Whether that sounds “better” or not is subjective.
 
WHY UPSAMPLING CAN IMPROVE AUDIO QUALITY
 
It may seem counter-intuitive that after upsampling to a higher sample rate and rendering, returning to a lower sample rate preserves the benefits of working at the higher sample rate. However, these benefits occur in the audio range, and as low a sample rate as 44.1 kHz has no problem reproducing sound in the audio range. Because upsampling processes at higher frequencies, when sample rate-converted back to a lower sample rate, the frequencies that could cause foldover distortion are no longer present. 
 
 
2015/06/30 14:15:53
Anderton
Here are some comments that Noel passed along to me from people who tested it:
 
“Much clearer open with more air.”
“Distinct difference in the quality of reproduction when Upsampling engaged.”
“Didn't realize DSP had issues operating at ‘normal’ sample rates. Thanks for opening my ears, Craig.” [in reference to a demo file I had prepared for the New Music Seminar and posted online]
“Everything sounded great, perhaps a little too great - the project sounded different clearer, more air, more open.” 
 
The last comment raises a definite issue: if you've been mixing a song and at the last minute decide to upsample and the sound is different, you may need to compensate for the change. For example with drums, foldover distortion might make the sound appear "fatter" so removing the distortion sounds subjectively less desirable.
 
Also, I have yet to hear any difference with sounds recorded into an interface and played back without processing. There has to be either signal generation or processing inside the computer itself to make any difference.
© 2026 APG vNext Commercial Version 5.1

Use My Existing Forum Account

Use My Social Media Account