[iXpertMan]

Hi!
 
I was just browsing through my Mac's applications (utilities section) when I came across a program called 'Audio MIDI Setup'. In its setting it states 'format', represented by a frequency and a bit size. These setting can be changed to different numbers, frequency from 44.1Khz to 96Khz and bit size from 16bit to 24bit.
Is this the upsampling that my Mac can perform or something else? Also what would it be best to set these setting to?
 

 
Thank you. 

 

[drews]

Yeah, this is software upsampling/downsampling (unless your source material matches it).  If you're playing CDs or ripped CDs, just leave it at 44.1/16.  If you have higher resolution music (from HD Tracks etc.) you can install BitPerfect which effectively changes this setting on-the-fly based on the track iTunes is currently playing to avoid resampling (iTunes playback normally uses whatever output format was active when the app was loaded).
 
Drew

 

[iXpertMan]

Thanks for clearing this up. But why would I leave the settings at 44.1Khz/16bit when I am playing music from a CD. Yes I understand that this is its native format, but oversampling is a good thing, so why not do it all the time?

 

[drews]

Oversampling in a DAC is different from upsampling from 44.1/16 to 96/24 (for example).  The latter is widely agreed to be a bad thing (the upsampling won't improve the original material and rounding errors due the conversion will actually make it worse).  It's akin to running a laptop with a 1280x1024 resolution in 1024x768 mode if that makes sense...
 
Drew

 

[iXpertMan]

I don't think the example of upscaling a number of pixels is the same as upsampling/oversampling audio. I don't understand the difference between upsampling & oversampling (as they to me, seem like the same thing), but I think that they essentially add extra points of sampling on an already existing graph, trying to improve the detail. Since we already have a graph and then just use extra points with averages we can 'create' more data. Can both upsampling and oversampling be represented with this graph?
 
http://www.audioholics.com/education/audio-formats-technology/upsampling-vs-oversampling-for-digital-audio/image_view_fullscreen

 

[drews]

That graph is showing a 2X sample rate...  I thought we were talking about audio that was sampled already at 44.1kHz - upsampling can't get back information that was not present in the 44.1kHz recording.  Upsampling to 88.2/24 etc. would at least not have the rounding errors (again similar to upsampling video to exactly 2X resolution) since it just adds bits but I'm not sure what the point would be (also you are transmitting more bits which increases jitter problems).  Upsampling is converting from a source rate to a different rate, oversampling is sample rate multiplication done internally by DACs as part of the conversion back to analog (to remove high frequency noise etc.).

 

[Chromako]

What I can say is that it won't hurt as long as you have a good dac that can handle the bitrate well. I think I noticed a subtle, pleasant, difference in sound, but it was small. 
 
I'm actually a fan of upsampling/interpolation of music (if done right), and adore the algorithms developed by iZotope. My philosophy is that the 44.1k Redbook protocol was chosen in part due to technological limitations back in ye olde day, so it's not sacred. 
 
A similar concept is that in Photoshop, I regularly bring out details that were surpessed in photographs by upsampling and applying advanced interpolation techniques. Try it. 
 
I think Fidelia has a trial mode? Give it a try...it's yummy and full of iZotope interpolation software. Except for the decidedly un-yummy GUI. 

 

[iXpertMan]

I am just using the line out of my Mac, which uses the Intel HD Audio card, and I am trying to improve sound quality with some upsampling.
As for the graph, lets say that my song is the top graph, who's native frequency is in fact 44.1Khz. So the second graph would represent that same audio upsampled to a frequency of 88.2Khz, right? Which in turn would lead to more details and possibly 'better' sound quality... I am not sure who would upsample the audio, either the DAC or computer software, but does it even matter, as it's still the same concept?

 

[drews]

I think the important thing to understand is that the samples (upper-right graph) have enough information to reproduce the original waveform back exactly (subject to 20-20kHz with 96 dB of range - look up Nyquist–Shannon sampling theorem).  If you upsample and interpolate more data points you're either going to produce the same waveform or a slightly deformed variation of it due to rounding errors (depending on the new sample rate).  You might like the sound of upsampled music better (or EQ'd or other effects like crossfeed or reverb or whatever) and that's fine of course but the point is upsampling will not restore "lost detail"...