[sdiy] Digital filtering (Oversampling and downsampling)

Andy Eakin aeakin1 at matrixblues.com
Tue Aug 4 18:58:08 CEST 2009


cheater cheater wrote:
> Huh?
>
> Oh, are you saying the flux capacitor connects between the cyclotron
> and the Z-Machine via the quantum loopback doohickey?
>
> ;o)
>
> D


Lol, something to that effect....

Really what it boils down to is a big "It depends".  In all honesty, if 
there is no specific need to go down to 48kHz I would say keep it 192Khz 
all the way to the DAC.  If that is not possible due to horsepower in 
the DSP or Latency, then the only real option is to use a FFT or more 
advanced algorithm again depending on your DSP Hardware and Latency 
tolerance.  Most DSPs should already have a basic basic FFT or more 
advanced CODEC in their libraries with the necessary hooks to make this 
easy, it really becomes a matter of personal taste as to the final sound. 

However if you ever find a perfect solution to this quandary of 
downsampling you MAY be assassinated by Ninjas hired by the Audiophile 
industry -OR- given a Nobel Prize.

On another note, I would be curious to see the amount of error 
introduced in Digitally creating a waveform then downsampling vs 
creating it at the intended bitrate.  Does 2x, 4x or 8x smooth out 
better? Where do we hit the point of diminshing returns?

Lets face it, Digital Signal Processing as a topic is about as close as 
most of us hobbyists will ever get to rocket science (unless that is 
your day job).

Andy



More information about the Synth-diy mailing list