Page 3 of 3
Re: Intersample peaks
Posted: Tue Jul 15, 2014 5:27 pm
by Drnkhobo
@Martin - why do you use FIR filters for the oversampling? I'm a bit lost here
Also you mention
if you look at the time response of the toolkit filter: a single pulse gets heavily distorted with lots of ringing.
I'm assuming this is detrimental to any signal, or specific to pulses..?
Re: Intersample peaks
Posted: Tue Jul 15, 2014 7:46 pm
by martinvicanek
Drnkhobo wrote:why do you use FIR filters for the oversampling?
Because I thought that most DAC reconstruction filters were linear phase. Now I realize how silly that assumption was!
RECONSTRUCTION FILTERS ARE ANALOG!
In other words, they are certainly not linear phase. So please forget my previous posts in the thread, it is all nonsense.
Drnkhobo wrote:Also you mention
if you look at the time response of the toolkit filter: a single pulse gets heavily distorted with lots of ringing.
I'm assuming this is detrimental to any signal, or specific to pulses..?
Not necessarily - I'd say it depends on what you do with it. The toolkit IIR filter has an outstanding frequency response. That's good because it efficiently suppresses aliasing without much high frequency attenuation. This brickwall filter response, however, comes at the price of phase distortion. That's bad because you lose coherence, making percussive sounds less punchy. For the purpose of metering I think the toolkit filter is not a good interpolator (although I do not know which filter would be, see above).
So my personal conclusion is: if you are concerned about intersample peaks, just stay some 1 to 2 dB below the roof (make it 3-6dB if you are
very concerned). However, don't bother to be more sophisticated than that unless you really know exactly what happens in the analog path after the DAC.
Re: Intersample peaks
Posted: Thu Jul 17, 2014 3:09 pm
by Tronic
Thanks Martin for your comments,
my request regarding explanations about oversamplig,
was right about your technique implemented for this specific context and with the FIR.
Re: Intersample peaks
Posted: Thu Jul 17, 2014 5:06 pm
by Drnkhobo
Thanks for the feedback Martin, I guess assumption is the mother of all . . .
I was thinking about IS peaks for use in a plugin. Limiter thingy.
Admittedly I have been stumped for a while on another issue but I'll keep that for another thread
It's embarrassing
