@Nathan: That is not correct. Linear phase filters do add delay, across all frequencies, equally, hence the name linear phase. What you are thinking of is labelled "zero phase" filters but those are basically just a processing trick that only works in offline filtering, so not during live performances.
ah, yes..the whole confusion is about using delay to alter phase responses (as is common practice in live sound) a technique labelled as phase alignment. You will understand this will only work in some cases.. About your 'processing trick only for offline filtering'...euhmm no not really..we definitely are using all kind of FIR based techniques in live performances, X-overs being the more simple applications. Multiband compression , beam steering the more complex.. And yes, we are limited by time.. :) so you do have to be careful what you use in what situation.
@@theimunten6325 Again, phase response and delay are two sides of the same coin. If you add a delay then you are literally increasing the slope of the phase response. Naturally, a flat delay is equal to a linear phase response. So you're right that such a delay cannot work in cases where you're trying to fix wildly varying phase vs. frequency problems. And about your comment on zero phase filters 'euhm no not really': yes, really. Zero phase filters are not causal. You cannot do zero phase filtering on live sound. The universe doesn't allow you to look into the future. Of course, you can do live filtering by turning zero phase filters into causal linear phase filters - that's trivial. Just add delay or in other words increase the slope of the phase response until the filter is causal.
Hey John. Thanks for your interest in the video. If I said that linear phase filters do not add delay, that was an error. Thanks for catching that! What I should have said is: The delay is constant across the whole frequency spectrum. If you'd like a deeper dive on this topic, I have a pretty cool interview with Michael John over here - www.sounddesignlive.com/myth-fir-filters-always-add-a-lot-of-delay-and-are-impractical-for-live-sound/
@@nathanlively Thanks for the interesting link. I've developed DSPs myself, for example I've developed a component that creates a minimum phase FIR filter based on a user-drawn magnitude response. It's also used in graphical EQs with a fixed number of bands. It's using a FIR filter exactly for the reason Michael mentioned, but he didn't go into detail. It's much simpler to design a FIR in the frequency domain (where you can directly specify the magnitude) instead of trying to approximate a number of IIR filter prototypes to match the desired response... as for example Room EQ Wizard does. :)
Within the premise off LTI (linear, time invariant) : min phase == phase response as dictated by magnitude response, max phase == the inverse phase response of the latter, Lin phase == those 2 glued together to combine to a zero phase shift..That last 1 gives you that infamous 'latency' (it is not latency!).. Read a bit about Fourier transform..but really do start to understand that Delay is NOT the same as phase shift
Linear phase is not zero phase shift. It's literally a linear phase shift, hence the name linear phase filter. Delay is absolutely equal to linear phase shift. It's literally defined as the derivative of difference in phase over frequency. "Latency" is a very broad term that can mean a lot of things. One can interpret the delay added by a linear phase filter as latency of the filter system.
@@xnoreq A side effect off the whole corona thing....now we have time to split the hairs..hahaha If we could agree that when we are not interested in overall time differences as we are examining phase differences one could say the phase shift is 0 (relatively in time) Latency: indeed a broad term in audio usually associated with the buffers necessary to process audio in CPU..if CPU are more powerful latency can be decreased.."latency" induced by FIR filtering can not by altered and hence shouldn't be called that to avoid confusion
@@theimunten6325 I do not think being accurate is splitting hairs. I don't see why one wouldn't be interested in delays because adding differently delayed signals is literally what a comb filter is/does ... because of the different slopes of the phase responses. For example, 1 sample of delay (at 44.1 kHz that is 22.6 us) is equivalent to a linear increase in phase vs frequency towards -180° at 22.05 kHz. Mix that with the original and you get a null at 22.05 kHz. Do 3 samples of delay (or 68 us) and you now get two nulls because phase linearly increases to -540° (that's 3 * -180°). I agree on the usage of the term latency. It shouldn't be used in time-invariant systems such as the filters we're talking about.
The phase graph represents the radial component of the audio waveform. Delay between out and in will cause a change in the phase graph, unless delay compensation is used.
@Nathan: That is not correct. Linear phase filters do add delay, across all frequencies, equally, hence the name linear phase.
What you are thinking of is labelled "zero phase" filters but those are basically just a processing trick that only works in offline filtering, so not during live performances.
ah, yes..the whole confusion is about using delay to alter phase responses (as is common practice in live sound) a technique labelled as phase alignment.
You will understand this will only work in some cases..
About your 'processing trick only for offline filtering'...euhmm no not really..we definitely are using all kind of FIR based techniques in live performances, X-overs being the more simple applications. Multiband compression , beam steering the more complex.. And yes, we are limited by time.. :) so you do have to be careful what you use in what situation.
@@theimunten6325 Again, phase response and delay are two sides of the same coin. If you add a delay then you are literally increasing the slope of the phase response.
Naturally, a flat delay is equal to a linear phase response. So you're right that such a delay cannot work in cases where you're trying to fix wildly varying phase vs. frequency problems.
And about your comment on zero phase filters 'euhm no not really': yes, really. Zero phase filters are not causal. You cannot do zero phase filtering on live sound. The universe doesn't allow you to look into the future.
Of course, you can do live filtering by turning zero phase filters into causal linear phase filters - that's trivial. Just add delay or in other words increase the slope of the phase response until the filter is causal.
Hey John. Thanks for your interest in the video. If I said that linear phase filters do not add delay, that was an error. Thanks for catching that! What I should have said is: The delay is constant across the whole frequency spectrum.
If you'd like a deeper dive on this topic, I have a pretty cool interview with Michael John over here - www.sounddesignlive.com/myth-fir-filters-always-add-a-lot-of-delay-and-are-impractical-for-live-sound/
@@nathanlively Thanks for the interesting link.
I've developed DSPs myself, for example I've developed a component that creates a minimum phase FIR filter based on a user-drawn magnitude response.
It's also used in graphical EQs with a fixed number of bands.
It's using a FIR filter exactly for the reason Michael mentioned, but he didn't go into detail.
It's much simpler to design a FIR in the frequency domain (where you can directly specify the magnitude) instead of trying to approximate a number of IIR filter prototypes to match the desired response... as for example Room EQ Wizard does. :)
Within the premise off LTI (linear, time invariant) : min phase == phase response as dictated by magnitude response, max phase == the inverse phase response of the latter, Lin phase == those 2 glued together to combine to a zero phase shift..That last 1 gives you that infamous 'latency' (it is not latency!).. Read a bit about Fourier transform..but really do start to understand that Delay is NOT the same as phase shift
Linear phase is not zero phase shift. It's literally a linear phase shift, hence the name linear phase filter.
Delay is absolutely equal to linear phase shift. It's literally defined as the derivative of difference in phase over frequency.
"Latency" is a very broad term that can mean a lot of things. One can interpret the delay added by a linear phase filter as latency of the filter system.
@@xnoreq A side effect off the whole corona thing....now we have time to split the hairs..hahaha
If we could agree that when we are not interested in overall time differences as we are examining phase differences one could say the phase shift is 0 (relatively in time)
Latency: indeed a broad term in audio usually associated with the buffers necessary to process audio in CPU..if CPU are more powerful latency can be decreased.."latency" induced by FIR filtering can not by altered and hence shouldn't be called that to avoid confusion
@@theimunten6325 I do not think being accurate is splitting hairs.
I don't see why one wouldn't be interested in delays because adding differently delayed signals is literally what a comb filter is/does ... because of the different slopes of the phase responses.
For example, 1 sample of delay (at 44.1 kHz that is 22.6 us) is equivalent to a linear increase in phase vs frequency towards -180° at 22.05 kHz. Mix that with the original and you get a null at 22.05 kHz.
Do 3 samples of delay (or 68 us) and you now get two nulls because phase linearly increases to -540° (that's 3 * -180°).
I agree on the usage of the term latency. It shouldn't be used in time-invariant systems such as the filters we're talking about.
Hi bro, sorry for the ignorance, the phase graph represents the frequency delay between the input and the output ?
The phase graph represents the radial component of the audio waveform. Delay between out and in will cause a change in the phase graph, unless delay compensation is used.
What about zero phase?
What is zero phase?
Wait...I can google that myself.
www.dsprelated.com/freebooks/filters/Filters_Preserving_Phase.html
whats the best for gaming bro ? i play fps like pubg please help :)
hmmm, I'm not following
Do you know when this linear cycle will end that we are in right now with the moon being in Pisces?