In the case of DACs, different use case: low pass filters instead of high pass. Practical examples are easy with DACs that offer selectable filters. Can also use Audirvana playback with own upsampling and filters to test things out.
I wouldn't trust a blanket statement on filter for all dacs. To these ears and on this dac here, minimum phase doesn't sound like shit. For others, all bets are off.
I think minimum phase makes a lot of sense for EQ filters. When correcting a speaker/headphone FR you're essentially correcting both the phase and magnitude response at the same time.
That article AFAIK is total bullshit. And no, minimum phase filters make no sense for EQ. The only reason one uses a minimum phase filters is to have as little latency and as little latency as possible. Otherwise they are total shit. I'm not just talking. I've been there and I know EE PhD's have been burned by this.
As far as practical examples is concerned I have process shit loads of signals with both linear phase and minimum phase filters. I think I even uploaded some examples here and/or in Changstar. I don't know what else to do. That's as f'ing practical as one gets.
I'm all for that. Which one sounds better is not the problem. That's completely a personal choice. But those guys are talking about phase shifts in the frequency content. And the technical claims are all fouled up. In terms of reconstruction a linear phase filter will always destroy the equivalent minimum phase. Always.
There are always going to be phase shifts once the signal goes through a filter. It's just that with a linear phase filter all frequency content adds up coherently and with the right "padding". It only results in latency. This is probably not obvious to most folks. But is very basic for DSP engineers.
But what about when we want to EQ a bump in the frequency response of a headphone with parametric EQ? We want to fix both the bump and the phase error that comes with it.
Anyway, gonna test that now.
Then again once you use FIR filters anyway you can correct the phase response, too. I'm talking about "simple" parametric EQ filters, but there's really no point to limit yourself to that. Plus you can fix phase errors from speaker crossovers, etc.
A PEQ is a set of cascaded bi-quad IIRs. There is no way in hell those are going to be linear phase. A FIR is superior when it comes to phase. But they tend to be much larger for similar specifications.
This is moving away from DAC filters and into EQ. But as far as that you don't necessarily care about the EQ by itself. You want the overall response to be linear phase.
Read these "rules" AND introduce
yourself before your first post
Being true to what the artists intended
(opinion / entertainment piece)
Comments on Profile Post by frenchbat