What Makes Dirac SO SPECIAL?

[Satanica]

I saw this today and think this is a good video about Dirac and also a bit of Audyssey and room correction in general.
 

 

[Keith_W]

FYI, a large portion of his video is based on Matthias Johansson's white paper which can be found here. Incidentally, Matthias Johansson (co-founder of Dirac) recently left his position as Dirac CEO and joined Harman. 

 

There are a number of errors in his video, especially re: the audibility of phase (about 13min). He makes no distinction between inter-channel phase response and intra-channel phase response. The former is audible, and the latter is audible within certain limits. There will be NO inter-channel phase discrepancy unless your DSP software is doing something inappropriate, for e.g. attempting to correct for excess phase caused by room asymmetry.

 

As for intra-channel phase discrepancy, the traditional thinking (Ohm's Acoustic Law) is that phase shift is inaudible with the exception of special test signals with headphones or speakers in anechoic chambers. The moment the speakers go into a listening room, it was thought there is so much phase soup that you have no hope of hearing a difference. However, it is currently thought that the audible threshold is (1) 15 deg of phase shift per ERB, and (2) phase shifts between ERB's which are far apart is inaudible. 

 

He said that mixed phase correction is better. That is true, but needs to be investigated with more nuance. MiniDSP/Dirac uses minimum-phase IIR at the bottom and linear-phase FIR at the top. Min phase means that the phase is rotated together with amplitude, so it can only be used to correct the min phase response. That is fortunate, because bass in rooms are mostly minimum-phase. The key word is "mostly". Min phase IIR can not correct excess phase in the bass region. For that, you need linear-phase FIR. The problem is, you then need a lot of taps and the MiniDSP (and also Audyssey) does not have that because it is computationally very expensive. 

 

I don't agree with mixed phase correction for the top. He himself says in his video that it is the speaker's anechoic response / CEA2034 that should be corrected. That's right. You're not going to get a proper anechoic measurement in your listening room no matter how many points you measure from. It is better than a single point measurement and trying to perform inappropriate correction on that, but it is probably worse than doing nothing at all (see Toole Chapter 7, he goes so far as to call it a farce). Attempting a mixed phase correction for the top whilst relying on improper measurements WILL result in weird phase issues that he describes later in the video where he says his friends ask him to turn it off. 

[Satanica]

miniDSP is not the only device that supports Dirac Live.

 

Dirac very much so takes latency into consideration, because Dirac is not just expected to work with just audio, but with both and video at the same time and in audio studios where real time tracking might be involved.

The main point of the video, is what does Dirac Live (and a couple of other systems) bring just over normal minimum phase EQ.

[Hi-Fi Whipped]

I don’t much about the dots and dashes but since I started using Dirac Live about 5 years ago, after a short learning curve it’s been the backbone of any system I’ve had that just brings everything together and makes listening so enjoyable.

 

Thank the audio gods for Dirac!