Using rephase and MSO together?

[kifeep]

TL;DR: Are there a priori reasons to think combining rephase and MSO wouldn't work? If not, what might be an appropriate strategy to do so?

--

I've been doing DRC using MSO and REW to generate IIR filters for a miniDSP Flex HT, and have been happy with the results.

But reading ASR, I've become FIR-curious. So last weekend I spun up a headless linux instance on an old 2012 mac mini I had, and got camillaDSP running. (Well, sort of running - at this point it's just processing zeros from and to null devices.)

So now I'm thinking about how to add it to my signal chain and I'm considering ordering a used M-Audio Fast Track Ultra 8R for this purpose. But before I spend any money, I'd like some reassurance that this isn't just a harebrained scheme.

A quick search on ASR for content that contain both 'rephase' and 'MSO' mostly surfaces posts in which they are just listed among software tools. This leads me to believe that few if any members are using both tools in the same system.

My general (naive) thought is to use rephase to generate FIR filters for camilla DSP to correct frequency and phase on each channel separately, and then return to MSO to integrate speakers and subs. The hope is that I might get better MSO results with the combination of a) more well-behaved inputs (especially w/r/t/ phase) and b) more available filter slots (since I won't need to reserve any for corrections between crossover and transition frequencies).

Does this make any sense? Would it be worth the effort?

Thanks.

 

[OCA]

Your main advantage with rePhase will be ability to add non-causal filters. All MSO filters are causal including IIR and allpass filters. Causal means having no content in the impulse in negative time. Adding filters with negative time content is tricky. In practice, if you go below -3ms (Dirac's limit is about -4.5ms) or so you'll start generating artifacts like pre-echo thus the use of non-causal filters are basically limited to crossover phase linearization for your speakers. There's almost nothing extra you can do for your subs.

 

[lincolnalmiron]

It is worth considering that CamillaDSP must process signals with sufficient computing power, especially if you plan to use long FIR filters.

 

[kifeep]

I ended up buying an inexpensive used AD/DA and hooked everything up. Unfortunately, I quickly abandoned the project due to the additional latency of the AD/DA and CamilaDSP overhead – before even getting to crafting any FIR filters. It probably would have been fine for music only, but was unacceptable on my system which is both music and HT.

But thank you both for your feedback.

It is worth considering that CamillaDSP must process signals with sufficient computing power, especially if you plan to use long FIR filters.

Just in case someone else finds this and wants to use similar equipment, the old mac mini running ubuntu server had way more computing power than needed. Using 128k dummy filters on 8 channels, it never exceeded about 4% CPU utilization.