Convolver for Digital Room Correction and EQ fir filters.

[pvog]

I would like to see support for convolution with a convolver like BruteFir on the WiiM Pro.
This can be used to load fir filter wav files from full room correction software like drc or Acourate.
But also as an easy way to load more complex eq filters from REW.

This would make WiiM Pro a very capable and cheap solution for full digital room correction.

 

[Brantome]

it is indeed very public and respectable... the irony is that it can nevertheless lead to certain difficulties like this story of acquisition-microphone for auto correction... ;-)

That’s maybe another reason that they’re investigating the use of the mic in their own remote - like iOS device mics, they’re a known entity. If they’re of sufficient quality for RC, that again makes it a viable solution for the less technically minded consumer.

 

[canard]

That’s maybe another reason that they’re investigating the use of the mic in their own remote - like iOS device mics, they’re a known entity. If they’re of sufficient quality for RC, that again makes it a viable solution for the less technically minded consumer.

request a heavy calibration in this case... (especially since the physical arrangement of the microphone in the remote control has a significant impact...)

 

[rrobot]

indeed very public and respectable..

Absolutely agree here, WiiM are to be commended for listening to their user community in ways few large electronics manufacturers ever do.

As FIR filters, i would be very surprised if Apple weren't employing a whole range of IIR and FIR filters in the homepod for example so the technology could i suspect actually be deemed mainstream and consumer...albeit in case of Apple or Sonos etc none of it is visible to the user. Im not comparing the homepod to the WiiM range here btw, just remarking on the increasing ubiquity of sophisticated DSP.

I can certainly see the goal is automated correction to appeal to maximum users. Whether convolution files is on that roadmap is for WiiM to judge i guess, but i personally think it would appeal to a reasonably sizeable niche group.

 

[Mr Ee]

reasonably sizeable niche group

i just googled this, and got Oxymoron

 

[rrobot]

and got Oxymoron

Fair point, that was a rather contradictory statement! But lets be honest, the group of users who will work out and apply parametric EQ is very niche. Few of the new features here are going to have widespread appeal but be very useful to a small cohort. Its not a huge conceptual stretch from parametric EQ to convolution filters though it may be technically. Will most users upload convolution filters? No way... but i think that conceivably applies to a lot of features which have been deemed worthy.

 

[planetti]

Dear FIR dreamers,

from my FIR convolution experience there are some hurdles to pass.

The hardware requirements for convolution is about the computing performance of a RPI 2B BruteFIR for 2 channels @192kHz, 65000taps. The miniDSP AD-chip, which is specialised on DSP calculations, provides e.g. in the OpenDRC for 2 channels @48kHz, 6144taps (which is already at the lower end of good results). For normal convolution the number of taps results in a frequency resolution per tap over the bandwith, e.g. 24kHz/6144taps, which is not very accurate at low frequencies. Some DSP therefore use IIR-PEQs for room modes.

For a symmetric algorithm the peak is in the middle of the impuls response, which means that there is a filter dependent delay for the sound reproduction, e.g. 65000/2 samples @48kHz result in about 0.7 seconds (so far from lip sync).

The next field of trouble is the filter generation. If there is a substantial time/phase correction (for which FIR filter are used for) with a high gradient, then the filter may produce preringing which results in an unpleasant noise before the tone. It is not easy to deal with these - you need lot of experience or spend many adaption loops. I do not know a one button software with convenient results. Dirac comes close, but takes a high licence fee.