• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

MiniDSP OpenDRC native equalizer "precision" (not a bug but might be interesting to someone)

Daltong

Member
Joined
Mar 9, 2022
Messages
47
Likes
5
So, in an adjacent thread I have had some issues with OpenDRC, it's weird resampling woes (mostly defeated) and slightly odd FIR filter behavior (decided just not to use FIR based EQ for now)

Decided to take a gander what would Amir's AeonRT EQ look "on the wire" at DAC output using same tools

Initially checked the sanity of the recording setup
(PC-toslink-OpenDRC-toslink-DAC->balanced connector to ESI record card) using test files and format settings that produce a good clean "flat" (deviations below 0.2 db) response with no anomalies (to ensure I'm doing everything right) without EQ

then run same with EQ

Result looks like this
Spectrum.png


Implemented using MiniDSP's "internal" EQ facility

While I am aware that it's generally normal for shelf EQs to cause a bit of a deviation before cutoff frequency, 0.5 DB "droop" between 50 and 90 Hz is quite a bit more than I expected (for the record if I "plan" this same EQ in rephase using minimum-phase filters, it predicts a "sag" less than 0.1 db in same region.

Also of note is that it overshoots the high shelf target value of 2, again, by 0.5 db (for reasons I can't even begin to understand)

I don't think it is a "bug" per se - probably just a matter of how a given EQ implementation + DSP processor work, but still kind of noteworthy IMHO and I decided worth a separate thread
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
So, in an adjacent thread I have had some issues with OpenDRC, it's weird resampling woes (mostly defeated) and slightly odd FIR filter behavior (decided just not to use FIR based EQ for now)

Decided to take a gander what would Amir's AeonRT EQ look "on the wire" at DAC output using same tools

Initially checked the sanity of the recording setup
(PC-toslink-OpenDRC-toslink-DAC->balanced connector to ESI record card) using test files and format settings that produce a good clean "flat" (deviations below 0.2 db) response with no anomalies (to ensure I'm doing everything right) without EQ

then run same with EQ

Result looks like this
View attachment 207346

Implemented using MiniDSP's "internal" EQ facility

While I am aware that it's generally normal for shelf EQs to cause a bit of a deviation before cutoff frequency, 0.5 DB "droop" between 50 and 90 Hz is quite a bit more than I expected (for the record if I "plan" this same EQ in rephase using minimum-phase filters, it predicts a "sag" less than 0.1 db in same region.

Also of note is that it overshoots the high shelf target value of 2, again, by 0.5 db (for reasons I can't even begin to understand)

I don't think it is a "bug" per se - probably just a matter of how a given EQ implementation + DSP processor work, but still kind of noteworthy IMHO and I decided worth a separate thread

I have no idea what you are talking about. It is very difficult to understand the supposed deviations because you do not show us a comparison of miniDSP and rephase.

Every time I've compared miniDSP to rephase they have been identical as long as you use proportional Q PEQs and enough taps. Although some platforms have different Q definitions for shelving filters miniDSP and rephase use identical definitions.

For the Aeon RT EQ here is a comparison of a measurement from an OpenDRC-DI running the miniSHARC-4x8-96 plugin and a modeled minimum phase FIR in rephase, they look pretty much identical to me and show none of the issues you are claiming. IMO, the supposed overshoot from the high shelf is just a slight level calibration issue as you are below 0 dB at ~500 Hz rather than exactly at 0 dB. There is a slight overshoot from the high shelf of ~0.2 dB because it is using a Q of 1.0 but not 0.5 dB as you are claiming.

Screen Shot 2022-05-18 at 9.49.04 AM.png


You seem to be making this way harder than you need to.

Michael
 
OP
Daltong

Daltong

Member
Joined
Mar 9, 2022
Messages
47
Likes
5
Yeah, I'm probably over-complicating things, fair

Meanwhile I also found a new firmware (apparently the one that gets placed into Program Files folder is old, and there's a newer one coming in the ZIP with installer itself, a fact that I only learned from OpenDSP tech support guy)

instead of 0.5 I now get 0.2-ish, (per REW, which I finally figured out and 0.3 as per RMAA) and like, I guess that will do, OpenDRC, that will do :)
 
Top Bottom