MDAC First Listen (part 00111001)

[DNM]

The MiniDSP boxes we tried had a very negative effect on sound quality - the issue seems to be the SharkDSP sample rate converter block used to allowing a single output sample rate and which ease the filter coefficient requirements.

The SharkDSP sample rate converter sounds REALLY bad.

Which MiniDSP's did you try?

I've used a 4x10HD for 3yrs or so and don't hear any material difference running a digi feed through the unit with all the filters off to an external DAC compared with direct to same DAC.

In fact, just to pacify the audio nervosa you've planted in my head with this post, I've just plugged an RCA coax cable from my Allo Digione to my DAC and A/B'd with the BNC output via 4x10HD > AES/EBU to same DAC and there's no discernible difference. Where you get this "REALLY bad" observation from is beyond me. I also find it somewhat distasteful that a chap who has taken peoples money and failed to deliver a product has the audacity to diss another manufacturers product on a forum, especially when said manufacturer actually delivers its products and provides exceptional VFM and service.

I have had a MiniDSP2x4HD between MDAC and amp to correct room modes. The result was not very promising. Less boom but much less lively. I am back to passive room / loudspeaker treatments. I guess double DA/AD/DA is not the way to go.

Its very easy to get DSP hopelessly wrong. Without knowing precisely what your problem is, its impossible to cure it. Correcting solely by ear is futile. Measurements are essential then you can manually create correction filters or generate and upload them from the likes of REW. I'm afraid DSP gets a bad name from people that haven't applied it correctly. I almost gave up but persevered and now wouldn't be without it.

 

[jansen]

I guess, I did it right. Umik mic measurements into REW for each single speaker, loaded into MiniDSP, slight correction of room mode frequencies around 35Hz and 70Hz. Or did I miss something?

 

[JohnW]

With regards to our experiences with the MiniDSP we used the hardware in "Flat mode" and at first could not understand what was wrong with the system sound quality - it was only after some time that we realised that the unit was up-sampling even when set to "Bypass" or Flat (No Eq).

The sound lost L-R sound stage and just sounded worst with the MiniDSP system in the loop... This is before ANY DSP correction.

Personally, I'd prefer to live with my system frequency deficiency then loose the "realism" to the sound quality - MDAC2 PRIMARY GOAL is sound quality.

Its easy to try for oneself by listening to the system with / without the DSPin the loop - set to Flat EQ (so your only listening to the Shark DSP up-sampling) - if ones happy with the sound then there no reason to loose any sleep over my post!

I'm not panning a product, only say why we are NOT including DSP into our hardware as it does not meet our goal of transparency (and I blamed the Shark DSP upsampling NOT the product) - This does not stop anyone from using an external DSP.

 

[DNM]

Well we still don't know which MiniDSP products you tested and it appears you have a convenient excuse not to deliver a product with DSP. So whats your excuse with DAC's and Streamers?

 

[DNM]

I guess, I did it right. Umik mic measurements into REW for each single speaker, loaded into MiniDSP, slight correction of room mode frequencies around 35Hz and 70Hz. Or did I miss something?

Impossible to say without knowing target range, measurements before and after etc. I would only use EQ per channel to correct bass. Auto correction from a single mic measurement over too high a frequency range can certainly suck the life out of the sound.

 

[ti33er]

Personally, have only found EQ to ‘flatten’ and take the punch out of my music ...perhaps if you have an extreme bass mode, could use to tame that a bit; otherwise sit closer to your speakers and further away from your rear wall 8-/

 

[jansen]

Impossible to say without knowing target range, measurements before and after etc. I would only use EQ per channel to correct bass. Auto correction from a single mic measurement over too high a frequency range can certainly suck the life out of the sound.

I measured from 20Hz to 120Hz and corrected the room mode frequencies. No auto correction.
But anyway, I am happier than before. I have extended the bass reflex tubes to reduce the room mode impact. Special service from the manufacturer.

 

[fusion5]

The DSP feature was one of the main reasons I signed up for this. It would have allowed me to easily correct a 25dB room mode regardless of the source. And now the feature is just gone?

 

[DNM]

I measured from 20Hz to 120Hz and corrected the room mode frequencies. No auto correction.
But anyway, I am happier than before. I have extended the bass reflex tubes to reduce the room mode impact. Special service from the manufacturer.

Did you uploaded REW correction filters from 20Hz to 120Hz or did you simply measure and apply your own EQ?

 

[timola]

Yes make the initial measurement with something like REW, create the correction, and then re-measure to confirm the efficiency of the correction.
Keith

I'm using DIRAC

The DSP feature was one of the main reasons I signed up for this. It would have allowed me to easily correct a 25dB room mode regardless of the source. And now the feature is just gone?

You could still do it from you computer, if you will be using one, REW is free.

 

[Purité Audio]

Use REW to verify Dirac’s correction, it might be interesting to see whether they agree.
Keith

 

[timola]

Use REW to verify Diracs correction, it might be interesting to see whether they agree.
Keith

Yes, I've been meaning to do that for a while.

Like I said on another thread these DSP vendors should simply allow you to measure with the correction in place...I wonder why they just don't do that. Dirac doesn't for sure.

 

[DNM]

Yes, I've been meaning to do that for a while.

Like I said on another thread these DSP vendors should simply allow you to measure with the correction in place...I wonder why they just don't do that. Dirac doesn't for sure.

Because its impossible unless you take the same sweeps from the exact same positions before and after. All you can do is record the REW Sweep and measure with the Dirac filter applied around the Sweet Spot and get a general idea of the effectiveness of the correction.

 

[timola]

Because its impossible unless you take the same sweeps from the exact same positions before and after. All you can do is record the REW Sweep and measure with the Dirac filter applied around the Sweet Spot and get a general idea of the effectiveness of the correction.

You say its impossible, then say how it would have to be done....I wouldn't mind doing all the measurements again with the filter in place, I'm quite good at it now.
I reckon there is a sinister reason why they don't allow it.

 

[timola]

Use REW to verify Dirac’s correction, it might be interesting to see whether they agree.
Keith

Keith, you always recommend just fixing the bottom end. This is where you say the major room issues are. I do understand this, but why not fix everything, there may be crossover point or a slight driver mismatch further up, why not fix that too?

Surely if you just correct the bottom, the software just multiplies the rest (above the bottom) with zeros so all the frequencies are manipulated anyway.

Lately I'm convinced DSP flattens the sound, I.e., takes some of the life out of it.

Your comments would be appreciated.

 

[JohnW]

Well we still don't know which MiniDSP products you tested and it appears you have a convenient excuse not to deliver a product with DSP.

We tried with the two MiniDSP boxes we have in the lab:-

1. DDRC-22D (MiniDSP DiracLive Technology)

2. NanoAVR

As the sonic degradation is native to the SharkDSP ASRC - I suspect it will be the same for any DSP unit that's based on the Analogue devices SharkDSP.

I believe it is possible to avoid using the ASRC but it adds complexity to the software as you need to have separate filter coefficients for each sampling rate.

My point is that if the sound is degraded before ANY DSP is performed (flat mode, just ARSC), then what comes afterwards is just a band aid as you cannot reconstruct the damaged information.

 

[timola]

We tried with the two MiniDSP boxes we have in the lab:-

1. DDRC-22D (MiniDSP DiracLive Technology)

2. NanoAVR

As the sonic degradation is native to the SharkDSP ASRC - I suspect it will be the same for any DSP unit that's based on the Analogue devices SharkDSP.

I believe it is possible to avoid using the ASRC but it adds complexity to the software as you need to have separate filter coefficients for each sampling rate.

My point is that if the sound is degraded before ANY DSP is performed (flat mode, just ARSC), then what comes afterwards is just a band aid as you cannot reconstruct the damaged information.

That's interesting. With Dirac (software on PC) I have filters at 16/44 and 24/?(96 maybe)....I wonder if then you should only play those files on that filter, otherwise it gets up/down sampled??? What a palava, I just like to play tracks and I have a big mixture of sampling rates on both mp3 and flacs....no wonder it sounds better without DSP.

 

[JohnW]

The DSP feature was one of the main reasons I signed up for this. It would have allowed me to easily correct a 25dB room mode regardless of the source. And now the feature is just gone?

I dont mind building a special version with your MiniDSP box customized for the FDAC and slaved units.

Also, I believe that REW filter coefficients can be run on the streamer hardware - the streamers Quad core 1.2GHz CPU with its GPU is more powerful then the Shark DSP.

I think Volumio and MoOde have DSP correction but I have many different software solutions floating around my head ATM so I could be wrong...

If DSP is so important I'll find away to interface to your DSP hardware.

 

[JohnW]

That's interesting. With Dirac (software on PC) I have filters at 16/44 and 24/?(96 maybe)....I wonder if then you should only play those files on that filter, otherwise it gets up/down sampled??? What a palava, I just like to play tracks and I have a big mixture of sampling rates on both mp3 and flacs....no wonder it sounds better without DSP.

Well I cannot be definitive here as we only have two boxes, but both seem to use ASRC and process at a fixed 96KHz irrespective of input sampling rate. Even at 96KHz input I suspect the ASRC is being used (I could be wrong).

 

[DNM]

You say its impossible, then say how it would have to be done....I wouldn't mind doing all the measurements again with the filter in place, I'm quite good at it now.
I reckon there is a sinister reason why they don't allow it.

How can there be a sinister reason? Dirac creates a correction filter based on 9 sweeps. How can you measure the results without taking sweeps from the exact same 9 positions then comparing the before and after? Perhaps I should of said its impractical rather than impossible but good luck trying!