Minidsp Flex Review (Audio DSP)

[Dan B]

"On the other hand.
If I'm trying to provide evidence that Phase shift is audible I can't do it with sighted listening tests because everyone with any knowledge of psychoacoustics knows that my (like everyone else's) ear/brain is subject to unconcious bias. We can only prove a difference can be heard by doing it blind and controlled."

I think it depends on the size of the difference. For example, you wouldn't need a 'blind and controlled test' to hear and validate the difference between both tweeters in a stereo pair of speakers turned on, and both tweeters turned off. The difference in sound would be easily identifiable, and which way is preferable would be easily identifiable, without a 'blind and controlled test'.

"Most importantly - if I'm evaluating new kit I'm only doing it for *me*. I'm not going to try to convince others that what I (think) I hear is correct. That they should buy the same kit because I think it sounds great."

On the other hand, if some change you made 'sounds great', you might want to recommend it to others.

In my opinion phase coherency makes an audible improvement, especially if phase errors in the drivers are also compensated for, which can all be done with the miniDSP 2X4HD or miniDSP Flex (and with a measurement microphone, signal source, and scope).

But it's very clear that most people who have never heard correct phase response coming from a loudspeaker strongly resist the idea that it is audible. It is not 'mainstream', and it looks like it will be a while before it becomes mainstream, if ever. But like with Gene Dellasalla from Audioholics, I'm satisfied that the difference is audible and preferable -- for me.

I'm perfectly OK with the majority of people who never heard a phase coherent speaker insist that phase coherency makes no difference, say it's all in my imagination, and never bother to check it out.

I think we can agree that this topic has reached its end.

 

[antcollinet]

"On the other hand.
If I'm trying to provide evidence that Phase shift is audible I can't do it with sighted listening tests because everyone with any knowledge of psychoacoustics knows that my (like everyone else's) ear/brain is subject to unconcious bias. We can only prove a difference can be heard by doing it blind and controlled."

I think it depends on the size of the difference. For example, you wouldn't need a 'blind and controlled test' to hear and validate the difference between both tweeters in a stereo pair of speakers turned on, and both tweeters turned off. The difference in sound would be easily identifiable, and which way is preferable would be easily identifiable, without a 'blind and controlled test'.

"Most importantly - if I'm evaluating new kit I'm only doing it for *me*. I'm not going to try to convince others that what I (think) I hear is correct. That they should buy the same kit because I think it sounds great."

On the other hand, if some change you made 'sounds great', you might want to recommend it to others.

In my opinion phase coherency makes an audible improvement, especially if phase errors in the drivers are also compensated for, which can all be done with the miniDSP 2X4HD or miniDSP Flex (and with a measurement microphone, signal source, and scope).

But it's very clear that most people who have never heard correct phase response coming from a loudspeaker strongly resist the idea that it is audible. It is not 'mainstream', and it looks like it will be a while before it becomes mainstream, if ever. But like with Gene Dellasalla from Audioholics, I'm satisfied that the difference is audible and preferable -- for me.

I'm perfectly OK with the majority of people who never heard a phase coherent speaker insist that phase coherency makes no difference, say it's all in my imagination, and never bother to check it out.

I think we can agree that this topic has reached its end.

No one is resisting, they are only asking that proper evidence of audibility is provided. Sighted listening tests are not that.

 

[Dan B]

I am doing it as stated in the manual, output three sending left and right channels. Sound outputs when I am just sending the L or R channel then goes silent when they are combined(visually stops sending info on the EQ bars as well). I have had an SHD in this system and it worked fine, I am familiar with the software and have reset to factory several times.

If you are saying that the woofer output drops to zero when both L and R are combined, but is OK with either L or R by itself, it might be that one of the XLR to TRS cables that go to the input of the Flex has its + and - signals wired reversed, or some connector in that path is wired reversed.

Even if the problem is somewhere else in your system, you might be able to fix it by reversing the + and - signals on one of the cables or connectors on the input path to the Flex (but not both).

 

[KMO]

I think it depends on the size of the difference.

This what I just heard Jack Oclee-Brown of KEF say about their "phase correction" feature, on the latest Darko Audio podcast.

In the LS50 Wireless II, they found was very hard to say it was a perceptible feature (despite being theoretically great and making square waves come out looking nice) and hence worth including, although they did.

In the LS60 Wireless, they've found the difference is perceptible, and they put that down to the starting point being worse - it's a 3-way, and has a bigger group delay due to the MF/LF crossover. The white paper shows a (simulated?) group delay of 1.5ms for a MF/LF crossover, compared to 0.2ms for a HF/MF. And he reckons the bigger absolute delay being corrected makes it more audible.

He then speculated that that bigger group delay in low frequencies of passive 3-way speakers may be a reason for some people preferring 2-ways.

 

[withoutsuit]

Again it doesn't have 8 output channels, just like the DDRC. You can output 4 channels and do whatever independent processing on those 4 channels you want, just like the DDRC.

Your example would imply the use of 6 outlet channels (2 x TOSLINK = 4 channels, 1 x Coax = 2 channels).

I am not sure where the confusion lies. What are you specifically trying to do?

Michael

This discussion is useless (no offense). I think you would understand it, when you try it. It's a kind of brainfuck. My Setup is stereo (two channel), but I use three different delays to drive it. And again: I'm not dumb - I use the DDRC a long time and it can handle it. The flex digital not, but the Flex RCA should do it. The reason is, that always 1 Toslink and 2 Coax are paired and not independently assignable, as it is via RCA (e.g. 1 Toslink + 3 Toslink and 2 Coax + 4 Coax). So if you want to use Output 2 (coax) to drive the main speaker, you have to route the right channel to Output 2 and the left channel to Output 1, altough you don't use Output 1 physically (no wire attached). Thats the "pairlogic" of the digital version. So if you want to plug in a subwoofer in Output 1 (toslink) and put a delay on it, it automatically affects output 2 (coax) and therefore the main speaker. Brainfuck, I know...

Thats what I tried to point out.

 

[Dan B]

This what I just heard Jack Oclee-Brown of KEF say about their "phase correction" feature, on the latest Darko Audio podcast.

In the LS50 Wireless II, they found was very hard to say it was a perceptible feature (despite being theoretically great and making square waves come out looking nice) and hence worth including, although they did.

In the LS60 Wireless, they've found the difference is perceptible, and they put that down to the starting point being worse - it's a 3-way, and has a bigger group delay due to the MF/LF crossover. The white paper shows a (simulated?) group delay of 1.5ms for a MF/LF crossover, compared to 0.2ms for a HF/MF. And he reckons the bigger absolute delay being corrected makes it more audible.

He then speculated that that bigger group delay in low frequencies of passive 3-way speakers may be a reason for some people preferring 2-ways.

Did he say that the "phase correction" is done using FIR filters? That's about the only way to achieve net zero phase shift in a speaker system that has greater than 6 db/octave slopes in its crossover. Did he mention a square wave looking like a square on the acoustic output of the speaker system?

Either way, phase correction (or phase coherency) may be starting to creep into mainstream high end.

 

[KMO]

Did he say that the "phase correction" is done using FIR filters? That's about the only way to achieve net zero phase shift in a speaker system that has greater than 6 db/octave slopes in its crossover. Did he mention a square wave looking like a square on the acoustic output of the speaker system?

The answer is yes to both, except he didn't mention FIR filters specifically in the podcast (which is here, starting at about 1:11).

The LS60 Wireless white paper says:

Phase Correction, as found in LS50 Wireless II and LSX, compensates for phase distortion by pre-filtering the signal through a special type of all-pass FIR filter. This filter has the reverse phase characteristics to the phase error introduced by the crossover and compensates for the system’s phase distortion, ensuring sound at all frequencies reach the listener at the same time.

(And there's a bit of other stuff about phase correction working especially well in KEF speakers due to the coincident source of the Uni-Q, and with the symmetrical-and-close-to-the-Uni-Q bass drivers of the LS60, meaning delay effects don't creep back in too much when you move off-axis, due to relative distance to drivers changing. Although that's probably a bit overplayed - a normal speaker with vertically-aligned drivers has no issues on the horizontal axis there, so you're only really gaining on the vertical, and maybe even losing on the horizontal with the LS60).

 

[mdsimon2]

This discussion is useless (no offense). I think you would understand it, when you try it. It's a kind of brainfuck. My Setup is stereo (two channel), but I use three different delays to drive it. And again: I'm not dumb - I use the DDRC a long time and it can handle it. The flex digital not, but the Flex RCA should do it. The reason is, that always 1 Toslink and 2 Coax are paired and not independently assignable, as it is via RCA (e.g. 1 Toslink + 3 Toslink and 2 Coax + 4 Coax). So if you want to use Output 2 (coax) to drive the main speaker, you have to route the right channel to Output 2 and the left channel to Output 1, altough you don't use Output 1 physically (no wire attached). Thats the "pairlogic" of the digital version. So if you want to plug in a subwoofer in Output 1 (toslink) and put a delay on it, it automatically affects output 2 (coax) and therefore the main speaker. Brainfuck, I know...

Thats what I tried to point out.

Seems pretty clear you don’t understand how this device works.

What specific components are you trying to use? Can you explain why using output 1/2 to your main speakers and output 3/4 to your sub would not work? Your comments make it sound like you think each output is mono? You understand that one TOSLINK output has two channels (either 1/2 or 3/4), not one?

Michael

 

[antcollinet]

This discussion is useless (no offense). I think you would understand it, when you try it. It's a kind of brainfuck. My Setup is stereo (two channel), but I use three different delays to drive it. And again: I'm not dumb - I use the DDRC a long time and it can handle it. The flex digital not, but the Flex RCA should do it. The reason is, that always 1 Toslink and 2 Coax are paired and not independently assignable, as it is via RCA (e.g. 1 Toslink + 3 Toslink and 2 Coax + 4 Coax). So if you want to use Output 2 (coax) to drive the main speaker, you have to route the right channel to Output 2 and the left channel to Output 1, altough you don't use Output 1 physically (no wire attached). Thats the "pairlogic" of the digital version. So if you want to plug in a subwoofer in Output 1 (toslink) and put a delay on it, it automatically affects output 2 (coax) and therefore the main speaker. Brainfuck, I know...

Thats what I tried to point out.

Not sure what you are expecting here. The device has 4 output channels. Digitally this only requires two stereo outputs.

You have two optical outputs (outputting all 4 output channels)
Those 4 channels are then duplicated on the two coaxial outputs.

Why are you expecting the coax outputs not to be duplicats of the optical ones?

 

[MCH]

Well, if he has two subs with, say, toslink input and they mix the two channels at input (if such a thing exists), then yes, the flex digital could have a more limited functionality in this case than an analog equivalent.
Maybe this is what he means, that the two pairs of channels cannot be separated as straight forward as with the analog minidsps.

 

[withoutsuit]

Well, if he has two subs with, say, toslink input and they mix the two channels at input (if such a thing exists), then yes, the flex digital could have a more limited functionality in this case than an analog equivalent.
Maybe this is what he means, that the two pairs of channels cannot be separately as straight forward as with the analog minidsps.

Thats what I'm trying to explain. It's two Main + 4 Subs, grouped as two pairs (double bass array). Goal was to get a completely digital setup to avoid the D/A-A/D conversion of the ddrc24, as my subs also get a digital signal.

 

[withoutsuit]

Seems pretty clear you don’t understand how this device works.

I do understand the device, you don't understand my explanation / setup (post before). All I wanted to do, is to warn others, who want to replace a DDRC24 with the flex digital and think it can do the same with it in every aspect.

 

[mdsimon2]

I do understand the device, you don't understand my explanation / setup (post before). All I wanted to do, is to warn others, who want to replace a DDRC24 with the flex digital and think it can do the same with it in every aspect.

Only took you 4 posts to explain what you were actually trying to do (mains + multiple subs) , I agree that it won't do what you want it do to do. I think how the device functions is pretty clear but I guess we will have to agree to disagree.

Michael

 

[Dan B]

The answer is yes to both, except he didn't mention FIR filters specifically in the podcast (which is here, starting at about 1:11).

The LS60 Wireless white paper says:



(And there's a bit of other stuff about phase correction working especially well in KEF speakers due to the coincident source of the Uni-Q, and with the symmetrical-and-close-to-the-Uni-Q bass drivers of the LS60, meaning delay effects don't creep back in too much when you move off-axis, due to relative distance to drivers changing. Although that's probably a bit overplayed - a normal speaker with vertically-aligned drivers has no issues on the horizontal axis there, so you're only really gaining on the vertical, and maybe even losing on the horizontal with the LS60).

Thanks for the links. The white paper shows an improvement in the impulse response, but it's not quite symmetrical. They 'undo' the phase shifts in their crossover with separate FIR filters. With the miniDSP Flex, the crossover can be implemented directly with phase-shift-free FIR filters, and phase shifts in the drivers themselves can also be compensated for.

 

[amper42]

On the RME ADI-2 DAC FS it shows the resolution of the audio file playing on the front display. When I use the FLEX I don't see a way to display the audio resolution of the file playing on the unit. Is there another way to verify the resolution of a Hi-Res audio file with the FLEX? Searching the FLEX manual for resolution didn't bring up much except that it can support up to 96kHz without Dirac. Thank you!

 

[dublinthedog777]

I inverted my subwoofer with my balanced Flex, which sounds like what you need as well, I was getting inconsistent FR readings when playing crossover content. I've tried looking into it more but I'm not sure why, maybe it has something to do with a TRS adapter -> RCA cable into a sub?
I've got one of the official miniDSP cables on order, I'l

I really dislike this thing, wish I could return it.

 

[Sluraad]

I really dislike this thing, wish I could return it.

Even with the inverted sub oddity, I love my Flex. It's a little more manual than my Denon X3700h but I am far more satisfied with the results.

 

[poxymoron]

A quick update on the problems I've been having with my new Flex Digital. Looks like the problem is not with the Flex but with my Edifier speakers. I've now tried toslink and coaxial connections from my Flex to both my Edifier S1000db and my Edifier S3000pro speakers and neither type of connection will work with either set of speakers. I forgot I had an unused Evolve soundbar stored away so I dug it out and hooked it up to my Philips TV and the Flex works as it should do with the soundbar using the toslink connection. So the Flex is working, or at least the toslink on the Flex is working, I presume the coaxial works too but I've nothing else with a coaxial input to do a test. Edifier say the speakers only accept PCM format. Does anybody know if the Flex outputs PCM? I can't find anything on this in the Flex manual. I would have thought that it did. Tomorrow I'll swap out the soundbar for the Edifiers and try the toslink connections on both sets of speakers just to confirm that they do work. Next step is to get a Topping E30 DAC and try again. All a bit odd I think.

 

[Blew]

Any difference will likely come from two areas.

1 - Capabilities of the DSP software, eg Camilla vs Dirac (or REW/Parametric EQ if that is what you are using)

From what I can tell, CamillaDSP potentially has advantages here due to not being constrained by the hardware limitations of the Flex, eg when run on a Pi4. It doesn't support Dirac though, but I'd personally prefer to try room correction manually using REW anyway.

2 - Crossover capablity - Frequency response, filter rolloff, delays/phase etc in the Flex Vs Amp.

In this case the Marantz amp that's doing the crossover is doing it in the analogue realm. So I assume filter rolloff doesn't apply there. I don't know how it measures for frequency response, but if I kept this setup I'd look to replace it with something like the Denon X3700H that we know measures well.

According to Amir, we shouldn't be concerned with phase when listening to music in a room. Thoughts?

In other words - the hardware is just a vehicle for whatever DSP you are running. It is the software that will determine the effectiveness of the result. (within limits - eg sampling rate capability/number of parallel filters etc).

So then, given the above, and assuming the DSP software is equally capable, it sounds like there's little to no disadvantage in doing DSP in a streamer/player via a stereo DAC and multichannel amp which performs the bass crossover in Pure Direct when compared to doing DSP in a Flex. Or is there anything else the Flex can do better given that it's outputting the separate analogue channels directly? Just want to make sure that I'm not missing anything here.

 

[deafenears]

Does anybody know if the Flex outputs PCM? I can't find anything on this in the Flex manual.

Yes, PCM, at least that's what my DAC says (it's fed streamer to Flex Digital to DAC).