Minidsp Flex Review (Audio DSP)

[abdo123]

@Ultrasonic I started collecting vinyl in the early 70's and still have a large (2000+) LP collection, many of which I can't find in digital formats. I would like to be able to listen to an occasional LP and introduce the kids to some long forgotten bands.

That's great! However if you're going to listen to LPs on a daily basis I would recommend something like the Parks Audio Puffin

Review: https://www.audiosciencereview.com/.../parks-audio-puffin-review-phono-stage.19795/

If you get the version with Toslink out then you never have to worry about things like gain or clipping. Also the puffin has a feature through which clicks and pops are removed in real time during playback.

 

[mdsimon2]

Interesting! I don't know if we can expect the Flex to behave exactly the same as the 2x4HD, but it's possible.

My only source is my Mac, and I configured Audio MIDI Setup so the Mac outputs 48 kHz. I figured this was likely the best way to set it up since CoreAudio does a very good job of resampling, and Dirac operates at 48 kHz. If I configure the Mac to other rates I can't tell any difference, so I don't know how valid my line of thinking is, but if the Flex behaves like the 2x4HD, then it looks like I'm set up for the better.

I think that from a resampling standpoint the Flex and 2X4HD should be close to identical as they use the same DSP.

To add some further quantification here are some measurements with a 2X4HD that show that the miniDSP resampler is better than the Mac CoreAudio resampler (but both are still very good).

44.1 kHz in to 2X4HD. -126 dB THD+N.

44.1 kHz resampled to 96 kHz in CoreAudio in to 2X4HD. -108 dB THD+N.

Michael

 

[Erici]

I'm new to FIR filters and just started dabbling in REW......Is there a recommended "How To for Newbies" somewhere on room correction without the Dirac option?

Here is a good ASR post on how to use REW:

Room EQ FAQ

Going down the rabbit hole of room EQ can be overwhelming and confusing, with critical hands-on information dispersed or hard to find. Here's a summary of targets and methods using REW, minimizing the fat and focusing on what's important. Olive/Toole (e.g. reference): Steady state room curves...

www.audiosciencereview.com

 

[JDubya]

Look at the multitone measurements. There the noise is at roughly -96 dB, so about 16 bits. Everything below is noise and intermodulation products. So we here have a device that can play one tone very well (18 to 19 bits equivalent), but when it comes to music, it only retains about 16 bits.

It’s purely the DAC performance.

I would not worry to much about that. You most likely won’t hear the difference anyway. 16 bits is still plenty, just not state of art.

Thank you, Voodooless.

Sorry, I'm obviously not versed in interpreting specs but would it be fair to say that the Flex effectively treats higher bit rate audio as 16-bit?
The other alternative I'm considering for a desktop system is the JDS Labs Element III. Based on pure dac performance (excluding the DSP features) using USB input and unbalanced RCA outputs alone, am I correct in interpreting from Amir's reviews of both that the DAC in the JDS Labs Element III outperforms the DAC in the Flex, or am I missing something?

JDS Labs Element III Review (DAC & Amp)

This is a review and detailed measurements of the JDS Labs Element III stereo DAC and headphone amplifier. It was sent to me by the company and costs US $449. As you see, the Element III has the treasured and much praised large rotary control which here, doubles as volume control and...

www.audiosciencereview.com

 

[levimax]

I have no issues with either sourcing or building an instrumentation grade linear amplifier for gain and impedance matching.

I built this "balanced transmitter"https://sound-au.com/project87.htm and used a 47K input resistor (designer approved of change) and feed the output to an ADC and do RIAA EQ using software and it works very well (keep the transmitter close to the TT with short interconnects and run the balanced as far as you like with no noise pick-up) and it very economical.

 

[PeteL]

Is there any point in high-res media, period? No.. how are your ears doing above 20 kHz nowadays?

It's a misconception that High Res purpose is to reproduce content above the audible band. Whether they are audible or not may be debatable, exemple to allow for easier non brick-wall filters with less ringing, etc. Most IMD graphs than Amir present are not 100% linear all the way to 20K and I suspect the're would be even more distortion in high frequencies if its dashbord was set to 44.1 instead of 192K, It would be good to see that too but he does IMD with 192K. Now, I believe it's not even that relevant here. In the end, Nyquist is about the theoretical limit of reproduction of a band limited signaL. This device is not about reproduction, it's about processing, the P in DSP. Processing is the main reason most recording and mixing engineer is standardised using Hi-Res Formats all trough the chain, until the final master that may be downsampled to RedBook CD formats or others. The more Data you have, when you start doing maths with these bits, the less rounding errors you get and the more precise every calculations are. Your 44.1 file may be theoretically perfect when it comes in, but when you start dividing it, summing it, phase shift it, well you get the point. Better to keep all the precision through all the calculations, and round it all at the end, It's always have been true in all applications of mathematics, and at least theoretically should be the case here too. Again, not sure of the audibility, but there are artifacts that can creep into what's theoretically audible to mathematical calculations, working with more margins should at least theoretically help with that.

 

[Ultrasonic]

Thank you, Voodooless.

Sorry, I'm obviously not versed in interpreting specs but would it be fair to say that the Flex effectively treats higher bit rate audio as 16-bit?
The other alternative I'm considering for a desktop system is the JDS Labs Element III. Based on pure dac performance using USB input and unbalanced RCA outputs alone, am I correct in interpreting from Amir's reviews of both that JDS Labs Element III outperforms the Flex, or am I missing something?

JDS Labs Element III Review (DAC & Amp)

This is a review and detailed measurements of the JDS Labs Element III stereo DAC and headphone amplifier. It was sent to me by the company and costs US $449. As you see, the Element III has the treasured and much praised large rotary control which here, doubles as volume control and...

www.audiosciencereview.com

I don't really see how they are alternative products, in that the reason consider the Flex is the DSP capabilities. If these don't interest you then there are loads of alternatives to consider.

 

[xykreinov]

Awesome! This is precisely what I wanted- SHD performance, but in a more compact and affordable package. Clearly, I wasn't alone. Very exciting product that stands out in a sea of mediocrity. No longer are you forced to use MiniDSP's fairly crappily performing 2x4 HD for smaller projects.

 

[xykreinov]

Now, I'd like to see if MiniDSP could make a new mobile DAC/AMP. My understanding is that the MiniDSP IL-DSP has largely worse performance than the Qudelix-5k. Having owned the IL-DSP, I was left disappointed with the lack of features / poor mobile support compared to the 5K, lack of power, and somewhat poor build quality- my IL-DSP irreparably broke within only a month or two of light use.

 

[JDubya]

I don't really see how they are alternative products, in that the reason consider the Flex is the DSP capabilities. If these don't interest you then there are loads of alternatives to consider

Fair enough. The crossover/subwoofer integration are the main DSP features of the Flex that I'm interested in. Absent the Flex, then I'd probably get the JDS Labs Element III and use a separate active crossover for the subwoofers. However, if the Flex can provide at least a comparable DAC to the JDS, then the Flex is the perfect small, single box solution for for my desktop. Either one would feed active speakers.

 

[amirm]

It would be great if @amirm can also measure performance for 44.1. All his measurements are in 48, 96, etc, as far as I noticed.

Dashboard is 44.1 kHz:

See the green writing at the bottom.

 

[voodooless]

Your 44.1 file may be theoretically perfect when it comes in, but when you start dividing it, summing it, phase shift it, well you get the point. Better to keep all the precision through all the calculations, and round it all at the end, It's always have been true in all applications of mathematics, and at least theoretically should be the case here too.

That’s what bit depth is for, not sample rate. More samples do not give you higher resolution in the passband . That’s what Shannon proved all those years ago.

And even if we’re true, the question was about high-res source, not the processing that would happen to it. If processing would benefit for higher sample rate, one could easily upsample it again for that purpose.

Mind you, some special filters do need high sample rate to be stable. These are things used in mastering, and in those cases, they will upsample the audio first by a factor of 4 or more. But these are very special cases. Depending on how much processing was done in the mastering, your track could have been through up and downsampling half a dozen times or more

 

[PeteL]

That’s what bit depth is for, not sample rate. More samples do not give you higher resolution in the passband . That’s what Shannon proved all those years ago.

And even if we’re true, the question was about high-res source, not the processing that would happen to it. If processing would benefit for higher sample rate, one could easily upsample it again for that purpose.

Mind you, some special filters do need high sample rate to be stable. These are things used in mastering, and in those cases, they will upsample the audio first by a factor of 4 or more. But these are very special cases. Depending on how much processing was done in the mastering, your track could have been through up and downsampling half a dozen times or more

I admit that I misread that you where strictly talking about the source file. Now yes, my limited example may have looked more Bit Depth relevant. But at the end of the day, in order to "up and down sampling half a dozen time" , well you will need internal processing that can do it, and assuming that mastering "special cases" is fundamentally different than what's being performed in DRC or Xover applications, is just an opinion. If not explain to me what's so special about these. We are manipulating harmonic content here, the bit depth of one sample does not give us harmonic content info so yes filtering digital manipulations don't rely only on bit depth any way you look at it, and I never said anything about giving higher resolution in the passband.

 

[Hemi-Demon]

So this device can't handle DSD/SACD iso playback, it has no MQA support (somewhat understandable), and also can't playback 24-192 files from Qobuz or HDtracks?

Seems like at least 2 of the 3 are huge features to leave out. Is this something that can be added in a firmware upgrade, or is the device limited by hardware and software coding capabilities due to DIRAC?

 

[voodooless]

I admit that I misread that you where strictly talking about the source file. Now yes, my limited example may have looked more Bit Depth relevant. But at the end of the day, in order to "up and down sampling half a dozen time" , well you will need internal processing that can do it, and assuming that mastering "special cases" is fundamentally different than what's being performed in DRC or Xover applications, is just an opinion.

It’s not just an option. Not all filters are the same. Specifically filters that are non-linear have aliasing issues and therefore are performed at vastly higher sample rate. For instance an overdrive filter. See here: http://blog.audio-tk.com/2011/02/01/electronic-the-purpose-of-an-oversampling-filter/

There are some cases though were linear filters also possibly benefit from higher sample rate, but that is due to stability. For instance a low pass with a frequency approaching Niquist. Funny thing: these happen due to bit depth issues. See https://www.quora.com/Why-does-a-di...e-specified-well-below-the-sampling-frequency . So adding bits for processing will also solve the problem.

We are manipulating harmonic content here, the bit depth of one sample does not give us harmonic content info so yes filtering digital manipulations don't rely only on bit depth any way you look at it, and I never said anything about giving higher resolution in the passband.

I have no idea what your trying to say here..?

 

[Ultrasonic]

So this device can't handle DSD/SACD iso playback, it has no MQA support (somewhat understandable), and also can't playback 24-192 files from Qobuz or HDtracks?

Seems like at least 2 of the 3 are huge features to leave out. Is this something that can be added in a firmware upgrade, or is the device limited by hardware and software coding capabilities due to DIRAC?

It can play 24/192 files but the sample rate will be downsampled to either 96 kHz (for a Flex without Dirac Live) or 48 kHz (for a Flex with Dirac Live).

My guess is there is zero chance of any of the features you're asking about being changed. None feel remotely 'huge' features to be omitted to me but obviously it comes down to what each individual wants or views as a priority.

 

[PeteL]

It’s not just an option. Not all filters are the same. Specifically filters that are non-linear have aliasing issues and therefore are performed at vastly higher sample rate. For instance an overdrive filter. See here: http://blog.audio-tk.com/2011/02/01/electronic-the-purpose-of-an-oversampling-filter/

There are some cases though were linear filters also benefit from higher sample rate, but that is due to stability. For instance a low pass with a frequency approaching Niquist.

I have no idea what your trying to say here..?

I'll let that go, apparently this is off topic. I find that strange, I was discussing the relevance of higher sampling rate processing, in DSP products, on a review trend on a DSP product that operate at 96k but the moderators don't like this discussion, at least here in this thread.

 

[AdamG]

I'll let that go, apparently this is off topic. I find that strange, I was discussing the relevance of higher sampling rate processing, in DSP products, on a review trend on a DSP product that operate at 96k but the moderators don't like this discussion, at least here in this thread.

Rules for Review Threads

New Rules for Review Threads Reviews of audio products are the crown jewels of our forum. In cooperation with kind members and companies that send in gear, and collective wisdom of the contributing members, we have created an incredible asset for audiophiles. As we cover nearly one new audio...

www.audiosciencereview.com

Go here for more general conversation about tech: https://www.audiosciencereview.com/forum/index.php?threads/minidsp-flex.28660/page-27#post-1086307

 

[testp]

View attachment 186103

As a lefty, it’s great that the control knob is on the Left hand side. Very unusual.

me too, but i hate it.. if it would be the first knob in the World, i'd like it, but im used to right side ones already...
nonetheless, i consider this my nr. one product for upgrade

 

[Bacchusoo7]

Can someone explain to me why this and the Onkyo TX-RZ50 that Amir reviewed today are in the same category? I know this was asked above, but no one seemed to answer it. Why is the Minidsp Flex considered AV? It doesn't do HDMI nor does it have the ability to decode any of the codecs in which AV generally comes. From the comments I see how it could be implemented into a music listening set up but I can't seem to figure out how one would use it as an AV processor in the same way one would use an AVR (sans the power amp section, etc.).

I'm not trying to be obtuse, I really feel like I must be missing something here. From my experience with the forums I know there are some insanely smart people here who understand all of this WAY better than I do. What am I missing?