Jump to content
  • The Computer Audiophile

    Grimm Audio MU1 Music Player Review

     


    The Grimm Audio MU1 first caught my eye, here on the Audiophile Style forum. Several members of the community began discussing the product in early 2020. However, a couple years went by before I realized I needed to bring an MU1 into my listening room for evaluation. This delay, from recognition to realization, was helpful in that the MU1 has since proven itself around the world as a stellar digital source over the last couple years. It's nice to write about new products, but it's easier to write about products that have had kinks worked out. 

     

    Speaking of easier, the Grimm MU1 has been both really easy to use and at times caused me a little consternation. For 99% of listeners, the MU1 will be absolutely simple, and fit right into their audio systems nicely. For me, someone who has to try every option, with multiple other devices, and attempt to break the product while also figuring out how it fits into the wider audio world, the MU1's distinctive feature set was tough to harness at first. I spent a couple weeks, and a long video chat with Eelco Grimm, attempting to understand what, why, and how the MU1 does what it does. I know, it's a digital music source, it shouldn't be complex for someone who eats, sleeps, and breathes this stuff. Perhaps I over thought it, when I should've just put on some music and listened. 

     

    The Grimm MU1 in the simplest of terms, is a high end digital source with built-in Roon core and FPGA upsampler, that outputs digital audio over S/PDIF, AES, Toslink, or a proprietary Grimm LS1 link, as well as inputs digital audio and sends it through reclocker. The Roon core runs on an Intel NUC with i3 processor, running Tiny Core Linux. The i3 was selected because Grimm wanted to keep the MU1 fanless and keep electrical noise low. The best way to do that, in the mind of Grimm engineers, was to use the i3 rather than the i7. 

     

    The MU1, delivered to me for review, contained an internal 2TB SSD to store some of my music library. This is a nice feature that I recommend for all MU1 users. When I pointed the MU1 at my NAS, containing 350,000 tracks, it wasn't a good experience. The library was too big and using a NAS for this, didn't work nearly as well as the local SSD. Roon server on the MU1 often restarted as it scanned the NAS and as it analyzed the library. I've run into similar situations with Roon running on an i3 Nucleus, where the application didn't fail gracefully. Rather than just slowing down, it entered an endless loop of ingesting too much data and restarting. Roon's recommendation for such large libraries is to use an i7 based Roon core (link).

     

    Note: The MU1 can also be a Roon endpoint if the core is run on other hardware. 

     

    The Roon core running inside the MU1 is an important feature for Roon users, but it isn't the most unique feature. That goes to the way in which the MU1 receives audio from the NUC and how it process audio before sending it out to an external DAC. I didn't grasp the MU1's process until after discussing it with Eelco. To say the Grimm team went the extra mile on this aspect, is an understatement. 

     

    Grimm needed to get the audio from the NUC to its own FPGA, which appears as a sound card/device to Roon, before it was output to a DAC. Rather than any of the common methods of doing this, such as I2S or USB (routed internally), Grimm wrote its own DMA controller. Grimm uses the PCIe interface for transferring audio between the FPGA and the NUC/motherboard, for extremely fast and stable communication. This is not a trivial task, and is well beyond the skill set of almost every audio company. 

     

    With its FPGA Grimm wanted to improve precision when processing digital audio. Due to the lack of processing power in most DAC chips, Grimm believes they lack the precision required to process audio using the best filters. Thus, the team designed its own digital filters and placed them in the FPGA. According to Grimm, when discussing digital filters, "incredible precision is needed and a lot of creativity to find the best solutions." The part abut creativity is what gets me most excited and buttresses my own experience when listening through digital filters designed by very creative audio engineers and software developers. Grimm is absolutely an engineering first company, but also realizes creativity is required to develop the best audio products. 

     

    Recapping the above paragraphs for clarity, the MU1 runs Roon, outputs the audio from the internal computer to an internal FPGA with custom filters and is capable of high precision DSP. The audio is then output over a traditional interface, in my case AES, to a DAC. Seems straight forward, and to the end user it is very straight forward. Roon running on an iPad, or similar, can control the music and a web browser can adjust any of the MU1 settings if needed. 

     

    This is where I ran into some complexity. Roon offers it's own DSP engine that includes up/oversampling and convolution, the MU1 offers oversampling at 2fs and 4fs, and the DACs I used offer up/oversampling using a number of filters, or no filters, and in the case of the EMM Labs DV2, it resamples everything to DSD1024. Which combination sounds best? Anyone with a free month or two, can run through all the possible combinations and find the answer that's relative to only him/her. That is until software or firmware is updated or a DAC is changed, etc... Hopefully readers get the point. If one wants to really dig into the MU1, it can't be done in a vacuum and the results have to be considered with curiosity rather than accepted as facts that can be extrapolated to other scenarios.  

     

    During this review I used the EMM Labs DV2 and the T+A DAC 200 digital to analog converters. The DV2 offers zero filter settings. The DAC 200 has several filter options including NOS. 

     

    The evaluation process I used involved setting Roon's upsampling to 2x or 4x, setting the MU1 to no oversampling, setting the DAC 200 to any number of filters or no filter, then listening, then setting Roon to no upsampling, then the MU1 to 4fs oversample, then listening, then trying a different setting or DAC, etc... The number of combinations is enough to drive anyone crazy. According to Grimm, its filters and DSP are superior, so DSP shouldn't be used prior to its FPGA. That confidence in its own filters and product is to be expected. But, I have to try "everything." Plus, the NUC inside the MU1 has plenty of horsepower for heavy DSP, and I had to find out if more horsepower would give it a leg up over the MU1 FPGA and custom filters. 

     

    Switching between the oversampling options of the MU1 is as easy as a single click. The MU1 switches almost immediately, making comparisons pretty easy. Switching Roon's DSP involves several clicks and is far from instant once the setting is changed. 

     

    All this talk of up/oversampling is halfway understandable and logical. Then, I have to thrown convolution into the mix. Using a 65,000 tap FIR convolution filter, I can take my system to a nearly textbook in-room response. This convolution runs within Roon, on the i3 CPU. The up/oversampling can run in Roon on the same CPU or the MU1's FPGA. If Roon does upsampling, it takes place before the convolution filter is applied. If Roon only does convolution, the MU1 does oversampling after the convolution filter is applied. Plus, then the DACs are going to up/oversample, unless the DAC 200 is set to NOS, but does it really work in NOS without the sample rate delivered at its max 705.6/768? Anyone one can see, there are enough options to drive people crazy. As Scott Galloway says, people don't want more options, they want more confidence in options presented. That's what I attempted to do while testing and writing this review, but found it's not possible. The results are too specific. 

     

    Note: Grimm says to set Roon to output Native DSD, while the MU1 converts that Native DSD to DoP for output to the DAC.

     

    Grimm-MU1-resized.jpg

     


    Digging into my Listening Experience

     

    The system I used for evaluating the Grimm MU1 was a T+A DAC 200 or EMM Labs DV2 DAC, into a Constellation Audio Inspiration Preamp, Constellation Audio Inspiration Mono amps, into Wilson Audio Alexia V loudspeakers, all cabled by Transparent Audio. The entire system, with room measurements, can be seen here.

     

    Note: The T+A DAC 200 couldn't receive DSD via DoP from the MU1 without loud static. Both companies are looking into the issue. Grimm uses the standard 0x05/0xFA Marker, so that shouldn't be an issue. I tested the T+A DAC 200 by sending it DSD via DoP from an Aurender N20 over AES, and it worked perfect. I also switched between the Consumer and Professional S/PDIF status bit, via the Aurender Conductor iOS app, and the DAC 200 handled both perfectly. I will update the review once I know more. 

     

    I really like the Grimm MU1, and have come to see it as a device similar to that of a dCS upsampler. Both use proprietary upsampling to 4fs over a single AES wire, with the dCS going to 8fs over dual wire AES. The concept of both is quite similar in that they reduce the load on the DAC, use sophisticated algorithms and filter designs, and act as a music source. 

     

    I will never argue over one's sonic preference, but I believe using the MU1 with an NOS DAC could leave a bit of performance on the table because the MU1 only oversamples to 4fs (176.4 / 192). CD players in the 1980s used 4fs oversampling, granted without the sophisticated filters and clocks in the MU1, but the fact remains that 4s is pretty low if a DAC doesn't take that incoming data and oversample to a much higher rate.

     

    I listened to the MU1 set to 4fs oversampling and Roon's DSP disabled, and compared that to Roon set to 4x upsampling with the MU1's oversampling disabled. This was a great test to show the extremely high quality of the Grimm filters and precision FPGA based DSP. I preferred the MU1 oversampling by a long shot over Roon's 4x upsampling. 

     

    Listening to Chris Isaak's cover of Only The Lonely, made this very apparent. Isaak's voice sounded silky smooth, and sucked me right in. The acoustic guitar was lush, with a wide soundstage, neither of which Roon's upsampling could match. Staying with acoustic music, I played Dave Matthews Live at Luther College. On Christmas Song, the 4fs oversampling gave the acoustic guitars a much fuller body and silky sonic texture when played through the EMM Labs DV2 DAC, and to a lesser degree through the T+A DAC 200 using the BEZ 2 pure Bezier interpolator. Based on my many hours of listening through and experimenting with the MU1, I believe it sounds best set to 4fs oversampling while playing acoustic music. This is its sweet spot. 

     

    Audio that sounded great through the MU1, but with oversampling turned off via its web interface, was what I'll call piano music. I listened to Larry Karush's album titled May 24, 1976 and Tsuyoshi Yamamoto's album Midnight Sugar extensively, using many different up/oversampling options. When only the DAC's oversampling was enabled, transients were bolder and definition in the bottom octaves of the piano was better. 

     

    On the title track to Midnight Sugar, there was also more definition in the top end and the transients were less sharp with the MU1's oversampling disabled. Specifically, the transient at 2:25 into the track. With 4fs oversampling enabled, the piano can rip the listener's ears off. It's overly sharp. Without MU1 oversampling enabled on this track, the T+A DAC 200 with bez2 filter sounded fantastic, as did the EMM Labs DV2. When listening to piano music, I preferred the T+A and EMM Labs filters and algorithms on their own. 

     

    My experience with the MU1's oversampling on and off seems unique, although not abnormal, as listener filter preferences are commonly based on musical content in systems that enable filter switching (dCS, HQPlayer, etc...). Others who use the MU1, set it and forget it (at either 2fs or 4fs oversampling). In my system, with my music selections, the MU1's oversampling benefits were very music dependent. Acoustic music was fantastic while piano music wasn't. Fortunately, disabling oversampling still enables the listener to take advantage of the MU1's stellar clocking circuit and clean digital audio output to the DAC. What I heard with all music, once I settled on my preferred filter setting, was terrific through the MU1. 

     

    Roon Convolution to Grimm MU1.pngOne note of caution. I'm a huge believer in convolution for room correction. A 65,000 tap FIR filter can deliver stunning results and literally blow away the room correction solutions of yesteryear. When using convolution with the MU1 and its oversampling enabled, the sound was off. Something just wasn't right. The issue could be related to a number of things, including my own preferences or Roon's conversion to 64 bit float and back to 32 bit before outputting to the MU1's FPGA, or something entirely different.

     

    Running only the convolution in Roon and the MU1's oversampling disabled sounded fantastic. I can't explain the reasons why I heard what I heard, but believe this note of caution is warranted. I would appreciate other views on using convolution with the MU1 and its oversampling enabled. I could be an outlier.

     

     

     

    Grimm-Audio-MU1-back-resized.jpg

     

     


    Wrap Up

     

    The Grimm Audio MU1 digital music source is absolutely the right product for many listeners in our wonderful hobby. If perusing the MU1 thread here on Audiophile Style is any indication, most people set it and forget it, at either 2fs or 4fs oversampling. The concept and validity of an external up/oversampling device has long been proven highly effective in many audio systems. The MU1 takes this a step further by including a Roon core and an advanced FPGA working to extend the capabilities of Roon, to deliver a pristine audio signal out to a DAC.  

     

    The sound I heard through the MU1 was fantastic, when I configured it to my taste. We must remember that when using well designed digital filters and oversampling, there isn't a right or wrong. It all comes down to personal preference. On some music the MU1 4fs oversampling was clearly an improvement. On other music I preferred the MU1's oversampling disabled. I guess this is why Grimm, and many other very high end DAC manufacturers, offer listeners a choice. 

     

    I look forward to coming updates to the MU1 that will enable it as a Roon Ready audio endpoint and UPnP renderer, among other enhancements. The MU1's Linux platform is powerful and extremely flexible. In the capable hand's of Grimm's engineers, the future for the MU1 looks very bright. 

     

     

    Where to Buy

     

     

     

     

    Product Information:

     

     

     

     

    Associated Music:

     

     

     

    Complete Audio System Details with Measurements  - https://audiophile.style/system

     

     

     




    User Feedback

    Recommended Comments



    Two thoughts:

    1) A friend of mine says that the Tambaqui DAC with the Grimm MU1 is the best digital system he's ever heard.

    2) For a few years, Grimm has been touting an upcoming MU2, which would add a DAC.  As it's been some time, I suspect the Grimm folks are having some issues.  Did you hear anything about a MU2?

    Share this comment


    Link to comment
    Share on other sites

    5 minutes ago, Flashman said:

    Two thoughts:

    1) A friend of mine says that the Tambaqui DAC with the Grimm MU1 is the best digital system he's ever heard.

    2) For a few years, Grimm has been touting an upcoming MU2, which would add a DAC.  As it's been some time, I suspect the Grimm folks are having some issues.  Did you hear anything about a MU2?


    Here is info about the MU2, straight from Eelco Grimm. 
     

    https://audiophilestyle.com/forums/topic/58754-grimm-audio-mu1-server/?do=findComment&comment=1215325

     

    Share this comment


    Link to comment
    Share on other sites

    Appreciate this review. The MU1 is the most consequential audio expenditure in a long time. I spend a lot more time enjoying music than I have in a long time. 

     

     

    Share this comment


    Link to comment
    Share on other sites

    9 minutes ago, soupcon said:

    Appreciate this review. The MU1 is the most consequential audio expenditure in a long time. I spend a lot more time enjoying music than I have in a long time. 

     

     

    Great to hear @soupcon

    Share this comment


    Link to comment
    Share on other sites

    8 hours ago, FredM said:

    After testing various settings in and outside the MU1, you mentioned that you preferred the non upsampling (0fs) setting in the MU1 with piano music. Calling the transient in the Midnight Sugar track ‘overly sharp’ and even ‘the piano can rip the listener’s ear off’. It got me curious and also puzzled. I’ll try to explain: Although I have preferred the 4fs upsampling since the first day with the MU1 (and actually never touch it since), I tried if I could reproduce your findings. Well, long story short: no. On the contrary actually. I downloaded Midnight Sugar the album (great album btw, thanks!), and must say that I clearly prefer the 4fs setting compared with the 0fs setting (didn’t brother with the 2fs setting). This said, I’m not questioning your findings. But find it interesting what makes that things differ. And thinking more about it, I wonder if the way of testing could have anything to do with it.

     

    Hi @FredM this is fantastic. This is what it's all about. Read what I have to say, then listen for yourself. Never take what I say about my experience, with my ears and my system n my room, as the gospel. 

     

    Part of what this boils down to is which initial stage upsampling is preferred by the listener, that of the MU1 or that of the DAC being used. There's no right or wrong. 

     

     

    9 hours ago, FredM said:

    With upsampling enabled to me it just sound more alive, with more air around the instruments. Also smoother, in a positive way. The strike of a piano key is easier to hear and the tones have a beautiful decay (longer lasting than in a 0fs setting). In more complex pieces, multiple tones just ‘flow’ next to each other and fade out gradually individually. In the specific section (2.25) the piano key is struck hard (and recorded close by?), but the clarity and flow remain. In the 0fs setting it sound (a bit) harsher for me, with less ‘flow’. I would call it an obvious difference (easy to pick in a blind test). Or could the higher Dynamic range (15) of the specific track have something to do with it?

     

    What this seems to say is that the upsampling to 4fs in the MU1 is either better than in your DAC or preferred by you more than what's in your DAC. All good. 

     

     

    9 hours ago, FredM said:

    With countless tuning combinations possible (filter settings, DSP, convolution setting, upsampling inside the DAC, etc), it’s hard to pin point things.

     

    Absolutely!

     

     

    9 hours ago, FredM said:

    Perhaps double upsampling (in MU1 and T+A DAC 200) can lead to mixed results?

     

    It isn't double upsampling technically. It's upsampling in the MU1 to 4fs, and in the DAC 200 to its max rate. The first stage is done in the MU1, the rest in the DAC 200, versus all in the DAC 200. Different filters and algorithms should sound different. Both of us have definitely heard this :~)

     

     

    9 hours ago, FredM said:

    Besides difficulty in reviewing the product itself with the external variables, as a reader it will also be more challenging to get an understanding of the performance of the product (like I have at the moment 😀 ). Perhaps my hifi tuning knowledge is far below average, but I wonder to what extend an average listener/ reader uses these tuning options in practice (and apply specific tuning for specific music).

     

    I hear you loud and clear. I really wanted to be able to deliver some clarity, but it just isn't possible with so many options. The only thing I could do was try many of them, and describe what I heard. It's great that you and others have described what you/they heard with the MU1 as well. More data points are good. 

     

     

     

    9 hours ago, FredM said:

    Looping back to the Grimm MU1, I think it would be interesting if you can try out the Midnight Sugar track with perhaps the tracks above and see how they sound in a minimalist setting (Without the DSP, filters, etc, etc.) Keep it simple. I think that would be closer to (most) users situation, better to understand and more valuable insights. 

     

    The MU1 is heading out the door today, back to @Analog Audio of Minnesota

     

     

    9 hours ago, FredM said:

    Please don’t get me wrong, at first I liked your to approach as a welcome addition to existing reviews. But some results seem to differ quite from what others experience (by MU1 users and other reviews). As a suggestion, perhaps it’s an idea to simplify the evaluation approach for gear and perhaps have a separate paragraph for those who have interest and understanding in advanced filter/tuning settings?

     

    No worries. I know you have th best of intentions and it all comes from a good place. The fact that my results are different, is kind of nice to me. I feel very different from most other writers. Not better, just different :~)

     

    All feedback and recommendations noted. It's all valuable. Thank you.

    Share this comment


    Link to comment
    Share on other sites

    6 hours ago, The Computer Audiophile said:

    The fact that my results are different, is kind of nice to me. I feel very different from most other writers. Not better, just different :~)

     

    All feedback and recommendations noted. It's all valuable. Thank you.


    Thanks for your responses @The Computer Audiophile 👍 
    I guess I’m trying to understand why -per type of music- you seem to prefer different upsampling settings (0fs / 4fs). But fully agree, we just have different findings, all good.
     

    Great that you’ve had the opportunity to experience the MU1. When it arrives back at @Analog Audio of Minnesota it can do its magic and create smiles on a lot faces again 😀, nice to occasionally see new MU1 users in the Grimm MU1 topic.

     

    Share this comment


    Link to comment
    Share on other sites

    20 hours ago, FredM said:

    I guess I’m trying to understand why -per type of music- you seem to prefer different upsampling settings (0fs / 4fs).

     

    This is actually very common. Here is page 22 of the dCS Vivaldi manual. You can see why people may prefer different filters and type of music may be considered when selecting a filter.

     

     

    page 22 dcs vivaldi manual.png

    Share this comment


    Link to comment
    Share on other sites

    Quote

    On the title track to Midnight Sugar, there was also more definition in the top end and the transients were less sharp with the MU1's oversampling disabled. Specifically, the transient at 2:25 into the track. With 4fs oversampling enabled, the piano can rip the listener's ears off. It's overly sharp. Without MU1 oversampling enabled on this track, the T+A DAC 200 with bez2 filter sounded fantastic, as did the EMM Labs DV2. When listening to piano music, I preferred the T+A and EMM Labs filters and algorithms on their own. 

     

    Very strange recording balance to the piano; I've never heard a live instrument present so much 'power' in the transients - did they place a microphone right on top of the right hand strings?

     

    Is this typical for how TBM record the sound of the musicians?

    Share this comment


    Link to comment
    Share on other sites

    20 minutes ago, fas42 said:

     

    Very strange recording balance to the piano; I've never heard a live instrument present so much 'power' in the transients - did they place a microphone right on top of the right hand strings?

     

    Is this typical for how TBM record the sound of the musicians?

    This is the specific version I listen to. 
     

    https://www.impexrecords.com/midnight-sugar/

     

    C2C763EB-998E-4540-8372-63CD3E1E7B71.jpeg

    Share this comment


    Link to comment
    Share on other sites




    Create an account or sign in to comment

    You need to be a member in order to leave a comment

    Create an account

    Sign up for a new account in our community. It's easy!

    Register a new account

    Sign in

    Already have an account? Sign in here.

    Sign In Now



×
×
  • Create New...