While this review is intended to provide insight in to the use and subjective opinion of the end result of the MiniDSP with Dirac Live, it is not intended to be a full tutorial on how to tune with Dirac Live. I have linked various documents for that purpose. My discussion in that regard will only be intended as supplemental. And, obviously, as with anything relating to audio in a room, your results may differ from mine. Thanks for reading and feel free to ask any questions. I'll do my best to answer them when I can.
For some number of years now, MiniDSP has been producing some very affordable, yet powerful products for home and car audio enthusiasts alike. I tested an original MiniDSP 2x4 way back in 2010. Though, at the time it was not quite ready for full scale implementation for the car audio market, it was an exciting glimpse in to what the future may hold. Fast-forward to now, August 2019, and I am holding the future almost literally in my hands with the MiniDSP C-DSP 8x12 with Dirac Live.
On its own the C-DSP 8x12 is an excellent product and the $499 MSRP makes it even more enticing. There are the staples of what we in car audio (and home audio) have come to adore such as time delay, level adjustments and parametric EQ. With the 6 analog inputs and 12 analog outputs, itís more than capable of taking a 2-channel input signal and splitting that up to 3-way active (tweeter/midrange/midbass) + subwoofer and still have enough channels to incorporate differential rear fill for those who want to do that. Then thereís the inclusion of a remote for preset selection and subwoofer volume control. As capable as the MiniDSP C-DSP 8x12 is on its own, what really steps things up to the next level is the addition of Dirac Live (add-on cost of $400 for the license and UMIK-1 measurement microphone). The C-DSP 8x12 with Dirac Live takes the MiniDSP 8x12 v2 and adds the following features:
|miniDSP is proud to introduce the new flagship C-DSP 8x12 DL in-car digital audio processor. Dirac Liveģ room correction, a 400 MHz SHARC floatingpoint processor and 32-bit AKM converters with Ė107 dB THD+N specification deliver pristine high-resolution audio in any cabin environment. Comprehensive I/O includes six differential low-level analog inputs for low noise pickup, high-level inputs for connection to head units with power amplification, and stereo SPDIF and TOSLINK digital inputs. On the output side, 12 RCA outputs combined with a full eight channels of Dirac Live processing make the C-DSP 8x12 DL adaptable for active crossover systems as well as multichannel/surround installations with subwoofer integration.|
This review will focus on the combination of the MiniDSP + Dirac Live.
Dirac Live: What Is It?:
In a nutshell, Dirac Live (DL) is an automated room correction utility. It works to adjust speaker response issues in both amplitude and time based on the speaker response and room's influence. Depending on the application it can be used to measure and correct a basic 2-channel stereo or a 7.1 surround sound system. You use a microphone and measure the system response at various positions (depending on the goal) and DL will take it from there. The ideal result is a speaker that acts as true to a point-source (single object emanating sound) as it can; within the limitations of the speaker itself, of course. What makes Dirac Live work is complicated. I wonít pretend to be versed in the algorithms that are running behind the scene and Dirac themselves arenít willingly providing the engineering level details. However, they do provide a medium level explanation of the system on their website, quoted here:
|By using mixed-phase correction we can enable impulse response correction. A loudspeakerís impulse response affects staging, clarity, detail and all spatial aspects of the sound. Dirac Liveģ is unique in that it improves the impulse response throughout the listening area, not just in a particular zone. The Dirac Liveģ algorithm suggests a target response appropriate for your listening environment and speakers, which you can of course adjust to your taste using our simple graphical interface.|
In the above quote, it is important to key in on the ďtarget responseĒ aspect. DL can be set to target any curve you desire. Depending on your own subjective preference you can tailor the final sound to be a tad more bass heavy, to be a bit brighter, or any other sound you wish to achieve. This will be discussed in more detail later.
MiniDSP + Dirac Live in Car Audio:
In the sound quality side of aftermarket car audio world there are predominantly two types of users:
- Casual hobbyist who just wants better sound. For this user type, the easier the process the better. Thereís little desire here to continually swap gear in pursuit of the highest level of sonic nirvana. Instead, they are looking for a solution that nets them the best possible outcome the first time with little to no need for upgrading or fiddling with settings.
- Hardcore enthusiast who is prone to continual Ďtweakingí. This user type may also be prone to gear swapping just to ďscratch an itchĒ and/or to improve their own understanding and ability to build a high fidelity sound quality system.
In general, Type 1 users may not attain the level of fidelity as Type 2 users. This isnít to say it cannot be done. Moreso, the level of desire and dedication to pushing the envelope as technology increases just isnít there. If they have a 95% solution in year 2019, odds are they wonít pursue the 98% solution in year 2020. On the flip side, Type 2 users are always chasing the infinitesimal improvement. Making that 95% become 98%, then 99% Ö but never quite reaching 100%. (All percentages here are for example only, of course. ).
I classify myself as a Type 2 user. I enjoy the ďchaseĒ so to speak. This is what helps me learn and what helps me be a better tuner. Years back, I had experience with JBLís MS-8 and while it was a solid step forward, there were some things it did that I just didnít care for and ultimately I felt the imaging wasnít as focused and placed as accurately as it should/could have been. There have been other auto-setup systems along the way (such as Pioneerís auto-eq, Alpineís Imprint, etc). These devices have always done one thing well but never have fully encapsulated a ďone and doneĒ setup. For the most part, tonality was always acceptable but the imaging wasnít up to par with a manual tune. At least in my experience. So, naturally, Iím somewhat predisposed to turning my nose up at an ďauto EQĒ device as new ones arrive. Dirac Live has changed my mind. Based on my experience with the combination of the MiniDSP + Dirac Live, these two types of user could be sufficiently merged so that someone in Type 1 achieves 98% immediately while those in Type 2 may be happy with the immediate 98% results and choose to not further chase the never-attainable 100% goal.
With that said, it is important to understand that DL is not impervious to all issues. It isnít perfect. While my results were quite impressive, the user has to understand that DL cannot fix all issues. For example, if you have set up your subwoofer channel to be out of phase (acoustically) with your midbass, DL will not be able to fix this. Letís dig in to the setup of the MiniDSP and Dirac Live so we can see some ways to get the best results.
MiniDSP + Dirac Live Setup and Implementation Example:
The majority of sound quality enthusiast systems are comprised of a 2-way or 3-way active ďfront stageĒ along with a subwoofer(s). Either of which the MiniDSP is more than capable of handling, as previously discussed. My system consists of a 3-way active front stage with trunk mounted subwoofers in infinite baffle configuration. Iíll provide results based on this kind of system.
Before I get too far, let me establish this one important term: Speaker. A Ďspeakerí to Dirac Live should be what would typically be considered a speaker in home audio: a device that covers 20hz to 20khz (as best it can). While one can use DL to calculate time delays and target an individual curve per each individual drive unit in an active system, I have found DL to perform best when processing the system from a left/right Ďspeakerí level rather than processing each individual drive unit of an active system. Meaning, the Ďleft speakerí would be the combination of the left tweeter/midrange/midbass and, in my case, the subwoofer (with signal from the left side only, in this example). The same for right side. If you are running rear speakers then the same applies. MiniDSP also advocates using the MiniDSP before DL; letting DL essentially clean up the combined active-driver response as a single speaker.
To be honest, my first few attempts with the combo were trying. Thatís because I didnít read the Application Note(s) MiniDSP provides. I mean, who does that? As a Type 2 car audio guy (tinkerer) I am naturally inclined to set it up my way and try how I think it should be done. I could tell you what I did wrong but to keep confusion down I wonít. Instead, I will focus on how to do it right per the Application Note. Before going further, I suggest you at least give a cursory look through the Application Note to familiarize yourself with the basic pre-DL setup of an active system. We will discuss what we do with DL a bit later.
To summarize the steps in the application note, what you have to do is set your system up in the MiniDSP first. Then run DL. Set up your time delays, set your crossovers and do some EQ for each active channel. What youíre trying to do is get the best results you can with a little pre-DSP work so that DL can have a great baseline to start with. As I mentioned earlier in this review, if you are providing DL a scenario it cannot correct (such as ďholeĒ in the crossover between two active drivers) then you are starting from a faulted position already and the results will likely not be as good as they could be.
The MiniDSP in my car was connected as so:
Headunit --> MiniDSP Channels 1 & 2 Inputs
MiniDSP Channels 1-3 Outputs --> Left Amplifier (left tweeter/midrange/midbass)
MiniDSP Channels 4-6 Outputs --> Right Amplifier (right tweeter/midrange/midbass)
MiniDSP Channel 7 Output (Summed from Channels 1 & 2 Inputs) --> Subwoofer
The application note states to:
- Select drivers and design enclosure
- Set up the routing menu
- Equalize the speaker
- Add Crossover (and set time delay; provided in another linked application note)
#1 & #2 above are so specific to the end user that I wonít be discussing those. That said, you should, again, be providing DL the best baseline you can which means using the correctly sized enclosure and making sure your installation is up to snuff (no rattling door speakers, for example).
This gets me to #3: Equalizing the driver response. Now, to be honest, one should also take the time to pre-measure each active driverís response and then use broad strokes to flatten out any large peaks or anomalies seen. However, I didnít do this my first go-round because, frankly, I was too giddy at the thought of this new toy and I went straight to Step #4. But let me add some info about this step here and explain why I think itís still important to do. The Schroeder frequency. Home audio enthusiasts and scientists often argue that no correction filters (manually or automatically generated) should be applied above the Schroeder frequency (the point at which the wavelengths are equal to or greater than the room's dimensions). To quote Linkwitz's site:
|The frequency fs is also called the Schroeder frequency and denotes approximately the boundary between reverberant room behavior above and discrete room modes below.|
Step #4: Set up my crossovers for each speaker driver. With the varied amount of options out there in both install and driver, there is no one-size-fits-all solution to crossover assignment. However, the important thing is to set the crossover point and slope to protect your speaker for your desired volume level. A portion of Step #4 includes setting time delay. The application note linked above suggests doing so by using a tape measure. This is where I deviated from the application note. Instead of measuring each speaker driverís physical distance from my listening position, I ran DL and let it tell me what the calculated time delay values were. I then used those and plugged them in to the corresponding output channelsí time delay. I figure doing this might offer me a slightly more accurate time delay value vs measuring. A couple inchesí offset in time isnít a problem for bass but for tweeters and midrange blending, it does matter. I canít say with 100% certainty the results were better than vs using a tape measure to calculate distance/delay but itís easier. To do this, youíll need to configure the input tabs for DL to be assigned to each individual speaker. Once you do this, youíll need to re-assign the DL channels so they are correcting for the combined active channelsí response (i.e., left and right Ďspeakersí as I discussed earlier).
So, now that the basics are set, itís time to run DL. Below is a link to the User Manual for setting up DL:
Additionally, MiniDSP provides an Application Note on using DL:
While the manual should be sufficient for most users, Iíll add in some anecdotes about my experience:
I used the ďchairĒ position in DL because I am only concerned with the sound at my seat. Given our off-center positioning in the car, you canít really have both seats sound fantastic. You choose a compromise of either excellent sound at one seat or (at best) good sound at both seats. The latter is really only in regards to tonality; the imaging will never be the same from both seats using standard stereo efforts.
DL procedures instruct you to first measure at the main listening position (MLP). You then are instructed to measure a number of other times from various locations within the seated area. For what itís worth, I initially tried measuring from a single mic location. I found measuring from 9 positions worked best. I can only presume because of the MLP weighting discussed below. So go ahead and measure from a few positions like DLís software instructs you to do if you want to get the best results you can. The first step, the MLP measurement, is crucial because DL uses this initial measurement as well as later measurements to sort out the good, bad, and ugly. So make sure to place the mic at a location that your head will be for the MLP measurement. For later measurements, based on my experiments and with some slight understanding of what DL is doing, I recommend to place the microphone in a 1 x 1 foot square and measure somewhat randomly. It is worthwhile to note DL does *not* average the responses. DL isnít looking for a specific pattern here as much as it is simply using the results of each measurement to determine what needs to be fixed and what doesnít, using the MLP measurement as the baseline.
As previously mentioned, I tried to use DL to measure and optimize each individual driver to a specific target curve a number of ways. The truth is, it just didnít work well for me. My best results were met by using DL in stereo mode with the subwoofer getting half of the summed signal as it was summed by combining Left+Right in the routing tab. DL suggests this and my results back this as being the best method. But, again, it might be useful for you to first run DL on each individual driver to let DL tell you the time delay and level values to set for the corresponding output channels in the MiniDSP. Then re-run DL in 2-channel mode after youíve made those adjustments.
Now that we have done all this and gotten DL to optimize the left and right response itís time to listen. Audio is always subjective. Sure, we all understand thereís a goal of achieving a reference level playback system but the truth of the matter is the environment plays a huge role in what we hear that there isnít a single target curve that will give you that. For that matter, you may not necessarily want that. So, letís say youíve listened to the system results and you just didnít like it. You want more bass. Or you want less bass. Or you want more ďpresenceĒ in your highs. At this point you have a couple options: a) revisit your MiniDSP settings; crossover, level adjustments, EQ, etc. Maybe thereís something wrong in those settings that you missed. If youíre confident you have your active channels set up appropriately then b) itís time to reconsider the target curve you chose. Of course, the solution could be a combination of both. The luxury of having the MiniDSP + DL is that you can make adjustments in the MiniDSP and/or re-run DL to match a different target curve. This is where the provided remote with ability to switch between four presets comes in handy. Now, you can use different settings, store them in separate presets and call them up later. You can try different curves and find what you like. Or, letís say you want to have two different tunes: one for cruising with the windows down and one for sitting in your driveway with the windows up. You can set up a preset for each of these. And the best part about this is that, since DL is targeting a specific curve, you wonít feel the need to reach for the subwoofer level adjustment and crank it up, which would otherwise wreck the midbass/subwoofer blending. Even when youíre cruising at 55mph with the windows down, if the sub is pulling to the rear it lessens the enjoyment factor (at least for me).
When I was using DL I tried a few different curves. Initially, I replicated my "manual" tune curve and had DL match that just to see how well it could match what I already had via my manual tune. It did a really good job of matching the tonality curve. Then I used the "Wisdom" curve provided with the DL software and modified it a bit to have a more boosted lower frequency response. I actually preferred this tonality curve over my own personal "manual" tune's tonality curve. Here's a screenshot of the left and right measured response (light blue) vs the DL targeted - modified Wisdom - response curve (orange) and the final, DL optimized result (green):
Here's the same screenshot but of only the DL targeted - modified Wisdom curve (orange) and the optimized response (green):
From what I found, DL has an internal delay of 1.30 milliseconds. This isnít important if you run all your signal through DL. However, it is important if you were to use DL for only some channels and not others.
I donít have a passive system. Iíd be curious to see how well DL works with a system using stock passive crossovers in the front stage, especially when the midrange and tweeter are separated by more than an inch or two.
Target Curves: There is no "one size fits all" solution. I implore you to try a few different settings and try to factor in how your system is set up and what components are used to help you determine a target curve that is logical for your situation. For example, I wouldnít tell someone using a 4 inch wideband to target a response that is flat or has a rising response above 2khz because the wideband is beaming above this point and it's response is naturally falling off as you go further off-axis; shooting for a flat or rising response above the beaming point could introduce distortion if DL were to try to boost the signal or it may simply limit output volume in order to boost the upper end. I think the parts of the system should dictate the response. I don't have the empirical data to prove this notion, however.
My contempt toward automated DSP boxes has been quelled by the combination of the MiniDSP C-DSP 8x12 + Dirac Live. This combo wasnít the hardest thing to use but wasnít necessarily second nature, either. Still, once I got the process down and understood the way DL worked I was nearly able to achieve in 15 minutes with DL what took me hours to achieve manually tuning my system (maybe that just means I stink but I hope not). You can try different setups and, depending on the complexity of your setup combined with what youíre giving DL to work with (i.e., the pre-DL tuning aspect) you may not have as fruitful results. That said, my setup is very similar to the majority of aftermarket sound quality car audio systems these days.
The end result with DL for me is: The imaging and focus is excellent. Especially higher in frequency. The image is rock-stable and I didnít notice any wandering of the vocalists as they would carry notes higher or lower. The tonality, again, is up to the user but after a few iterations what I ended up with was excellent. In fact, using DL enabled me to find flaws in my own manual tune and improve them. The depth is impressive; my car seemingly lends itself toward good depth (not bragging, just relaying feedback I routinely get) but it's improved with this DL tune. Basically, everything above about 300hz was impeccable. Which, by nature implies below 300hz wasnít excellent. Ok. Why? So, midbass in the car is tough. The width of the car is right around the wavelength where most midbass tops out which means that what you wind up with is a series of peaks and nulls from each side. Some of these peaks/nulls complement each other when played in stereo and they help to fill in well. But you are still left with some that draw attention to themselves in one way or another. I spend the overwhelming majority of my time on tunes focusing on midbass and sub. Almost to a fault. So, I donít expect DL to nail this area but I canít give it a complete pass, either. So, letís talk about thatÖ There's still some left side bias towards the ~70-80hz midbass suckout. Every car has this suckout somewhere in this frequency range. Itís more noticeable especially in the cars that cross their midbass too low. Thatís why I cross my sub a little higher than most and let my sub fill in this null to some degree. As expected, DL cannot fix this issue but I believe with some additional time and experimenting with crossover frequency and slope I can get DL to at least mimic what I have with my manual tune. Additionally, the same thing that drives the 80hz null drives the other modes in the < 300hz region and, unfortunately, thereís just not much DL can do to fix it. In order to really do anything here, I think DL would need to measure in stereo and use artificial intelligence to determine how to adjust the EQ to balance the peaks/nulls. Of course, with the MiniDSP I can go back in after the fact and make corrections to the left and right midbass channels. And, if Iím being honest, these kind of issues wouldnít bother the large majority of people. I havenít heard a car with what Iíd consider ďperfectĒ midbass yet. So we are all fighting this fight. Therefore, Iím more apt to overlook this troublesome area but the fact that I still can do this better will not permit me to ignore it completely. While the left/right midbass imaging wasnít what it could be, the subwoofer tonality and blending to the midbass was excellent. No pull to the rear (in my final iteration). No overhanging, droning, resonant notes. Just clean bass.
Based on my experience with this combo thus far, I believe for a newcomer this combination of the MiniDSP + Dirac Live is a definite recommendation. Heck, forget newcomersÖ I have years of experience and the combination provided me with a tune that comes close to my own in less than 15 minutes. Even though it's not quite perfect, I am legitimately impressed with the results of DL in my car. It doesn't completely outdo my own manual tune with hours and hours of time spent but it gives it a run for it's money, no doubt. I'd have no problem recommending this but with the caveat that if you feed Dirac Live something it can't fix like an hole in the response or a poorly designed enclosure then you may not be as pleased. All of the components work to create the whole; if you have issues with the system before DSP then you can expect to have those after DSP. However, if you have taken the time to build and install the components in an optimal manner (as optimal as it can be in car audio) and take the time to pre-configure the speaker driver's responses and interaction before DL I am confident you will be very, very happy with what Dirac Live can do. And while I can't say with 100% certainty the MiniDSP + DL combo will be the end-all solution for everyone, I can say that I think it's certainly worth giving it a shot.
Some parting shots: