Home Theater Forum and Systems banner
Not open for further replies.
1 - 1 of 1 Posts

· Banned
5,013 Posts
Discussion Starter · #1 ·
miniDSP DDRC-22D Dirac Live® 24/96 Room Correction Audio Processor Review

miniDSP Website
Available Direct From miniDSP

by Wayne Myers


The recently introduced DDRC-22D, DDRC-22A, and just announced DDRC-22DA are miniDSP's entries into the room correction market making use of Dirac Live room correction technology. The Dirac Live technology has been available for PC- and Mac-based systems for years from Dirac Research, founded in 2001. The technology is also licensed for use in cars, smart phones, and other audio products. It is a thoroughly implemented and robust technology which uses mixed-mode filters, making time and impulse response correction possible. The miniDSP entries are two channel models, standalone products which can be inserted in a home or professional system at various points in the signal chain to implement room correction. miniDSP is a Hong Kong based company specializing in digital signal processing products for audio applications.

Room Correction Technology

Mention digital room correction (DRC) in the presence of a group of serious audio hobbyists, and you are sure to get a lively discussion started. Arguments will range from "It doesn't work," or "It is always a bad idea," at one extreme to "It performs magical transformations" at the other. The more reasonable middle ground is that, applied judiciously with proper speaker and room preparation where possible, and with attention to detail in the calibration stage, DRC will often provide a significant improvement to the listening experience.

Note: There is a freeware, command-line-driven software product called Digital Room Correction, or DRC. For our purposes, we will be using "DRC" to talk about room correction in general.

The basic idea is that imperfections in loudspeaker design combined with real-world acoustical challenges in the listening room lead to a less-than-optimal listening experience for many, if not most, listeners. DRC attempts to compensate for and correct those imperfections to make the listening experience as close to perfect as possible. Bear in mind that even the definition of "perfect" in this situation is arguable. A calibrated measurement microphone is used to analyze a test signal at various positions around the listening position, then the resulting data is analyzed using some proprietary process, and a set of correction filters is generated to be placed in the signaling path as pre-compensation to correct for those speaker and acoustical imperfections.

Some DRC implementations focus on the frequencies below 200 Hz and leave the higher frequencies which are responsible for the characteristic sound of the speakers untouched, or relatively so. Some implementations purposely do not attempt time correction, while others purposely do; there are arguments supporting either approach. Some allow the user to select the amount or severity of correction applied. Most advanced, or professional, implementations allow the user to apply a desired target curve to the resulting filter set. An ideal approach will never be agreed upon.

The result? With some preparation and care, and a little experience, most users get real and worthwhile improvement quite easily. Those last two words are key, the reason people buy and use DRC products is to achieve big improvement with minimal fuss. A lot of engineering goes into making those products as easy to use as possible and as foolproof as possible. Proper preparation and care, the details of which are beyond the scope of this review, usually involve dealing with acoustical matters and careful speaker setup.

The Dirac Live approach to DRC involves frequency and time correction, using mixed-mode filters to attempt re-creation of the original impulse response of the delivered signal. The Dirac Live team has been at this for some time, and their product is mature and well proven, with many fans and adherents.


The DDRC-22D which I received for evaluation came boxed with a power supply, a UMIK-1 calibrated microphone with cable and accessories for room analysis, a nifty collapsible mic stand, two 6-foot long Toslink cables, a USB cable for connecting the DDRC-22D to a PC running the Dirac Live software (only necessary during calibration), and a pair of rack-mount ears. All software is downloaded from the miniDSP web site.

The DDRC-22D unit is a sturdily built metal chassis and is compact enough to easily find a place in a typical equipment rack or on a shelf. The single control knob on the front is a little exposed, not unusual for gear of this type. Also on the front panel are indicators for active input, Dirac Live active/bypassed, and indicators for which of the four downloaded filter sets is active. I would also have liked to see a readout for gain.

The control knob gives front-panel access to volume, input selection, and filter set selection, but not the active/bypassed selection, which I ended up wishing for during the evaluation process. All functions can be accessed via popular programmable remotes, and a utility is included to help with the remote learning process.

The rear panel has inputs and outputs, in this case TOSlink, SPDIF digital coax, and AES-EBU on Neutrik 3pin XLR. This evaluation made use of the TOSlink and SPDIF ins and outs.

As indicated , the unit has rewriteable memory to accommodate four stereo filter sets. When connected to the host PC by way of USB, with the Dirac Live software active, the filter set can be selected from the PC and made active or bypassed.

The Dirac Live software makes use of an impressive and functional user interface that did just about everything I could imagine wanting it to do. The user can retake any of the individual measurements of a measurement set, create/save/load/apply custom target curves, save and load measurement projects and, of course, determine which filter sets are installed in which of the four hardware memory locations in the DDRC-22D unit. The application is set to work only with the supplied UMIK-1 measurement mic; there is no option to use another type of measurement microphone.



No code has to be inserted here.

Associated Review Equipment
  • Music Server: foobar2000, Yohng VST Wrapper, Reaper ReaPlug VST Plugins, Phenom II x6 1055T, 8 GB memory
  • M-Audio FireWire 410 Audio Interface (TOSlink out)
  • 4x2 Remote Control Digital Optical TOSlink Audio Fiber Optic Selector/Splitter by Specialty Audio Video
  • Onkyo TX-SR705 Receiver
  • Crown Xs500 Power Amp
  • OSD Audio ATM-7 Digital 7-Zone Dual Source Speaker Selector with Remote Control
  • Martin Logan ESL Speakers
  • Home Theater Direct Level Three Towers
  • NHT Model 5 2-Way Bookshelf Speakers
  • Polk PSW10 Subwoofer, used with the NHT Model 5 Bookshelf Speakers
  • Beyerdynamic MM-1 Measurement Microphone
  • Roland Quad Capture Audio Interface

Evaluation Approach

The Dirac Live DDRC-22D was used in various combinations of the following situations:
  • Three different sets of stereo speakers, one with a subwoofer.
  • Two different target curves.
  • Two different mic setup patterns, "my way" (1 analysis point) and "by the book" (9 analysis points).
  • Comparison to Audyssey MultEQ XT.
  • Comparison to a hand-tuned room correction example.
Having played a bit with mic setup patterns for DRC implementations like Audyssey MultEQ (XT and XT32), and with listening position measurements in general, I have settled on a simplified measurement approach for my own use. It is effective, accurate, and quick to use. This is used in the "my way" (1-point) measurement approach as a reference.

The back of the listening chair, a recliner with a flat back which extends to top-of-head level, is covered with a thick, plushy blanket which remains there during measurements and listening to reduce reflections from the chair back. Soundstage and image clarity are improved during listening. When measuring, the measurement mic hangs by its cable at the center-of-head position, a couple of inches in front of the blanket. But even a soft blanket is slightly reflective, so a rolled-up hand towel is placed under the blanket so that it bulges out just far enough to barely touch the microphone tip. Voila, no reflection, and the frequency response in the area of the listener's head varies very little. A single measurement is taken, which is very smooth, no peaks or dips from reflections, and that is it. Listening tests confirm that the measurement result, correction filters, and the listening result all track beautifully.

Note: Having the UMIK-1 hanging by its cord for measurements will introduce some measurement error, the mic being 90 degrees off its reference axis and reading 1 to 2 dB low at 9 kHz for a typical UMIK-1. This would lead to extra boost at 9 kHz in the correction filter set. My target curve has enough loss at that frequency that the brightness is kept in control.

This technique - "my way" (1-point) - was used for a reference with each of the three Dirac Live speaker setups. Of course, Dirac Live setup was also completed "by the book," and most listening was done with this filter set. Results of both approaches are included in the following analysis.

Ease of Operation

Software and hardware setup were very straightforward. The online documentation covers this all in detail. My evaluation unit required a slightly different approach, but instructions had been written out in an email and when I had questions or a minor problem along the way, support via email was close to instantaneous.

Getting all of the pieces of a complex audio system to work together, especially in the digital realm, can sometimes be a little daunting. This is no reflection on the DDRC-22D or the Dirac Live software at all, simply a reality of the audio world. In my case, I ran the output of the DDRC-22D through the digital input of another audio interface and had to get the audio interface to sync to the 96 kHz output rate of the DDRC-22D. This should not be a problem with most DACs or AVRs which would normally follow the DDRC-22D in a typical setup.

I had the Direct Live software running concurrently with Room EQ Wizard, foobar2000, and Reaper in various combinations on the same laptop, and all behaved very nicely together, not always the case with audio software. To me this is a sign of a mature and carefully crafted product in the Dirac Live software.

I was very pleased with the software interface. It was extremely intuitive, well laid out, easy to follow without having to refer to documentation. At the appropriate stage, it includes a 3-D diagram of the mic pattern to be used for the analysis measurements, from three different perspectives.

Each of the nine analysis measurements can be deleted and retaken if desired. The order in which the measurements are taken is fixed. Actually, any number of measurement points from one to nine may be used, and only the absolute position of the first is critical, so the adventuresome user can play with the analysis mic pattern at will. An on-screen representation of the three sweeps taken for each measurement allows the user to see any major anomalies or noise interference.

Once the analysis sweeps are completed, the user can either use the default target curve or create or import her/his own. Target curve creation with the Dirac Live user interface is mouse click easy. Double click to add a node, or as many as you wish, to the curve and drag it where you want it. It is that simple. Target curves can be saved and recalled at will, so there is no limit to the number one may work with, although most experienced room tuners rely on two or three at most.

A project, including all of the measurement data, mic calibration curve, and target curve, can also be saved for later reuse. With a single click, the user applies a target curve to the analyzed and averaged input data, and a filter set is generated. The expected response is shown along with the input data and target curve. Here you can see the averaged left & right analyzed signals, the target curve, and the expected result curve. An Impulse Response view can also be selected.

The filter set can then be drag-and-drop loaded into one of the four memory slots in the DDRC-22D. At that point, one is ready to hear the corrected result.

Analysis Mic Patterns

The Dirac Live analysis software allows for up to nine measurement mic positions for an analysis data set.

The first measurement position is special. The Dirac Live software uses this first measurement to determine critical timing and distance information. It must be located where the center of the listener's head would be located at the Primary Listening Position (PLP). The other eight measurement positions should, in my experience, form a pattern with left/right symmetry around that first central point. They tell the Dirac Live software about acoustical variations in the vicinity of the PLP.

There are two versions of this pattern shown in the Dirac Live software and documentation, one for a chair with a single listener and one for a sofa with several listeners. The engineers wisely chose not to try to accommodate broader seating arrangements, almost a sure recipe for disappointing results.

A minor point: the sofa mic pattern shown in the documentation is not left-right symmetrical. I recommend always using a pattern that is left-right symmetrical for best soundstage and image clarity. My tests all used the chair mic pattern.

Other users have mentioned preferring the sofa setup pattern to the chair setup pattern, that it gave a more pleasing soundstage result. My own experience with different DRC implementations has been that more widely spaced mic patterns yield a more diffuse soundstage and softer imaging, not my preference at all. I will, when I am able, run Dirac Live setup with a wider mic pattern and report on the result.

Are nine positions enough? The answer depends a little on your philosophical priorities and what you hope for the averaging process to accomplish. Are you trying to optimize your sweet spot with DRC or make it bigger to include more listeners? DRC implementations tend to perform sweet spot optimization quite well, the averaging process assures correction is for an "acoustically typical" ear location in the sweet spot. The sound at the sweet spot is improved and the listener still has some freedom to move his head around. Using the averaging process to include more listening positions, usually the reason for needing more than eight or nine mic positions, only means that none of the listening positions sounds as good as it could.


Following are the frequency response measurements...
  • Without Dirac Live
  • With Dirac Live (1-Point), 2 target curves for the ESL
  • With Dirac Live (9-Point)
It appears from analyzing these plots that part of the Dirac Live design strategy was, above 250 Hz, to correct with less severity, allowing the contours that represent a speaker's characteristic sound to remain while eliminating wider variations that detract from the listening experience. Below 250 Hz, the frequency response is flattened as much as possible.

This was seen with all three speaker examples, and strikes me as a perfectly-compromised approach. The mids and highs contain a speakers "voicing" signature. Below that you just want even response, no boomy notes and none that are shy or disappear.

It is also apparent that the more severe dips in response above 250 Hz have been left uncorrected, wise because these anomalies are pretty much untamable, varying widely within a small area, and are generally not heard anyway. The attackable dips below 250 Hz are sometimes eliminated completely.

All these details together speak of a well-thought-out and implemented design intended to correct that which can be corrected with minimum downside, and to do so in a way that is as musical as possible and respects the fact that many listeners like the sound of their speakers, and simply want to tame their environment a little. From what is visible in the measurements, and what I heard in listening tests, I say hats off to the Dirac Research engineers for a product excellently done.

The following screenshots are for the ESL speakers: using the default target curve with "my way" 1-point analysis, custom target curve with 1-point analysis, and custom target curve with "by the book" 9-point analysis.

The impulse response diagrams indicate a clear progression toward near-ideal, the by-the-book 9-point analysis, and very close matching between left and right stereo channels. The impulse response shown was determined by mathematical conversion. The impulse diagram is interesting in that it contains all frequency and phase information in one nice little waveform. It is a bit hard to interpret, but fairly easy to compare at a glance and see that one is "more ideal" than another.

Test Tracks and the Hair-Splitter Grading Scale
  • Atoms For Peace - Default
  • B-52's - Ain't It A Shame
  • B-52's - Revolution Earth (mono, for quick image clarity check)
  • B-52's - Vision of a Kiss
  • Beethoven - F. Reiner - Chicago Symphony Orchestra - Symphony No. 7 In A Op. 92 - 2. Allegreto
  • Cincinnati Pops Orchestra - Star Trek (the movie) - Main Theme
  • Joni Mitchell - California
  • Melody Gardot - Baby I'm A Fool
  • Mindy Smith - My Holiday
  • Nickel Creek - Ode to a Butterfly
  • Radiohead - Weird Fishes--Arpeggi
  • Radiohead - Packt Like Sardines In A Crushd Tin Box
  • Radiohead - Pyramid Song
The following scale is meant for hair-splitting. Grades 4 through 6 are all very close to the same, representing listening qualities that most users would be happy with, differences small enough that direct A-B comparison would usually be needed to hear them.

  • 6 = Superb
  • 5 = Excellent
  • 4 = Very Good
  • 3 = Good
  • 2 = Fair
  • 1 = Poor

Listening Tests: Frequency Response

All three of the speaker sets used for evaluating the DDRC-22D hardware were set up with little or no toe-in, so the wide listening angle called for some high-frequency correction (except for the Level Three Towers with their wide dispersion).

As the measurement plots show, Dirac Live was able to resolve and match the frequency response to the desired target curve with very little error. I like that the resulting frequency response was not exactly flat to target. One can see the sonic signature of the speakers showing through just enough to retain its individual character while getting rid of distracting variations. And with the NSM bookshelf, the rough area of integration with the subwoofer is nicely flattened out.

The default target curve suggested by Dirac Live has an almost straight downward tilt from +3 dB @ 30 Hz to -3 dB at 16 kHz. I prefer just a little more brightness than that, so my target curve was flattened from 600 Hz to 6 kHz, then rolled off to -3 dB at 16 kHz. With all three of the test speaker sets, the result was very listenable. The sweet spot allowed for plenty of head movement and regular listener position adjustments for comfort. Significant variations due to placement and acoustics appeared to have been compensated for, and the individuality of the speakers remained intact. Dirac Live seemed to be living up to its promises. Bear in mind that frequency response correction alone is the easiest of the tasks to be accomplished by DRC (in some cases the only one undertaken). The remaining categories to be discussed are much more difficult. But we will not minimize what has been accomplished so far, and accomplished very well.

The centered synthesizer on Ain't It A Shame stands out as it should, left behind in the mix by many speakers. Guitar tones, like those on Weird Fishes and Ain't It A Shame are accurately represented. Orchestral instruments, brass instruments, and vocals all were as they should be. I could find no instance of instrumental or vocal tonality which seemed less than accurate. Overall the sound of the Dirac Live correction had no sound to speak of, which is what you want to hear.

Listening Tests: Soundstage and Image Clarity

A great soundstage results from complex factors. Having dedicated countless hours to soundstage optimization and equalization, it is clear to me that EQ or DRC can be used to improve an already-good soundstage, but they cannot create something from nothing. Where a superb soundstage already exists, DRC implementations tend to degrade it slightly. It seems a reality that the finest soundstage performance is reserved for those who tweak their systems by hand endlessly, and that is as it should be. There should be some reward for fanatical attention to detail.

DRC is for a bang-for-buck type of audio system addition, a way to get great results with minimum bother. Dirac Live accomplishes that extremely well.

Two of the speaker sets were set up for the best possible image clarity and soundstage. The Level Three Towers were purposely set with zero toe-in, slightly off from their known best orientation, an additional challenge for the Dirac Live correction to overcome.

A subtle effect was noted with some of my Dirac Live results, one which I have heard with other DRC products, too, that I will refer to as a sort of mosaic effect. The best soundstage is seamless and cohesive. At times the Dirac Live soundstage seemed pieced together, like a mosaic of soundstage pieces that do not quite match up where they join. On Weird Fishes, the snare drum sound through the intro has a small field of reverb extending to its right. Without Dirac Live, that reverb sounds like a natural extension of the snare drum. With Dirac Live, in two of my three test cases, it seemed not to belong with the snare drum, almost like it was out of phase. Overall, the soundstage slightly loses its ability to totally convinced a listener of its natural reality.

This is not a major complaint, but it might matter to some listeners. The effect is subtle, and might not be noticed without direct A-B comparison. In my case it could be compared to the hand-EQ soundstage reference with only a couple seconds of switching time. As previously mentioned, the effect is a common artifact with DRC in general. I expect that many listeners would consider it a miniscule sacrifice compared to the more obvious benefits from using Dirac Live.

MartinLogan ESL

The ESL started out with the best soundstage and image clarity. Both were superb with no EQ or correction. The ESL had optional hand-tuned EQ to compare to, two bands and one shelf applied to mid and high frequencies, also delivering superb quality level.

Dirac Live "my way" (1-point) sounded much like my hand-tuned setup, only not quite as open and refined. Dirac Live "by the book" (9-point) gave an excellent soundstage, very close to the quality of the hand-tuned, even wider. I liked the extra width, but that mosaic effect was a detractor.

Image clarity suffered slightly with Dirac Live. The drumstick clicks on Weird Fishes went from pinpoint sharp with rock solid placement to softball-sized with position slightly uncertain in the soundstage. These were not major concerns, many would not consider them problems at all. Hair-splitting aside, the Dirac Live soundstage and image clarity on the ESL pair were fantastic, really fun to experience.

The width and distribution of the soundstage struck me on Default. In some ways it almost seemed better organized than the original.

Melody Gardot's voice on Baby I'm A Fool was a little larger with Dirac Live and had a slight tendency to wander. It is a very difficult voice for most speakers to localize. I enjoyed the way the orchestra was separated a little farther from her voice. Mindy's voice on My Holiday was also a little soft with a tendency to wander, and the strummed guitar at the beginning of Vision of a Kiss lost its punchy impact.

I think the lesson here is don't mess with perfection. Those who already have a first-rate soundstage, not easy to come across, are not likely customers for gear like this anyway.

But different listeners have different preferences, and there were some real treats to be heard with the same setup. The second movement of the Beethoven symphony never sounded better. With this recording I have gone back and forth on how well I like the micing and mixing. Using Dirac Live, it almost sounded as though the entire orchestra had been re-seated for sound. The Star Trek track also sounded marvelous with Dirac Live. Perhaps for mixes which are more dense in nature, like a symphony orchestra, Dirac Live provides separation of images in a way that speakers alone are not able to. I found myself wishing I could keep the eval unit just for listening to orchestra tracks with.

No code has to be inserted here.

HTD Level Three Tower

Starting out with the purposely less-than-optimally-set-up Level Three Towers, I expected to hear the greatest improvement with Dirac Live active, and that was the case. As the table of results indicates, the soundstage without correction was not bad, although nothing to be impressed by. Soundstage quality with Dirac Live my way (1-point), was much better, wider, more open, much more distinct. With by the book (9-point), it was better still, better organized and better balanced.

The degree of seamlessness was OK at all test settings. The ability to fool the listener into believing the music was live was not quite there.

Imaging was a little soft without Dirac Live, and went a little softer still with both of the Dirac Live setups. The four drumstick clicks at the beginning of Weird Fishes are an instant giveaway. Other imaging reference points are Cindy's locals on Ain't It A Shame and Mindy'a vocals on My Holiday. All were significantly softer and broader with Dyrac Live, and harder to localize. This is not a huge surprise. The ability of DRC to improve image clarity is limited at best, and here the imaging ends up still quite respectable along with a very nice soundstage. Considering the setup starting point for those speakers, the end result was more than acceptable, and would certainly have been in the excellent range with a better initial setup.

No code has to be inserted here.

NSM Model 5

The little two-way Model 5 bookhelf speakers are imaging lasers. Image clarity on the Weird Fishes drumstick clicks was superb, pinpoint sharp in all test cases, including Dirac Live. Their soundstage was quite good without and better with Dirac Live, widening and opening up. The final soundstage was not the biggest I heard with Dirac Live in that room, but was impressively nice.

And completely seamless. Dirac Live could do no wrong with the Model 5 bookshelves, and I wanted to just sit and soak in the sound.

It was delightful to have a test case where Dirac Live improved all listening categories with no downside. That made me wonder if the other speaker sets might also be capable of more with Dirac Live, given the time to experiment and tweak them further. I am sure the technology is capable of more than I had time to squeeze out of it, but then part of the idea is to get results quickly, so that is what I went for.

No code has to be inserted here.

Listening Tests: Dynamic Response

With the Level Three Towers, dynamics really stood out beautifully. On the intro of Vision of a Kiss, the centered, strummed guitar and bass should both be punchy and full of impact. With Dirac Live, that punchiness really stood out.

Along with the softened imaging from the ESL pair, Dirac Live softened the dynamic impact of those same instruments.

Listening Tests: Clarity and Detail

There also was no noticeable lack of detail or clarity with the use of Dirac Live. Orchestral spring plucks on the Beethoven symphony had an intimate, close-up quality. I always listen carefully to the string picking on all the acoustical instruments on Ode to a Butterfly. The closely-miced guitar, banjo, fiddle, stand-up bass, and especially the mandolin all contain so much detail that it is hard to take it all in. Application of Dirac Live took nothing away from any of these instruments that I could hear.

Female vocals, especially Joni Mitchell's on California and Melody Gardot's on Baby I'm A Fool, hand an energy, clarity, and intimacy that was very enjoyable. Dirac Live left all this perfectly intact.

Listening Tests: Audyssey MultEQ XT Comparison

The Martin Logan ESL speaker setup was also corrected with Audyssey XT, using a 7-point left-right symmetrical version of the 9-point mic pattern. The frequency response correction was very similar, but other differences were easy to hear.

The soundstage with Direc Live correction was more spacious, more open. The images making up the soundstage were more distinct and separate with more space between them. The Aydyssey XT soundstage was a little more run together. It had a slight edge over Dirac Live in terms of natural cohesiveness.

A fairer comparison would have been between Dirac Live and Audyssey XT32, which was not available for this test. I have heard numerous speakers with XT32 correction applied after having been set up for optimum soundstage and image clarity, and based on that experience I would say that the performance gap between Dirac Live and Audyssey MultEQ XT32 is small. Only an A-B test will tell for sure. I am hoping to procure an AVR with Audyssey XT32 for a direct comparison, but did not want to delay the review for that purpose.


I can enthusiastically recommend the miniDSP DDRC-22D as a truly Pro-Level DRC offering. The sonic ultra hair-splitters are probably more interested in their own hand-tuning techniques over DRC, but even they enjoy a break once in awhile. For most listeners, miniDSP's DDRC family provides a lot of tweaking flexibility while delivering top-notch results quickly and easily.

Dirac Live with the NSM Model 5 test case could do no wrong. The other test cases might have come closer to perfection with a few more tries. Remember that a better starting point gives better final results. There is no substitute for attention to detail when working with DRC.

But the main point is the bang for the buck that Dirac Live delivers, delivering great sonic results with minimum trouble at a reasonable cost from a company that stands ready to keep you happy.

1 - 1 of 1 Posts
Not open for further replies.