(Randolf Werner, 01/2023)
Today digital room correction is widely available in home entertainment equipment. Almost any AV Receiver offers some calibration method. Manufactures have developed many different systems. One of the most famous and advanced once is Dirac Live. However digital room correction is not limited to surround sound, if done properly also 2 channel Stereo can benefit a lot from it, especially in critical room and loudspeaker combinations.
In “Lowther Fidelio in my living room” I have described my efforts to optimize a Lowther Fidelio in my living room. So far the best the result could be achieved by applying a parallel notch filter to address midrange problems, using a Helmholtz resonator to address an outstanding 45 Hz room mode and using Yamaha YPAO to fix remaining issue in the bass section. My YPAO experience was rather disappointing, it could not detect the outstanding 45 Hz room mode and does nothing in the midrange nor treble section. So I considered YPAO more as icing on the cake than a real game changer. Anyway, I recently started looking into Dirac Live and it turned out to be a very big improvement. To me Dirac Live and Lowther Fidelio complement each other perfectly.
Even loudspeakers manufactures have started to bundle their products with Dirac Live. Quadral offers active loudspeaker AURUM ALPHA and AURUM GAMMA with integrated Dirac Live. ART OF SOUND bundles many of their horn loudspeaker with Dirac Live running on miniDSP DDRC-24. Bohne Audio combines their 2-way active speakers utilizing huge ribbon tweeters with a miniDSP SHD acting as crossover and Dirac Live processor. Dynaudio’s new active streaming loudspeakers Focus 10, 30, 50 are Dirac Live enabled (limited or full bandwidth license must be purchased separately from Dirac Live).
The equipment I used consists of the following components:
There is growing number of options to run Dirac Live room correction in a very wide price range. Here is a small list of the more affordable options you may want to consider (with no claim to completeness):
The Dirac Live Calibration software is the same in all of these cases, however it is not so easy to choose the most suitable Dirac live processor option. In my case a Yamaha R-N803D with Dirac Live instead of YPAO would be the ideal solution, affordable, solid build, good quality and supporting a full feature set. Anyway, there is no such device yet available on the market. Currently most Dirac Live implementation are available in AV Receiver products. Only a few 2 channel Stereo amplifier offer Dirac Live. Alternatively, you can consider adding Dirac Live to your existing equipment via Dirac Live Processor software running an PC or one of the miniDSP products.
To evaluate if Dirac Live is of any benefit in my environment at all I used the Dirac Live Processor for Windows with a 14 day free trial license. This is the most cost-effective starting point, PC and UMIK-1 were already present, so I was able to run a 14 day trial without any extra costs. I strongly recommend starting that way. The results were so impressing that I started looking for a permanent solution. The Dirac Live Processor for Windows is great if all your sources come from the PC, but for an every day usage it doesn’t meet my usability expectations.
The Arcam SA30 and JBL SA 750 are probably the closest match to a “Yamaha R-N803D with Dirac Live”. Anyway, I decided against it. The build quality doesn’t match my expectation for a device in that price range. There seemed to be a lot of software issues in the past. Arcam Music App looks very basic to me. Some minor features (second speaker pair connectors, tone controls, Bluetooth) are also missing.
The Pioneer VSX-LX505 is a bargain, you get a whole lot of features for your money. It has recently become available in the EU. There are a lot of positive reviews out there (but also poorly measured on www.audiosciencereview.com). At the end it comes with some compromises typical for most AV Receivers, especially in the power amp section. Fortunately, Lowther’s are very easy to drive, (very high efficiency, 8 Ohm and uncritical impedance curve). The NAD T 758 V3i is probably not much different (but also poorly measured on www.audiosciencereview.com). The Denon AVC-X3800H, AVC-X4800H and Marantz Cinema 50, 40 might be an alternative in case you prefer a Class A/B amplification (audioscience reviews: Denon AVC-X3800H, Marantz Cinema 50, Denon AVC-X4800H). Arcam AVR 5 and 10 are other options when looking for Class A/B amplification (audioscience review: Arcam AVR 10).
Any of these devices will resample the input to some fixed frequency to perform the DSP processing regardless of the input sampling rate, many of it probably to 48 kHz, a few perhaps to 96 kHz. This is common practice in any DSP solution and should be good enough for the audible frequency range. Anyway in measurements you can see the effect, e.g. in a sharply dropping frequency response at 20 or 40 kHz when DSP processing is enabled.
In general I would recommend to checkout technical reviews on Audio Science Review and Soundstage! for finding an appropriate device. Both provide very detailed measurements. Especially look out for results related to the digital / DSP mode of the devices, since this is what you actually will use when running Dirac Live. Unfortunately often the sampling rate and DSP filter capabilities of a device are not obvious when operating in Dirac Live mode. Since Dirac Live will operate in the +/- 10 db range some extra dynamic headroom is welcome to safely avoid distortion of the signal or details getting lost in the noise floor.
Since I am still very happy with the Yamaha R-N803D I decided to go for the miniDSP DDRC-24 for now.
The measurement microphone should be Dirac Live certified and of reasonable quality. Each miniDSP microphone comes with an individual calibration file at 0 and 90 degree. Less expensive microphones like the ones typically bundled with AV receivers may be less accurate. The following graph shows the individual calibration files of my miniDSP UMIK-1 at 0 and 90 degree and a generic calibration file for a NAD microphone:
In case of question, I would recommend to use the UMIK-1 or UMIK-2. Pioneer and Onkyo support using the microphone which comes with the product for Dirac Live. Denon and Marantz require to use a UMIK-1 for Dirac Live calibration.
In Germany HiFi-Selbstbau offers an affordable microphone calibration service. The following graph compares the original miniDSP calibration files of my UMIK-1 at 0 and 90 (blue, red) degree with calibration files for my UMIK-1 generated by Hifi-Selbstbau:
“VON WEGEN GENAU | Muss ich meine Messmikrofon kalibrieren lassen?” from Frank Rohde is a nice video with more background information on microphone calibration.
There are lots of reviews and extensive documentation available on the miniDSP DDRC-24 and the miniDSP 2x4HD (identical hardware but without Dirac Live license). Currently it is the most cost-effective way to integrate Dirac Live as standalone hardware into your system. Early revisions used a AKM AK4626AVQ A/D and D/A converter (2xADC 96 kHz, 24Bit, 102db + 6xDAC 192 kHz, 24Bit, 106db), my version (V2 R1.01 from Apr 12, 2021) uses a Burr-Brown / Texas Instruments PCM3168A (6xADC 96 kHz, 24 Bit, 107db + 8xDAC 192 kHz, 24 Bit, 112db). The differential voltage output of the DAC directly drives the unbalanced RCA output via a low noise NJM2060 operational amplifier. Both revisions utilize an AD ADSP-21489 DSP. Internal processing is 32 bit with sampling rate of 96 kHz, but only 48 kHz in Dirac Live mode. Inputs are digital USB and toslink as well as analog unbalanced RCA (2 or 4 Vrms). Digital inputs accept up to 192 kHz sampling rate. Outputs are 4 RCA channels (typically L+R and eventually subwoofer). It can be configured in many other ways as well, e.g. as a freely programmable DSP by switching to miniDSP 2x4HD mode. 4 filter presets can be stored on the device. Via an optional infrared remote you can switch between the 4 presets, enable/disable Dirac Live, control the output volume (0.5db steps), mute the volume and switch between the 3 input sources. The miniDSP can also learn control codes from other infrared remotes in case you want to control it from another remote. It is configured via the miniDSP plugin or the new miniDSP device console running on a PC connected via USB to the DDRC-24. The standard Dirac Live software is started from within the miniDSP plugin or the new miniDSP device console to perform the Dirac Live measurement and filter design. It comes with a simple 12 VDC external power supply.
The DDRC-24 can be integrated into the existing environment in different ways, e.g.
In case you consider to modify your integrated amplifier for integrating the DDRC-24 I can recommend two sources:
Option 1) works with any setup but since I mostly use the streaming capabilities of the Yamaha R-N803D it was not very appealing for me. Option 3) doesn’t work with the R-N803D and probably many other amplifiers with (partially) digital pre-amp. The R-N803D has 2 line REC-OUT and input terminals, but you cannot use it as a tape loop. They will only output the signal source that is also actually passed to the power amp. I studied the service manuals in detail and I think it is possible to
However, tweaking the R-N803D turned out to be more complicated as expected, especially due to
Therefore, I decided to take a different approach for now:
Since I have an idle AX-590 I use the REC-OUT terminal of the R-N803D to pass the signal to the DDRC-24. The DDRC-24 output is passed to the AX-590 CD input terminal. This way I can use R-N803D as pre-amp and the AX-590 as power amp. This works very convenient since the AX-590 also has infrared remote control, which reacts on the same codes as the R-N803D. Therefore, I can use the R-N803D remote control to also control the AX-590 volume. The DDRC-24 runs with fixed master volume level. Since the DDRC-24 can learn remote codes I could even trigger the Mute functionality via the R-N803D remote control. Last but not least the AX-590 has switched AC output terminals which can be used to drive the DDRC-24 power supply. The power on delay of the speaker protection relays is long enough to allow the DDRC-24 to startup first and avoiding any noise when doing so. So, the end user experience is almost exactly the same as with the R-N803D alone, you just have to switch on/off two devices instead of one.
The power amplifier sections of the AX-590 and R-N803D are very similar, partially even identical with identical power rating (2x100W, 8 Ohm, 20Hz-20000Hz, 0,019% THD) although they are 22 years apart from each other:
I could not detect any audible difference between the 2 amplifiers in a direct A/B comparision.
Later on, I switched to a different approach. I added a Yamaha WXC-50 streaming pre-amplifier (running in player / fixed volume mode). The WXC-50 has almost the same streaming capabilities as the R-N803D but only offers one RCA analog and one Toslink digitial external input, which is sufficient for my use case. This enables me to route all my digital and streaming sources via the WXC-50 to the DDRC-24 Toslink digital input and it’s analog RCA output to the R-N803D (running in Purce Direct mode). This way I can avoid the unnecessary extra D/A and A/D conversion and get rid of the second AX-590 amplifier. The DDRC-24 is powered on/off automatically by the R-N803D via a master/slave power strip and the WXC-50 is switched on/off by the R-N803D via its 12V trigger input. By default, WXC-50 and R-N803D use the same remote control IDs, however one configure both devices to use an alternative remote control ID set.
In general, I like the DDRC-24 a lot, however there are a few things I don’t like:
Dirac Live always uses multi point measurement, which combines the measurement results from 9, 13 or even 17 microphone positions. For all measurements I used the 13 positions setup. You need to decide whether to point the measurement microphone directly towards the speaker base and use the 0 degree microphone calibration file or to point the microphone to the ceiling and use the 90 degree microphone calibration file. For surround system measurements usually the microphone in pointed to the ceiling to capture the sound equally from the speakers in all directions. For two channel Stereo systems it can be beneficial to point the microphone towards the base of the 2 stereo speakers, therefore I used this setting for my measurements. After creating a Dirac Live filter, I verified the result using REW. For the first 3 measurements I used
The upper 3 curve show the result without Dirac Live and the lower 3 curve the corresponding result with Dirac Live (all 1/12 octave smoothing):
We can see that Dirac Live results is very flat (following the target curve with an intended slight decay) and almost identical for all 3 setups. So, from a frequency response perspective neither the parallel notch filter nor the Helmholtz resonator are necessary anymore. However, if we compare the waterfall diagram of the Dirac Live result with and without Helmholtz resonator
we can see that Dirac Live cannot fix the poor reverberation time at 45 Hz in contrast to the Helmholtz resonator. Therefore, I removed the parallel notch filter from my setup but kept the Helmholtz resonator.
In 2020 Dirac Live Bass Control introduced multiple independent subwoofer support to further improve the low frequency result. Most recently in 2023 Dirac Live Active Room Treatment (ART) was released which utilizes all channels of a multi-channel setup to generate anti sound to improve poor room reverberation times below 150Hz. A short demo of the generated anti sound can be found on the CES 2023 demo. This may make my Helmholtz resonator obsolete but would require a multi-channel setup.
So far, I used the default Dirac Live target curve and automatically calculated correction frequency range, which sets an upper limit for the correction range at 13Khz in my case. To see if we could improve the high frequency rolloff I adjusted the upper limit to 16 and 20 Khz. The graph uses 1/6 octave smoothing and shows the default Dirac Live target curve and the REW measurement for an upper correction limit set to 13, 16 and 20 Khz:
We can see the slight linear decay of 5 dB in the default target curve from 20Hz to 20 kHz. The measured result follows very closely the target curve. The high frequency rolloff due the Lowther drive unit limitation and of axis measurement can be improved as well. I noticed no negative side effects (e.g. DSP artifacts) of doing so and therefore decided to set the upper correction limit to 20Khz. The effect of this is very subtle and not easily audible for me, since these very high frequencies (often also called “air” or “sparkle”) are more felt than really heard. The resulting spectrogram is:
The default Dirac Live target is already rather flat. Often people emphasis the bass section a little bit more. However after listing a long time to various music and various volume levels I decided to do the opposite and use a flat target curve below 400 Hz. The graph shows the default and my modified target curve and the measured final result (1/6 octave smoothing):
There is no objectively best target curve, in depends on personal preferences and listening volume. The graph shows some target curve examples (Harman, NAD, Dirac Live default, mine):
In the latest version of the Dirac Live mobile app a so called “auto target curve” feature was introduced. The idea is to not just strictly follow a given target curve but to instead auto generate a specific target curve which tries to eliminate the problems caused by the room but preserving more of the loudspeaker characteristics. Since the mobile app version cannot be used with miniDSP and the feature is not yet available in the Windows version I could not test this feature so far.
For a good stereo imaging it is crucial that the signal from left and right loudspeaker are correlated as closely as possible (in time and frequency domain). The next graph shows left and right measurement without (bottom) and with my final Dirac Live filter (top) (using 1/12 octave smoothing):
We can see that left and right channel are now more closely correlated in the frequency domain.
As one can expect from the measurements the Dirac Live effect on the sound is very obvious. Especially the midrange section is very different and significantly better than using the parallel notch filter. Stereo imaging is improved as well and finally the bass section is a little bit more controlled. So far with the combination of Lowther Fidelio, Helmholtz Resonator and Dirac Live I could achieve the by far best result in my living room.
I finally measured the frequency response correction of the DDRC-24 with Dirac Live filter enabled from USB input to line output of the DDRC-24 for left and right channel:
We can see Dirac Live equalization is limited to at most +/-10 db, which is fully used to compensate a) the high frequency rolloff and b) a drop around 200 Hz. Although these 2 corrections may look like the most important / outstanding ones they are not. By carefully listing they are audible with specific music, but the corrections in the midrange are far more important and immediately audible with almost any kind of music.
Last winter my father passed away. There were 2 loudspeakers in the estate I build for him about 35 years ago. Since they were still in good condition, I decided to restore them by replacing all capacitors in the crossover and finally did some measurement with and without Dirac Live. The loudspeaker uses a hexagonal pillar bass reflex cabinet with 2 KEF B110B SP1057 bass/midrange drivers and a Dynaudio D-28 dome tweeter. It is a 2,5 way design with one B110B in the foot of the pillar working up to 220Hz, one B110B at the top of the pillar with a reflector cone working up to 3500Hz and a Dynaudio D-28 tweeter. The crossover is 2nd order butterworth network design.
Also, this type of speaker could be significantly improved in my room using Dirac Live: