Details concerning my DIY instrument and the room it now inhabits can be found at viewtopic.php?f=21&t=13444.
About two years ago, when my HW instrument was still located in a typical living room, the various Forum and personal communications of Jan Loosman and Patrick (pat17) helped me to greatly improve the amplified sound of my organ through the incorporation of ARC2 digital room correction. I also noted at the time Patrick’s report concerning the Dirac Live alternative for digital correction to computer sound output.
After moving my HW organ to a much larger space early in 2014, I struggled to establish an ARC2 correction that was genuinely satisfying. The struggle was exacerbated when I replaced one 12” subwoofer with a pair of 15” Rythmik units. In late December, 2014, I decided to take advantage of the Dirac Live 30-day free trial offer to see how it would perform. As in the past, Martin Dyde kindly played a key role (on 26 December, no less!) in pointing me toward using ASIO4All as the internal multichannel driver that I needed in order to connect all 6 of my HW audio outputs to the Dirac Audio Processor.
The first step in using either ARC2 or Dirac Live is to take measurements of the computer’s audio system output through a microphone connected to the computer. It took three very different microphone placement layouts for me to find a successful scheme for taking measurements for Dirac processing for my space. I admit that I never used the final, successful Dirac microphone placement scheme in taking ARC2 measurements, which may explain a portion of the huge improvement in sound that I have obtained using Dirac Live instead of ARC2. During my efforts with Dirac measurements, I was surprised to learn that the Dirac Live “auditorium” measurement mode only takes one set of speaker readings, which approach turned out to be woefully inadequate. In the end, though, I am thrilled with the results that I have obtained using Dirac Live:
1. Frequency response (Dirac’s term: “Magnitude response”):
In the past, ARC2 definitely improved the evenness and clarity of bass response, but particularly for some 16’ stops in some sample sets (especially for “round,” fundamental-dominant stops), a residual, excessive boominess frequently remained around 50Hz, usually being most noticeable on one or more of the notes G through B of the 16’ octave. This was true in both the typical living room setting and, to a lesser extent, in the larger room. I have not experienced any such boominess with the Dirac correction that I now have, regardless of stop or sample set. In fact, the note-to-note balance and color of sound from the loudspeakers is, for the first time, so close to what I hear through high-quality headphones that I am not certain which is the superior representation. I suspect that eliminating every aspect of the potential for boominess required the phase correction that Dirac accomplishes (point 2, below). Note: in order to work with Dirac Live, I purchased a USB microphone, UMIK-1, through CSA Acoustics, who for a modest surcharge provided a custom calibration curve for the microphone, spanning 5Hz to 25kHz, that could be loaded into and applied by the Dirac Measurement Tool.
2. Stereo image, impulse response, phasing, speaker delay:
The Dirac measurements resulted in very fine level adjustment between the output of the six channels (3 stereo pairs) that I employ: the “quietest” channel is unchanged by the Dirac processor, and the other five are variably reduced by a small amount (each less than 2db) to match it. This Dirac adjustment successfully takes place despite the fact that the three stereo output pairs are very different in frequency response. This fine balance of loudness helps stereo imaging.
Undoubtedly the more potent enhancement of imaging comes from controlling the time relationships between the channels, including between the subs and the other loudspeakers. As I understand it, two levels of adjustment are made. First, some channels are delayed in order to correct and clarify the apparent center of stage. Second, the phasing of sound between channels is adjusted with some dependence on frequency (perhaps in frequency bands), which Dirac refers to as “mixed phase” correction. I can state that the results that I obtained have been profoundly beneficial. The sound imaging is so clear that the speakers essentially disappear as discrete sound sources, beginning at a very modest distance away from them. What emerges in space is a broad but detailed “stage-full” of sound. The improvement in phase relationships even enhances apparent dynamic range and “body,” not just clarity or position. Plenae and Tutti are more robust and “present.”
One unexpected – but in retrospect quite logical – improvement first emerged among the tremulants of PAB Gravissimo. I am not an avid tremulant aficionado, but it always seemed to me (especially after moving to the larger space) that the tremulant effects in PAB Gravissimo were too understated, to the extent of being essentially inaudible on some flues. With the Dirac Live mixed-phase correction, these PAB tremulants have “come to life” in a way that is very real and appealing. Since this initial PAB finding, I have learned that all HW-modeled tremulants have benefitted in realism and appeal. I wonder how much of the disappointment sometimes voiced in the Forum concerning modeled tremulants actually reflects a need for better output phasing?
It is also noteworthy that the sound of old sample sets benefits greatly from the Dirac Live correction, being sometimes improved to the point of approaching more recent ones in sound quality.
3. List of claims from the Dirac Live user guide, with comments:
•Improve the imaging of your sound system [I strongly agree.]
•Improve the clarity of the music [I agree, from single rank to full chorus.]
•Make voices more intelligible [I “agree,” though Dirac’s claim may actually mean human voices.]
•Produce a tighter bass [I agree.]
•Reduce listening fatigue [I agree, and to a surprising extent! – also, see Appendix, below.]
•Improve the timbre [I definitely agree.]
•Remove resonances and room modes [no specific knowledge/opinion]
•Reduce early reflections [no specific knowledge/opinion]
4. Miscellaneous improvements over ARC2:
A. Reduced latency: with my previous ARC2/Reaper ReaRoute setup, Hauptwerk measured “Sound delay” as 11.4ms (at 48kHz sampling), but with Dirac Live in my system the HW “Sound delay” is 6.0ms.
B. Convenience #1: with Dirac, all six channel corrections are contained in one file for the system-startup Dirac Live app which is automatically “on” for audio output from the computer, though the Dirac corrections can easily be disabled for headphone use. In contrast, for each organ session with ARC2, I had to start Reaper and then manage the ARC2 correction files for each of three stereo tracks. (Among varied attempts, I never succeeded in “teaching” my ARC2/Reaper setup to remember and automatically apply the three different correction file assignments; instead, all three Reaper tracks always booted with a single repeated ARC2 file, and I had to manually reassign 2 of the 3 the files upon each Reaper startup – and not always the same 2.)
C. Convenience #2: the entire digital chain now automatically adjusts to whichever sampling frequency (44.1kHz or 48kHz) is requested by HW. With ARC2/Reaper ReaRoute, I had to reset Reaper manually.
5. Summary:
Conversion to Dirac Live processing did cost (with the additional microphone) the equivalent of a relatively high-end sample set, but for me it has made ALL sample sets sound magnificently higher-end.
6. Appendix:
I followed with some interest the HW Forum discussion regarding perceived variability in the loudness of HW output from one startup to another (viewtopic.php?f=17&t=12967&p=96328&hilit=loudness#p96328). Before moving into the larger space (and therefore before the “loudness” topic thread was opened), I too had experienced such apparent variability, to the point that I sometimes changed and later reversed sample set gain settings. I agreed with various comments that some or all of this effect may be psychological, but I also wondered about my personal experience that the “variability” in sound that I perceived included not only apparent changes in loudness, but also subtle changes in “satisfaction” from one occasion to another, sometimes even suggesting a variably present, vague “harshness.” This in turn made me think about changeable phasing between channels, perhaps dependent upon different computer startups.
I have now been using a successful Dirac Live correction for nearly two months. Whatever the perceptions or realities were that I experienced prior to Dirac Live – with or without ARC2 – I can unreservedly report that with a correctly measured Dirac Live processing template in place, I have perceived NO such occasional “harshness” or slight dissatisfaction, and no variability in loudness. Perhaps the perception vs. absence of a subtly harsh patina corresponds to at least some of the Dirac claim of “Reduce listening fatigue.”
About two years ago, when my HW instrument was still located in a typical living room, the various Forum and personal communications of Jan Loosman and Patrick (pat17) helped me to greatly improve the amplified sound of my organ through the incorporation of ARC2 digital room correction. I also noted at the time Patrick’s report concerning the Dirac Live alternative for digital correction to computer sound output.
After moving my HW organ to a much larger space early in 2014, I struggled to establish an ARC2 correction that was genuinely satisfying. The struggle was exacerbated when I replaced one 12” subwoofer with a pair of 15” Rythmik units. In late December, 2014, I decided to take advantage of the Dirac Live 30-day free trial offer to see how it would perform. As in the past, Martin Dyde kindly played a key role (on 26 December, no less!) in pointing me toward using ASIO4All as the internal multichannel driver that I needed in order to connect all 6 of my HW audio outputs to the Dirac Audio Processor.
The first step in using either ARC2 or Dirac Live is to take measurements of the computer’s audio system output through a microphone connected to the computer. It took three very different microphone placement layouts for me to find a successful scheme for taking measurements for Dirac processing for my space. I admit that I never used the final, successful Dirac microphone placement scheme in taking ARC2 measurements, which may explain a portion of the huge improvement in sound that I have obtained using Dirac Live instead of ARC2. During my efforts with Dirac measurements, I was surprised to learn that the Dirac Live “auditorium” measurement mode only takes one set of speaker readings, which approach turned out to be woefully inadequate. In the end, though, I am thrilled with the results that I have obtained using Dirac Live:
1. Frequency response (Dirac’s term: “Magnitude response”):
In the past, ARC2 definitely improved the evenness and clarity of bass response, but particularly for some 16’ stops in some sample sets (especially for “round,” fundamental-dominant stops), a residual, excessive boominess frequently remained around 50Hz, usually being most noticeable on one or more of the notes G through B of the 16’ octave. This was true in both the typical living room setting and, to a lesser extent, in the larger room. I have not experienced any such boominess with the Dirac correction that I now have, regardless of stop or sample set. In fact, the note-to-note balance and color of sound from the loudspeakers is, for the first time, so close to what I hear through high-quality headphones that I am not certain which is the superior representation. I suspect that eliminating every aspect of the potential for boominess required the phase correction that Dirac accomplishes (point 2, below). Note: in order to work with Dirac Live, I purchased a USB microphone, UMIK-1, through CSA Acoustics, who for a modest surcharge provided a custom calibration curve for the microphone, spanning 5Hz to 25kHz, that could be loaded into and applied by the Dirac Measurement Tool.
2. Stereo image, impulse response, phasing, speaker delay:
The Dirac measurements resulted in very fine level adjustment between the output of the six channels (3 stereo pairs) that I employ: the “quietest” channel is unchanged by the Dirac processor, and the other five are variably reduced by a small amount (each less than 2db) to match it. This Dirac adjustment successfully takes place despite the fact that the three stereo output pairs are very different in frequency response. This fine balance of loudness helps stereo imaging.
Undoubtedly the more potent enhancement of imaging comes from controlling the time relationships between the channels, including between the subs and the other loudspeakers. As I understand it, two levels of adjustment are made. First, some channels are delayed in order to correct and clarify the apparent center of stage. Second, the phasing of sound between channels is adjusted with some dependence on frequency (perhaps in frequency bands), which Dirac refers to as “mixed phase” correction. I can state that the results that I obtained have been profoundly beneficial. The sound imaging is so clear that the speakers essentially disappear as discrete sound sources, beginning at a very modest distance away from them. What emerges in space is a broad but detailed “stage-full” of sound. The improvement in phase relationships even enhances apparent dynamic range and “body,” not just clarity or position. Plenae and Tutti are more robust and “present.”
One unexpected – but in retrospect quite logical – improvement first emerged among the tremulants of PAB Gravissimo. I am not an avid tremulant aficionado, but it always seemed to me (especially after moving to the larger space) that the tremulant effects in PAB Gravissimo were too understated, to the extent of being essentially inaudible on some flues. With the Dirac Live mixed-phase correction, these PAB tremulants have “come to life” in a way that is very real and appealing. Since this initial PAB finding, I have learned that all HW-modeled tremulants have benefitted in realism and appeal. I wonder how much of the disappointment sometimes voiced in the Forum concerning modeled tremulants actually reflects a need for better output phasing?
It is also noteworthy that the sound of old sample sets benefits greatly from the Dirac Live correction, being sometimes improved to the point of approaching more recent ones in sound quality.
3. List of claims from the Dirac Live user guide, with comments:
•Improve the imaging of your sound system [I strongly agree.]
•Improve the clarity of the music [I agree, from single rank to full chorus.]
•Make voices more intelligible [I “agree,” though Dirac’s claim may actually mean human voices.]
•Produce a tighter bass [I agree.]
•Reduce listening fatigue [I agree, and to a surprising extent! – also, see Appendix, below.]
•Improve the timbre [I definitely agree.]
•Remove resonances and room modes [no specific knowledge/opinion]
•Reduce early reflections [no specific knowledge/opinion]
4. Miscellaneous improvements over ARC2:
A. Reduced latency: with my previous ARC2/Reaper ReaRoute setup, Hauptwerk measured “Sound delay” as 11.4ms (at 48kHz sampling), but with Dirac Live in my system the HW “Sound delay” is 6.0ms.
B. Convenience #1: with Dirac, all six channel corrections are contained in one file for the system-startup Dirac Live app which is automatically “on” for audio output from the computer, though the Dirac corrections can easily be disabled for headphone use. In contrast, for each organ session with ARC2, I had to start Reaper and then manage the ARC2 correction files for each of three stereo tracks. (Among varied attempts, I never succeeded in “teaching” my ARC2/Reaper setup to remember and automatically apply the three different correction file assignments; instead, all three Reaper tracks always booted with a single repeated ARC2 file, and I had to manually reassign 2 of the 3 the files upon each Reaper startup – and not always the same 2.)
C. Convenience #2: the entire digital chain now automatically adjusts to whichever sampling frequency (44.1kHz or 48kHz) is requested by HW. With ARC2/Reaper ReaRoute, I had to reset Reaper manually.
5. Summary:
Conversion to Dirac Live processing did cost (with the additional microphone) the equivalent of a relatively high-end sample set, but for me it has made ALL sample sets sound magnificently higher-end.
6. Appendix:
I followed with some interest the HW Forum discussion regarding perceived variability in the loudness of HW output from one startup to another (viewtopic.php?f=17&t=12967&p=96328&hilit=loudness#p96328). Before moving into the larger space (and therefore before the “loudness” topic thread was opened), I too had experienced such apparent variability, to the point that I sometimes changed and later reversed sample set gain settings. I agreed with various comments that some or all of this effect may be psychological, but I also wondered about my personal experience that the “variability” in sound that I perceived included not only apparent changes in loudness, but also subtle changes in “satisfaction” from one occasion to another, sometimes even suggesting a variably present, vague “harshness.” This in turn made me think about changeable phasing between channels, perhaps dependent upon different computer startups.
I have now been using a successful Dirac Live correction for nearly two months. Whatever the perceptions or realities were that I experienced prior to Dirac Live – with or without ARC2 – I can unreservedly report that with a correctly measured Dirac Live processing template in place, I have perceived NO such occasional “harshness” or slight dissatisfaction, and no variability in loudness. Perhaps the perception vs. absence of a subtly harsh patina corresponds to at least some of the Dirac claim of “Reduce listening fatigue.”
Don Vlazny