Wed May 29, 2013 11:05 am
Report on ARC2 with Hauptwerk: extremely pleasing results
I am indebted to Jan Loosman and the others who have contributed to this ARC thread (including, of course, Thomas Gentry, for my initial HW launch). I first learned of ARC as a result of the resuscitation of this thread in April, 2013. It has taken me a while to assemble the software, hardware, DAW learning curve and overall project time needed to apply ARC2 correction. The following is an early report.
1. LATENCY: ARC2 correction DOES slow things down.
[Note: all references to milliseconds report the “Sound delay ms” measurement provided by HW.]
When I converted my DIY console to HW in 2010, my initial sound card was an M-Audio Delta 1010LT. At that time, I found that I could run my system (i7-930 dual quad) with a buffer size of 768 but not 512. I subsequently upgraded to an RME HDSPe AIO sound card, including its accessory analog output card. The RME interface provided a distinct improvement in sound quality, but it did not support a buffer size of 768, so I set the size to 1024, which gave Sound delays measured by HW in the low 20’s (ms). This was acceptable.
Although I was immediately pleased with the sonic result of ARC2 correction, I also “felt” a new latency in playing that was roughly similar to the least (Manual I) of the intentional pneumatic-modeled delays present in the OAM Sauer/Dortmund sample set. When I looked at the HW Engine panel, I saw that adding ARC2 correction had doubled the Sound delay to more than 46ms (I think it was 46.7ms). I was already so enamored of the ARC2 improvement in sound that I had to try to reduce the latency by setting the buffer size to 512, something I had never done with the RME interface. Resetting the buffer size from 1024 to 512 in the presence of ARC2 correction has lowered the Sound delay to 21.3ms, which is once again acceptable. I have yet to experience a full-blown polyphony/CPU problem with this smaller buffer size and the RME card, although the full Tutti of PAB Gravissimo can run both of these parameters (Polyphony and CPU utilization) “into the yellow” with as few as five moving parts. The Rotterdam Tutti (without rear channels) sounds superb with ARC2 correction but does not challenge polyphony/CPU as fully. Perhaps having the RME card on the PCIe bus allows my system to outperform my prior experience with the 1010LT on the PCI bus. In any event, it continues to be certain that ARC2 correction slows things down: at the same buffer size of 512, the Hauptwerk “Alt 3” that I have reserved for headphones (i.e., without ARC2 correction) shows a Sound delay of 12.3ms instead of the 21.3ms I consistently obtain in the presence of ARC2 correction.
2. SOUND: what a pleasure!
A. Similar to descriptions elsewhere in this thread, I have also obtained a tremendous increase of clarity in the sound not only of bass notes but well into treble range. I do not personally—in my specific room, etc.—experience any of this as either a result or an adjunct of increased or inappropriate “brightness,” which may be a concern for others. The ARC2-corrected bass that is present is more crisp and potent. Subbass and French Flûte pipes are under control even below 8′ pitch: it has occasionally startled me how much better the ARC2-corrected sound allows me to “experience” an actual pipe of this sort instead of becoming lost in the mental interpretation of an overrepresented sine-wave fundamental. I would be sold on ARC2 for clarity alone.
B. A second benefit for me—one for which I was hoping and which has been well fulfilled—is that my two dissimilar pairs of stereo speakers now give nearly identical sound. I should mention that this has been most distinctly demonstrated using the exceptionally helpful CalibOrgan sample set made available by Al Morse. (Thank you, Al!) I run two pairs of stereo speakers: a pair of first-year Advents that I refurbished in 2000 or so, and a smaller pair of AR bookshelf speakers that I bought used but in good shape. The Advents are connected to an RME analog output through the stereo line-level crossover of a Regnar subwoofer, and all stops below 8′ are assigned to this “2.1” assemblage. Prior to ARC2 correction for flat response, the Advent and AR speakers were easily (and undesirably) distinguishable through much of the keyboard range. That is no longer the case, particularly above tenor C. By the way, none of my five speaker boxes is “ported.” They are all of a design once referred to as “acoustic suspension,” which I have always enjoyed partly because of its tightness of bass response.
C. Lastly, because the room response to lower frequencies is under control, I have been able to reset the overall volume of the organ upwards, resulting in a more satisfying presence and dynamic range.
3. NUTS AND BOLTS: miscellaneous details.
A. Microphone preamplifier.
Not surprisingly, the condenser microphone included with ARC2 requires a preamp that can provide the 48V field across its sensor. However, although the RME card has a balanced line-level input that is appropriate for an already pre-amplified microphone signal, it does not contain a filter to remove the 48V. Mixer panel inputs may have such a filter, but I would not expect a computer sound card to have one. After some investigation, I decided that it made most sense to acquire the Rane MS1S single-mike preamp. It was incrementally more expensive than other options, such as an inexpensive filterless preamp with a requisite external output filter, but the Rane unit incorporated both the 48V filter and superlative preamp performance which I believe is reflected in the degree of success of the overall project.
B. Robustness of ARC2 measurement and application.
As suggested above, I am not using near-field/studio speakers—instead, the speakers I have were intended to enhance dispersion. When I ordered the ARC2 product, I was looking forward to any tips the user guide might give concerning microphone placement during the measurement procedure. I found that although the ARC2 guide is appropriately helpful, translating it to my reality required a leap of faith. For example, the guide refers to measurements possibly being taken only “inches” apart in order to represent the soundfield of typical mixing studio positions. In addition, there was no discussion of room asymmetry in the guide, but various constraints prevent my organ from being centered along either axis of the room it is in. I am happy to write that at least in my circumstance, the overall ARC2 measurement/correction procedure appears to have been less fragile or fussy than one might infer from the guide. I used the front center of the organ bench as position 1 (the ARC ‘key’ position), the ends of the bench as positions 2 and 3, and then I widely distributed 7 additional measurement points across the room, arranged symmetrically along the centerline that extends out from the organ setup. I was careful to use (within an inch or two) the same locations for measurements from both of the speaker sets (Advent “2.1” and AR). Despite this limited, off-center measurement procedure, the entire room has benefitted with remarkable evenness.
C. Measurement level and “balance.”
The test signal from the ARC2 measurement application is generated at a much higher level than the Hauptwerk signal is presented to my soundcard. As a result, for the room measurement procedure the output volume on each power amplifier (two Cambridge stereo amps and the subwoofer panel amp) had to be set at much lower levels than usual. The problem with this is that I have no method for analytically “matching” the subwoofer volume to the Advent amplifier volume throughout their range, as if to adjust them equally in tandem. During my first “take” of measurements, the subwoofer was quite overrepresented, which resulted in a faulty ARC2 correction. Once I was more accustomed to the intended sound of the test signal (partly from measuring the AR speakers alone), I was able to “match” the subwoofer and Advents aurally, within what I now believe to be a widish acceptable range, and this resulted in a good if not ultimate correction file. I might mention that for me, I had to proceed through measurements with some quickness because I used layers of blankets to muffle the modest fan noise from the computer. (Also, HVAC, appliances, and electronic media were off, traffic outside was fortunately absent, etc.)
D. L/R Combined Correction.
As Jan recommended, after I licensed my copy of ARC2, I was careful to download from IK Multimedia the latest version of ARC2. With this version, I have spent a little time using the LR Combined Correction facility that Jan and Pat discussed in the thread. This combined correction obviously relies on some sort of alternative calculation based on the raw room/speaker data recorded through the microphone. I am not convinced that this alternative calculation offers any benefit in my circumstance; I expect to discontinue its application.
E. REAPER.
I have read that REAPER is designed as a compact, efficiently operating DAW. My experience is in agreement. Early on, when I simply ran HW through REAPER (without any processing), I experienced no increase in latency at all. This may not be surprising, but I was particularly pleased to observe an additional “efficiency” that had been a concern of mine from the beginning: even after ARC2 was fully functional, I observed no competing RAM utilization from the addition of REAPER, the plug-ins for 2 tracks, and the correction files. This meant that sample sets that I have been loading to within 1GB of RAM capacity (as reported in HW) required no recaching and loading at a smaller size in order to make room for ARC2 correction.
F. A novice in REAPERLAND: inexpert observations/solutions.
Having never before seen a DAW, and working with a version of REAPER (4.402) noticeably past the one (3.7.5) discussed in detail in the HW4 guide, there are a few tricks that I had to stumble across to make things work. I apologize for all apparent/actual density on my part.
1. I found Jan Loosman’s recommendation to employ REAPER’s ReaRoute to be spot-on correct. Unfortunately, it took me a while to get everything put together, addressed and recognized properly, mostly because of issues below. I do believe that ReaRoute is a very elegant method of applying REAPER/ARC2 to Hauptwerk in a PC/Windows system.
2. Perhaps there is a method by which the following “difficulty” can be avoided, but this is the extent of my understanding and experience:
The REAPER Project Sample Rate setting (viz. 48K or 44.1K) exerts preeminence over all others settings where Sample Rate can be declared or determined:
* Over REAPER’s own Sample Rate setting in REAPER Preferences
* Over the ASIO Sample Rate setting (for me, the RME/Hammerfall setting)
* Over any “request” made by Hauptwerk for a Sample Rate change
A REAPER “Project” that has been set up to process signals from Hauptwerk will have a Sample Rate setting of its own. This Project Sample Rate setting can be conveniently changed in the Information window for that Project. When it is changed, it automatically and instantly changes the ASIO Sample Rate of the interface. In contrast, resetting the ASIO Sample Rate has no effect on the REAPER Project Sample Rate. In addition, if the Project Sample Rate in REAPER does not match—IN ADVANCE—the Sample Rate of a HW sample set that is to be loaded, the HW load will abort (near the END of the load!). The REAPER Project Sample Rate will not change to accommodate the HW Sample Rate “request.” Manually changing the REAPER Project Sample Rate takes 2 seconds or less—unless you first need to be reminded to do so by an aborted HW load.
3. I agree with Jan Loosman that it is also a very quick, minor task to assign a different ARC2 correction scheme to a second (third...) REAPER track, in case two or more distinct corrections need to be applied. In my experience, when REAPER starts, it takes whatever ARC2 correction file was saved as being assigned to the track 1 ARC2 VST (whether or not it is the “original” or “oldest” correction available) and assigns the same correction to all tracks. Opening the ARC2 dropdown to change the ARC2 correction for track 2, etc., is a quick chore for the beginning of a Hauptwerk session.
4. The “Record” function of a REAPER track must be enabled (but no recording actually started) for a HW signal to be passed through ReaRoute for amplification.
5. Unless ARC2 installation is customized, it is necessary to “Add” (by browsing) the directory address of the folder that contains the ARC2 VST plug-in to the directory field in the REAPER Preferences\Plug-ins\VST[plug-ins settings] window. REAPER will thereafter list ARC2 for selection when clicking on an “Fx” slot to add the VST for each track in a Project.
G. Hauptwerk voicing.
I do diverge somewhat from Jan on this issue. I will continue to use Hauptwerk voicing in conjunction with ARC2 correction. There are four reasons for which I applied Hauptwerk voicing prior to ARC2:
* Improve balance (via amplitude adjustments), usually to “smooth” one or more notes within a rank, but also for returning PAB Gravissimo in detail to the original organ’s inter-stop and inter-divisional relationships
* Repair the color of an individual pipe (viz. ameliorate an excessively buzzy Gedackt or a dull Prinzipal or reed, etc.)
* Remove clearly artifactual noises, such as a ‘click’ or other pipe-unrelated, high-frequency sound
* Reduce excessive post-release ringing of select individual pipes (viz. a strong, round flute) in a single-release sample set
I have barely begun to peruse my sample sets with ARC2, but my experiences so far indicate the following, in order:
° I will continue to use amplitude adjustments, though with ARC2 I have seen the need to change (reduce) many and eliminate some. I have even noticed that some of the amplitude adjustments I made in the past make even MORE sense following ARC2 implementation, now that the correct timbre is restored and the adjustment is not merely an approximation complicated by wrong sound. Perhaps my amplitude adjustments are not a purist’s approach, but my limited performance abilities are too quickly challenged by what I do or do not hear clearly. (Come to think of it, that’s ANOTHER good reason for applying ARC2 correction.)
° So far, I have revisited only a few “color” adjustments (I never did make many), but I found that it has been appropriate to remove all of them under ARC2. This is more satisfying than surprising.
° Voicing to remove artifactual noises was something I needed and did only rarely. I have already found a couple of instances in which when I remove this corrective voicing, the previously offending noises are absent with ARC2 correction applied. This is a little surprising, but again quite satisfying. I believe that I am not the first to experience this repair of noises by ARC2 correction.
° I developed my anti-ringing gimmicks for a few Flute notes in Forcalquier. I have played Forcalquier with ARC2 and (without actually experimenting) found no reason to remove the repairs. This may be another stable application of HW voicing for me in conjunction with ARC2 correction..
Final summary: I have explored so far only a small percentage of sample sets with ARC2 correction, but every experience has been positive, even exciting. The project has proven itself to be worth every penny and was specifically more pertinent than any option of speaker upgrade or reconfiguration could have been.
Cheers!
Don Vlazny
Don Vlazny