Using unconventional stereo pairs with Hauptwerk, e.g. quadraphonic organs via 5.1 surround sound systems

Author/date

Rudolf Cardinal (rudolf@pobox.com), Feb 2025. Version of 2025-02-22.

Index

• Purpose
• Reminder
• Overview
• Prerequisites
  • Equipment
  • Setup
• Principles of Hauptwerk audio routing
• Specimen configuration for multiple stereo pairs
  • Idiosyncrasies
  • Audio Device settings
  • Audio Mixer settings
    • Master mixes
    • Intermediate mixes
    • Primary buses
  • Bus groups
  • Rank routing: an example of a quadraphonic organ
  • Setting up other virtual organs
  • Testing
  • Volume explanation
    • Basic psychoacoustics
    • Hauptwerk audio flow and volume correction
    • Relevant volume corrections
• The end result?

Purpose

Hauptwerk is virtual pipe organ software. That is, it takes input from physical organ keyboards (manuals and pedalboards) via MIDI, and makes sounds that reproduce recorded physical pipe organs.

This short page deals with a particular way to set up Hauptwerk audio, via a consumer-type surround sound (e.g. 5.1) system. The main innovation is to “overlap” stereo pairs, creating more stereo spatial locations. The principles are extensible to any other audio system with more than two speakers, where channels are individually addressable.

Note: Using a “standard” consumer surround sound system is NOT what Hauptwerk experts recommend. The normal recommendation is to arrange all speakers in stereo pairs—often lots of pairs—plus at least one subwoofer. The typical recommendation is to use studio monitors, i.e. self-powered speakers. However, sometimes other practical reasons force a different setup, such as the use of an audiovisual (AV) receiver. If that’s the case, how can we get the best out of it?

Reminder

This process involves reconfiguring Hauptwerk audio, so there is always scope to mess things up. You might want to launch Hauptwerk in an alternative config, and edit that. For example, if you normally launch “Hauptwerk” from the Windows desktop, you could launch “Hauptwerk (alt config 1)” (or 2, or 3) instead, and edit that config. That way, your original configuration will be untouched.

Overview

I have a rectangular room with a typical 5.1 system: an AV receiver with five full-bandwidth channels connected to unpowered speakers, and one powered subwoofer for low-frequency sounds. It has approximately this speaker layout:

                                     SOUTH        

  Front_L                            Centre     [Subwoofer]            Front_R
                               +----------------+
                               | |||||||||||||| |
                               | |||||||||||||| | CONSOLE
                               +----------------+
E                                ||||||||||||||                               W
A                              +----------------+                             E
S                              |    Organist    | BENCH                       S
T                              +----------------+                             T





  Rear_L                                                                Rear_R
 
                                     NORTH
    

This isn’t to scale, but it will suffice. Note also that subwoofers are effectively un-directional/omnidirectional, because humans are poor at localizing low-frequency sounds, so I’ll omit the subwoofer from diagrams that follow.

If we use a stereo signal to a conventional AV receiver, it is likely to be able to spread that signal to all 5.1 speakers. That’s good—it’ll sound much better than just using the front L/R pair—but it will still give just a sense of left/right stereo; only one stereo channel is in use.

If we have a signal to our receiver that conveys more than stereo (e.g. using multiple channels on an audio interface, and a 7.1 multichannel input option on our receiver), it is simple in Hauptwerk to arrange speakers in standard, non-overlapping, stereo pairs. So we might get two stereo pairs, front and rear, like this, plus the subwoofer, but with the centre speaker unused:

  <-------------------------------------------------------------------------->
  Front_L                                                              Front_R
                               +----------------+
                               | |||||||||||||| |
                               | |||||||||||||| |
                               +----------------+
                                 ||||||||||||||
                               +----------------+
                               |    Organist    |
                               +----------------+





  Rear_L                                                                Rear_R
  <-------------------------------------------------------------------------->
    

However, we can also create overlapping stereo pairs. That allows us to set up a quadraphonic (quadrophonic) 4-manual organ, for example with the Great division to the South, Swell to the East, Choir to the North, and Solo to the West; plus some distribution of the Pedal division—again, plus the subwoofer. (For a real quadraphonic organ, see e.g. the Kuhn/Goll organ at the St Laurenzenkirche, St Gallen, Switzerland in a video tour by the Scott Brothers Duo, or Jonathan Scott’s performance of the Nutcracker Suite on it.) The next diagram illustrates five stereo pairs:

  <-------------------------------------------------------------------------->
  <------------------------------------><------------------------------------>
^ Front_L                            Centre                            Front_R ^
|                              +----------------+                              |
|                              | |||||||||||||| |                              |
|                              | |||||||||||||| |                              |
|                              +----------------+                              |
|                                ||||||||||||||                                |
|                              +----------------+                              |
|                              |    Organist    |                              |
|                              +----------------+                              |
|                                                                              |
|                                                                              |
|                                                                              |
|                                                                              |
|                                                                              |
v Rear_L                                                                Rear_R v
  <-------------------------------------------------------------------------->
    

We can also create multi-speaker groups and assign a single rank (set of pipes) or division (group of ranks controlled by a keyboard) to a group of individual speakers, and Hauptwerk will assign different individual pipes to different speakers, making the sound move around as you play a scale. Overall, there are quite a few possibilities; I’ll illustrate a couple.

Prerequisites

Equipment

You need Hauptwerk with a basic console, and a sound system with at least 3 speakers to make use of this method. You need hardware that can address the speakers individually, not just e.g. “expand” stereo sound to more than two speakers.

My equipment is as follows:

• A 4-manual organ console, from Nottingham MIDI Organs.
• A PC running Windows and Hauptwerk, Advanced Edition, v9 (2024). The Advanced Edition is necessary for the kinds of sound routing we’ll be doing. All of the descriptions below are for Hauptwerk v9; equivalent functionality, but slightly different labelling, is present in earlier versions of Hauptwerk.
• A Behringer UMC1820 USB audio/MIDI interface (“audio interface”).
• A Marantz SR6011 AV receiver. This can produce 11.1 surround sound; but, more importantly, it can receive a 7.1 multichannel analogue input signal (as well as an S/PDIF optical input, and others). Multichannel analogue input is not entirely common in receivers, but it’s important for maximum Hauptwerk flexibility.
• For my setup:
  • The speakers are 5.1, as illustrated above.
  • The first two channels on the audio interface (channels 1–2) are connected to headphones.
  • Eight channels on the audio interface (channels 3–10) are connected via an 8-way analogue cable to the 7.1 inputs of the receiver, an input renamed “Multichannel”.
    • This is the main input used for Hauptwerk.
  • The optical output of the audio interface (channels 11–12) are connected via a fibre-optic S/PDIF cable to the receiver. This input is renamed to “Optical” on the receiver, and set to “Multi Ch Stereo” mode (meaning that it will use all 5.1 speakers when sent stereo input—but this is “faking” it, not true 5.1 control).
    • While S/PDIF cables can carry compressed proper multichannel audio, such as via Dolby encoding, most Windows-based programs and/or sound cards can’t cope with this, and certainly Hauptwerk doesn’t support it as a multichannel option. Stereo is the norm for Windows sound such as OS noises, web browser audio, and the like.
    • Windows is configured to use this output. Having this as a separate connection also means that Windows is silent when the receiver is switched to the Hauptwerk input, so organ playing cannot be interrupted by operating system alerts.

Setup

• You should have set up and balanced your speakers for distance from the listening position and for volume (for example, using the Audyssey system, a microphone supplied with the receiver and software built into the receiver). A given test signal should sound equally as loud from each of the 5 primary speakers. If you can’t achieve this, don’t worry; you can compensate later in Hauptwerk.
• You should have an ASIO driver installed for your audio interface, primarily because this substantially reduces the latency from keypress to sound. (This may be a driver supplied with the audio interface, as for the Behringer UMC1820, or a generic driver such as ASIO4ALL.)
• You should already have Hauptwerk up and running, producing sounds in response to the organ console.
• In Hauptwerk 9, advanced options are hidden by default, but you will need some. Go to the Touchscreen menu → Settings → [General] Preferences … → Simplify. Then:
  • UNTICK “Use simple audio routing…”.

Principles of Hauptwerk audio routing

Some tutorials:

• Drew Worthen, 2021, https://www.youtube.com/watch?v=QycUYnhI7OU (very clear)
• Sonus Paradisi, 2020, https://www.sonusparadisi.cz/en/blog/understanding-hw5-mixer/
• Richard McVeigh, 2022, https://www.youtube.com/watch?v=FlT44XjWNbs
• Richard McVeigh, 2023, https://www.youtube.com/watch?v=F0_Hlif2vp4

Some “prior art” on using Hauptwerk with 5.1 surround-sound systems:

• Iain Stinson (2019–2020), on Hauptwerk 5 with 5.1 speakers, using 2–5 stereo pairs, but not overlapping.
• A 2022 discussion on organforum.com, but I don’t think that the overlapping stereo idea is suggested.

I’ve not found the “overlapping stereo pairs” method elsewhere.

You should set up complex audio in this order:

1. Touchscreen menu → Settings → [General] Audio device … [we’ll call this AUDIO DEVICE]
2. Touchscreen menu → Settings → Audio routing, reverb and voicing/panning settings → [General settings (for all organs)] Audio mixer (for selecting device channels, reverb, audio routing) [we’ll call this AUDIO MIXER]
3. Touchscreen menu → Settings → Audio routing, reverb and voicing/panning settings → [General settings (for all organs)] Audio mixer bus groups (advanced; for multi-channel output) [we’ll call this BUS GROUPS]
4. Touchscreen menu → Settings → Audio routing, reverb and voicing/panning settings → [Organ settings (for this organ)] Rank routing to audio mixer buses/groups (advanced; for multi-channel output) [we’ll call this RANK ROUTING]
5. Touchscreen menu → Settings → Audio routing, reverb and voicing/panning settings → [Organ settings (for this organ)] Rank voicing and perspective panning [we’ll call this RANK VOICING]

The audio flow is as follows (from top to bottom):

(Then there are 128 mixer presets, which encompass a large number of these settings each. Ignore for now.)

For editing, open these LEFT TO RIGHT (to mimic sound flow left to right across the screen):

• (Rank Voicing, not necessary for core setup)
• Rank Routing
• Bus Groups
• Audio Mixer

Then, effectively, you set things up by working from right to left on your screen.

Specimen configuration for multiple stereo pairs

The key feature is that we create some single-speaker devices with unusual stereo balance, then recombine them to create new stereo pairs. Hauptwerk can use non-consecutive channels for stereo (v9 Advanced User Guide, page 37) but only by remapping them; and we want to use both conventional and unusual pairs, so we have to use a different approach.

• The standard stereo pairs are front (L/R) and back (L/R). If you face forwards, the L/R channels are as you’d expect.
• We create an extra pair that is left (back/front). If you face left, the L/R channels are as you’d expect.
• We create an extra pair that is right (back/front). This one has an even more unusual orientation: if you face right, the L/R channels are reversed (R/L). This is only done so that this stereo pair can be combined with the left pair, and the notional L/R stereo channels then map coherently to “back/front” (rather than mapping to two crossing diagonals). Reversal of this stereo channel in Hauptwerk is unlikely to be noticeable, just as you would likely not notice the exact reversal of all pipes in a rank on a real organ (which are usually not in exact chromatic order anyway), but could if necessary be corrected for in the rank settings.

Idiosyncrasies

My front centre speaker is too loud by default. The receiver is set to adjust it, but I’m not convinced that this adjustment applies to multichannel 7.1 input mode. Therefore I have an adjustment CentreCorrection = −14 dB, applied to this speaker in a couple of places in the configuration below. Ignore it if this doesn’t suit your needs, or adapt the method for this or any other speaker if you have one that is somewhat unbalanced.

Audio Device settings

We use simple, obvious channel names:

Audio Mixer settings

Here we set up many of the simple devices. Individual order is not wholly important, but there is a hierarchy across primary/intermediate/master buses, so if you want a primary bus to address unusually configured devices, those devices need to be configured as intermediate or master buses.

Note that mono(phonic) outputs appear a long way down the list compared to stereo(phonic) outputs! But they are there.

Master mixes

Intermediate mixes

Here we see the unusual individual speaker setups.

These level adjustments are not arbitrary (except for the centre speaker adjustment C, described above), but are based on a principle. See volume explanation below.

Primary buses

Note:

• Primary buses are necessary to produce sound.
• We route all primary buses to (a) the Main_recording_headphones bus, so it comes out via headphones and can be recorded; (b) the Subwoofer bus, so the bass is heard well (remember subwoofers are omnidirectional); (c) some other speaker(s).
• There are more configured here than you probably need!
• The order is definitely unimportant here.

See volume explanation below for a discussion of the level adjustments. The aim is that all resulting bus groups have the same effective volume as a standard single stereo pair, and just differ spatially.

Bus groups

I’ve used a different naming convention for bus groups. You should only need to set mixer/bus group settings once, but you may have lots of virtual organs to map to bus groups, so I find it helpful to have bus groups with standard names even for bus groups that route to a single destination.

Rank routing: an example of a quadraphonic organ

Now we’re ready to play!

This example sets up the Peterborough Cathedral virtual organ, from Audio Angelorum, in a quadraphonic style, with each manual division in a different compass direction, plus a more diffuse distribution of the Pedal division. Note the console positioning in the diagram above: in the centre, facing south. The “console noises”, such as key and stop noises, are routed to the single speaker by the console. Each division has quite a few ranks, but they are grouped alphabetically; * denotes a wildcard.

By default, perspectives other than #1 do nothing.

Setting up other virtual organs

It is then very quick to apply this system to other organs, just using RANK ROUTING.

Testing

• During testing, route key/pallet/stop noises to SILENCE_primary, so you can use Hauptwerk’s activity lights (e.g. in the Audio Mixer) to see properly where the pipe sounds are going.
• Note that divisions “borrow” on this organ, e.g. some Solo stops borrow from the Great. Watch out for this if you route all Solo ranks and then wonder why non-Solo speakers are in use.
• It’s useful to route a single clearly localizable reed stop/rank, e.g. the Solo Tuba 8′, to TEST_primary, and play with the destinations for that bus.
• To test stereo channels, it can help to pick three adjacent keys and assign them to full left, stereo centre, and full right. In the RANK VOICING screen, select a rank (e.g. that Solo Tuba 8′); then in the Adjustment drop-down box, pick “All perspectives: overall: amplitude stereo bal (pct)”. The sliders now correspond to stereo balance. On the Peterborough virtual organ, “Octave 3” of this rank corresponds to the octave based on middle C. You could therefore set E to have a stereo position of −100%, F to 0%, and G to +100%. Click Reset afterwards to cancel your changes.

Volume explanation

Basic psychoacoustics

• Sound power/intensity is usually measured in decibels (dB). This is a logarithmic scale: an increase of 10 dB is a tenfold increase in physical power; a decrease of 10 dB corresponds to a tenth of the original power.
  • Strictly, sound power is the energy per unit time, e.g. at the source of the sound (or the power passing through a surface enclosing the source). It does not depend on distance.
  • Sound pressure (level) (SPL) is typically measured at a point, in decibels, and will depend on distance from the source. It is defined with respect to a reference sound pressure in air (20 µPa, roughly the threshold of human hearing at 1 kHz), such that 0 dB corresponds to the human hearing threshold at 1 kHz.
  • The relative power of P versus a reference power P₀, in decibels, is 10·log₁₀(P / P₀).
  • Therefore, a doubling of power corresponds to a change of +3.01 dB. Similarly, a three-fold change is 4.8 dB, a four-fold change is 6.0 dB, and a five-fold change is 7.0 dB, to one decimal place.
• Playing the same sound twice simultaneously—for example, using two speakers at the same volume, instead of one—corresponds roughly to a doubling of sound power (assuming no constructive/destructive interference), and therefore an increase of +3 dB. And chopping a signal in “half” (e.g. using only one of a pair of stereo speakers) likely corresponds roughly to a halving of power, and a change of −3 dB, I would surmise. (Confirmation: see Wikipedia on sound pressure / multiple sources.)
• We have seen that decibel is defined in terms of power P as 10·log₁₀(P / P₀). But power is related to the square of the corresponding field quantity, F (e.g. voltage, current, or pressure), where P = F². See Kuhn (2020) on digital signal processing. Therefore, the decibel quantity can also be written 10·log₁₀(F² / F₀²) = 20·log₁₀(F / F₀).
• Computer audio signals are translated directly (linearly) to voltages, not power. Therefore, a doubling of a digital audio signal is likely to correspond to +6 dB, and a halving to −6 dB.
• Humans perceive loudness on approximately a logarithmic scale: see the Weber–Fechner law and Stevens’s power law (Wikipedia; Stevens 1957). An increase of 10 dB corresponds, roughly, to a perception of “twice as loud”. And a change of −10 dB corresponds roughly to “half as loud”. The unit of subjective loudness is the sone.
  • Approximate derivation via Stevens’s power law: perception = k·I^a where I is physical intensity (e.g. sound pressure level) and a = 0.67 (per Wikipedia on the power law) for auditory tone sound pressure, per link above; therefore, a doubling in perceived loudness corresponds to a multiplicative change in intensity (ΔI) given by 2 = ΔI^0.67, or log(ΔI) = log(2) / 0.67, or ΔI = 2.81. This multiplicative change in sound pressure level corresponds to 20·log₁₀(2.81) dB = 8.98 dB.
  • Alternatively: Stevens’ original 1957 review (per link above and see sone) used an exponent of 0.3. That would suggest that for ΔX, a change in a different quantity, 2 = ΔX^0.3, or log(ΔX) = log(2) / 0.3, or ΔX = 10.08; so that appears to be an exponent relating to decibels, and indeed the original paper appears to use that (e.g. pages 163/164 and Figures 2/4). The original source is Stevens (1955, doi:10.1121/1.1908048), which gives the very clear equation log L = 0.03·N − 1.2, where L is loudness in sones, and N is the number of decibels; also states it as L = k·I^0.3 (L loudness, I intensity proportional to square of sound pressure); and makes explicit that the exponent 0.03 is log₁₀2. This paper also summarizes the data on which this equation was based.

Hauptwerk audio flow and volume

Hauptwerk’s basic audio concept is of a stereo channel. In the configuration outlined above, we fiddle around with this somewhat. But imagine that at any instant, a stereo signal is represented by two “level” values, a left value and a right value, and let’s suppose that these levels are each represented by numbers in the range 0–100. The numbers will then change rapidly (e.g. at 44.1 kHz) to produce sound. But at one instant, we can imagine a stereo signal such as L = 50, R = 50 (balanced), or L = 90, R = 10 (louder on the left). And then we can think about the way this signal flows through Hauptwerk, from ranks to audio devices.

• Basic stereo output. When Hauptwerk sends a stereo signal (e.g. L = 50, R = 50) to an actual stereo pair of speakers, I presume the channel connected to the left speaker gets the left signal (e.g. L = 50) and the right speaker gets the right signal (e.g. R = 50). Of course, that is likely to happen within the audio interface rather than within Hauptwerk.
• Routing and copying. When Hauptwerk routes signals to or between buses/mixes, or copies a signal to more than one bus, I presume the signals are copied unchanged.
• Monophonic output from stereophonic input. When you specify a MONO output, my guess is that Hauptwerk averages across the L and R channels to give a monophonic (M) output: M = (L + R) / 2, I presume. Therefore, if L = 50, R = 50, the output will be M = 50. If L = 10, R = 90, the output will be M = 50. If L = 50, R = 0, the output will be M = 25. (The Hauptwerk v9 Advanced User Guide is not specific about the method; it refers to “a down-mixed mono signal” on page 35.) The upshot is that using a speaker in “mono” mode probably does not change the average signal/power delivered to that speaker, but one must be mindful of what goes into it; see the next point.
• Stereo balance. Hauptwerk does not provide, directly, a way of taking just the “left half” of a stereo signal and sending it to a single speaker. So what we have done in the configuration above is to set the stereo balance fully to the left (or right) for some buses/mixes. When Hauptwerk sets the stereo balance fully to the left, it does not “move” or blend the right-channel signal into the left as well; instead, it cuts out the right-hand signal. This is not clear in the manual, but the tooltip (the pop-up help you see if you click “?” and then the stereo balance box) says: “Adjusts the stereo balance, with 0 being normal, −100 being the left channel only, and +100 being the right channel only.” Hauptwerk’s stereo balance values range from −100% (full left) to +100% (full right), so we could call this percentage p, range [−100, 100]. Let q = p / 100, making q in the range [−1, 1] (−1 full left, 1 full right). And we can use the notation if(a, x, y) to mean “if a is true, then x, but if a is false, then y”. Then I guess that Hauptwerk uses approximately the following method:
  • L_output = if(q ≤ 0, L, [1 − q]·L) ;
  • R_output = if(q ≥ 0, R, [1 + q]·R) .
  Therefore, if a signal was roughly “centred” before this process, setting its stereo balance fully in one direction will lead roughly to a halving of signal (a quartering of power; −6 dB).
• Bus groups. By default, many ranks are roughly centred in the stereo channel. So if you route them to a stereo bus connected to a stereo pair of speakers, you might get, approximately, each note played on both speakers. If you play a 4-note chord, you’ll get four notes coming from each speaker. However, if you use the same two speakers, in mono mode, as two primary buses in a bus group, the pipes are distributed between the primary buses within the group (e.g. Hauptwerk v9 Advanced User Guide, p. 38; the actual algorithm used is chosen via RANK ROUTING, as described on page 41). Something like this:

            BUS GROUP           versus    STEREO PAIR
            L_mono    R_mono              L         R
            C-G       E-C                 C-E-G-C   C-E-G-C
          

  So, relative to a stereo pair, I’d expect that a two-bus group like this will probably cause (roughly) a halving of apparent power, or a change of −3 dB. But empirically it is −6 dB! Some other comparisons:

  Bus group	Speakers per pipe	Versus	Speakers per pipe	Expected power factor	Expected dB	Empirical dB
  n × {primary bus to 1 × mono speaker}, regardless of n	1	1 × {primary bus to stereo pair}	2	0.5	−3	−6
  n × {primary bus to 1 × mono speaker}, regardless of n	1	group of 2 × {primary bus to stereo pair}	2	0.5	−3	−6
  n × {primary bus to 1 × mono speaker}, regardless of n	1	1 × {primary bus to 2 × stereo pair}	4	0.25	−6	−12
  n × {primary bus to 1 × stereo pair}, regardless of n	2	1 × {primary bus to stereo pair}	2	1	0	0
  n × {primary bus to 1 × stereo pair}, regardless of n	2	1 × {primary bus to 2 × stereo pair}	4	0.5	−3	−6

  For reasons I do not understand, the adjustment e.g. for the first row is 6 dB, not 3 dB. Presumably I have interpreted something as power when actually it corresponds to a field quantity. The empirical values are from me listening for equal loudness, and are used in the setup described above.

Relevant volume corrections

Therefore, we use approximately these corrections:

• When we use a fully lateralized stereo channel, that halves the signal strength (quarters the power); therefore, we compensate directly by an adjustment of +6 dB for all of these, and we don’t need to think further about that.
• When bus groups are specified, and when those bus groups address >1 primary bus, there is an effective volume change, but no level adjustment is possible to the group; instead, you need to adjust the corresponding primary buses. For most sensible combinations, we can adjust to the “standard” loudness of a single stereo pair by (a) doubling the signal (+6 dB) that primary buses send to mono speakers (without altering the signal they send to the master/subwoofer mix), and (b) halving the signal (−6 dB) that primary buses send to double stereo pairs. This is shown in the table above.
• An example of an individual speaker correction (turning the centre speaker down a bit) is also shown above; see CentreCorrection.

Of course, you may want to vary these to get the best results for your setup.

The end result?

It works. The organ sound comes alive!