Spogg's organ...

[Spogg]

This is all Hugh Banton’s fault!

He aroused my interest in pipe organs, so I’m making a pipe organ synth. I just had to. It will be called the Quilcom Rank and I was hoping to have it ready for Christmas 2018, but it might not be. So here’s a taste of the sounds I’m getting, with some seasonal stuff included.

This is a link to a Dropbox folder from where you can play the pieces directly.

https://www.dropbox.com/sh/159d332uwpw3 ... ig2Ka?dl=0

I’ve deliberately taken a different approach to Hugh’s excellent HB3 organ. I’ve used physical modelling for wind and pipe resonances and additive synthesis for the tone itself, with amplitude interactions between the sources.
More specifically, there are NO samples or sampled single cycle waveforms involved.

The very best reference site, where I got my information from, is http://www.pykett.org.uk/ and this is a treasure trove of background aspects of pipe organ technology and history. Some of the preset sounds were taken from his hard-won spectrum graphs for single real pipes, and set up on the additive synth. I also compared real pipe sample sets with my work to help tune the settings inside.

No synthesiser, or sampler for that matter, will be indistinguishable from the real thing and the reasons are very well explained by Colin Pykett. But I’m quite pleased with where I’ve managed to get to.

I’m not much of a musician, and certainly not a proper organist, so these are taken from free midi files, and I’ve had to guess at the appropriate registration settings. So, if you know the subject, please excuse my naivety. I think that, in the hands of a skilled and experienced organist, much greater sounding music would be achievable.

So I wish you all a very happy and peaceful Christmas.

Cheers

Spogg

[tulamide]

Wow!

This is a very, very good approach! When I heard Bach's Toccata I was blown away. I also took the chance to compare it with a very good recording of a real Pipe Organ session.

https://youtu.be/ho9rZjlsyYY

And that's when I realized that it's basically two things that seperate the two. One is a minor issue: If you listen to the original pipe organ, you will notice that the perceived loudness of the individual pipes (just listen to higher and lower notes) is even, while on your organ some note ranges are perceived louder than others. I'm sure, you can manage to work that out.

The second one, and that might be a work of weeks if not months, is a proper reverb sound. When listening to your example, you can hear the pipe sound coming straight from the center, then spreads out to the stereo range due to the reverb. In the recording of the real pipe organ, you'll notice a different behaviour. The lower the notes, the more they are unlocatable. This is due to the room's reflections. The slower the sound wave, the higher the chance that such reflections arrive at your ear earlier than the source sound.

I think this could be best solved with Convolution Reverbs. I have listed a few sources of free IRs from various places, at least one church or cathedral exists per link. You could use them first with Reaper's ReaVerb, to see how they match the desired outcome.

https://fokkie.home.xs4all.nl/IR.htm
https://www.voxengo.com/impulses/
https://freesound.org/people/jmuehlhans/sounds/220752/
http://georgievsound.com/free-downloads/

If that doesn't help, here's an online Impulse Response Modeler. It is synthetic, of course, but features very good tweaks and might produce usable results. License is free to download and use any IRs generated, but no selling or distribution of said IRs on its own.

https://www.wiregrind.com/impulse-response-modeler/

Hope it helps!

[Spogg]

Thanks tulamide

I really do appreciate your detailed comments

The balance of low and high notes’ apparent volume is called scaling, in organ parlance. This was down to the organ builder and they all sound different, even between individual ranks (sets of pipes) because historically the design was more an art than a science. Some ranks (trumpet reeds for example) are expected to sound more strident than others, so the higher note, and harmonics, are louder. The performer will select the registration (what stops to use where) to suit the organ, its environment, and the chosen music.
Colin Pykett analysed this scaling and found that generally the pipe diameter (which affects SPL and harmonics) halves approximately every sixteenth note, gradually of course. This means that the lower pitched pipes are usually thicker than the higher ones.

Since I haven’t published the schematic yet (I still have the user guide and a video to do) you wouldn’t know that I have a switch for turning on overall scaling and two knobs for scaling the odd and even harmonics individually. The latter excludes the fundamental, since this is the base value.
So you could achieve what you suggest with the synth. Of course I used my own ears and inexperienced judgement to get sounds and registrations that I personally favour, but there will be much flexibility for the user to adjust what they prefer.

The Rank module contains Martin’s recent FDN reverb, which I think sounds nicer than his Freeverb. So what you hear in my demos are up to 5 instances of the plugin, each with that reverb running! OK, I was being lazy Obviously one would use several plugin instances to simulate different sounding ranks and typically just one favourite reverb on the Master bus.

In the manual I intend to advise the user to turn off the inbuilt reverb, in every plugin instance, and use an external reverb of their choice to simulate the space and response they’re looking for. As you know, I favour having an inbuilt reverb for auditioning and sound design, but there’s no way I could please everyone with whatever I might choose.

I will look into convolution reverbs from your links though, so many thanks for that.


Cheers

Spogg

[Spogg]

Well, I’ve spent a couple of hours messing around with the Bach Toccata.

I increased the scaling to reduce the higher frequency notes’ amplitude and also tried several church/cathedral IR files in Reaverb, without the inbuilt reverbs turned on.

I uploaded what I thought was the best (into the original music folder on Dropbox - the original is still there untouched) but to me they just sound different. You see I’m no good at assessing such things, and I always seem to prefer the last one I heard

Anyway, I’d be interested to hear your opinion.

Cheers

Spogg

[tulamide]

I just listened to the new recording, so here's my first impression.

The overall volume/loudness is much lower than before, so that alone will make it appear different.

Also, the usage of good reverb support is meant to make a difference. In reality, people never heard the pipes without its surroundings. Most, if not all, pipe organs are found in churches/cathedrals. Which makes us being used to that specific sound.

However, in this case, the richness of a church reverb didn't happen. Maybe the used one had a too short tail, or it was recorded from a too close position? Normally you sit like 10+ meters away from the organ. But the reverb is so intimate, almost as if lowpassed and with very little amount in the mix. In this specific case the original reverb was the better fit.

Do you listen with speakers or with headphones? They are mostly changing a sound based on their material and quality of the tranducers. I'm using the same headphones (I mean the same model, bought it several times over the years) for almost 30 years now and know exactly how something has to sound with tehm, to have a decent, pretty linear mix. For me it was the low-mid/low frequencies that were over-emphasized, not the higher ones. This was even intensified now. For example, at 5:32, when the lower notes kick in, it's almost as if another instrument is playing with amplification. Those notes kill the other ones. Compare it with the video I linked to.

Listen from 5:09-5:41 in the video and 5:25-5:56 in your second toccata mp3. Those are the same passage. Don't you hear how nicely the low notes sit in the background, just supporting the higher notes, wheras in your mp3 the low notes take over the whole passage?

I know you said it could be done by the user as he/she sees fit. But you also asked for my opinion. I hope it is ok!

[Spogg]

Thanks tulamide, once again I really do appreciate your opinions and comments on this.

I prepared all the pieces using my Corsair 2.1 speakers only. This I guess means the bass is handled by a completely different signal path so I would get a different effect using 2.0 headphones. I will try it!

I tested with 6 different promising IRs and rated them according to my own expectations. So the one I chose was the best for me, but I was still not completely happy with it.

Whilst I’m pretty happy with the Rank itself I would say we’re talking more about mixing and mastering and that’s where I fall flat on my face. You have the experience that I lack, big time.

Colin Pykett makes many points about taking samples, and where the listener is seated in relation to the ranks. If you are close to the pipes the sound colour and spacial effects are very different to being much further away, plus there are frequency dependant cancellations and summing effects in a real church. I know this from my old church-going days. I once sat directly in front of the Swell department and the sound was horrible. That was why they were the least popular pews, as I discovered.

If you are seated close to a department (set of ranks), the direct sound is with you before any reverb is heard, like in my original. But further away the screechiness, chiff and wind sounds are very much reduced and the reverb is more apparent and, as you say, may be heard at a higher relative level to the source sound. Plus, as you said, localisation is reduced as you hear the sound reflected and obfuscated from many surfaces. Add to this the fact that every individual pipe is in a different location and the problem of accurate synthesis becomes monumental. This is a big problem for Virtual Pipe Organs that almost always use sampling these days. Where do you put the microphones? If you want a dry sound, so you can create your own space later, how do you eliminate the auditorium acoustics? The same goes for IRs. The impulse should originate up in the pipe ranks but where should the IR be recorded?

My focus at this time is to get the Rank finalised, a user guide and video made. I think and hope there is sufficient flexibility to achieve a wide-ranging sound source.

Once I publish it I would love to hear your mixed and mastered version, but I realise that would take a lot of time, so I have no expectations, just interest.

I’ll delete the Reaverb version from Dropbox since neither of us is impressed!

Cheers

Spogg

[Spogg]

Dammit tulamide, I just had to have another go at this

It’s on Dropbox in the same folder and called Bach Toccata remix 1.

This is what’s different:

I remixed on headphones (Sennheiser HD 280 Pro).
I added another plugin instance set to a bright reed pipe sound.
I reduced the extreme panning, to be much more subtle (the original was for speakers but it sounded awful on headphones).
Re-balanced the tracks to sound more even or linear (to me anyway).
All inbuilt reverbs turned off and the free Reverberate CM by LiquidSonics applied to the Master track. This is a convolution reverb that comes with many IRs including a “Cathedral”. I re-shaped the tail decay at the end, to match what I heard on the YouTube version. It actually has an unexpected end whereby the tail seemed to close down in a linear fashion from about 4 seconds. Of course that may be down to YouTube’s encoding of audio.

Now when I compare this with the original I do prefer it, but if you have the time and inclination I would really appreciate some feedback (or from anyone else). I’m not fishing for compliments and won’t be offended at all by any criticism which can explain how I could improve it.

Cheers

Spogg

[tulamide]

This is a stellar improvement! The sound is so much more balanced. The pipes don't fight against each other anymore. And isn't it funny, that of all people I am still not convinced of the reverb? I was so sure that a proper convolution reverb would nail it, but there's something in the sound of those churches, that I am missing.

I looked for more videos, to find newer ones. Newer videos have a better sound quality due to Youtube's compression methods that improved over the years. Here's one from 2018. I don't like that version, the artist rushes through it and even adds a few notes. But it is a recording from the famous Silbermann-Orgel in the Freiberger Dom. Built between 1710 to 1714, it serves for over 300 years now. 40 organ stops. Relatively small church for such a big organ. Yet the reverb is much more present. I think, what I am missing can be heard from 2:11 to 2:30, when he plays the fast notes. There's like a shimmer, additional frequencies, maybe from the late response that now almost works like an echo. I can't explain it. But overall the sound is denser.

https://youtu.be/tRZrhg1fm50

However, that is a minor thing, and even not in the scope of your organ. So, well done!

[Spogg]

Many thanks again for the feedback and I’m glad you can hear an improvement

The issue I can see with convolution is that the impulse has to be generated in the pipe rank and ranks are all in different places too, even down to individual pipe placement. So the overall response and tail will never be a perfect fit. You’d need each and every pipe to have its own IR, and that’s just silly. So we always end up with a compromise, a sort of average which is perfect for none. Given that the IR I used was probably just taken in the body of the church (who wants to climb a tall ladder and fire a gun?) it won’t even be roughly correct. But it’s all I’ve got to work with.

Since my last post I’ve looked into what actual organists look for in their reverbs for “electronic” organs. There seems to be an agreement that there should be a “sustain” part at the start, where there is little fading of the reflections. This could be the impression given by the arrival of different delays of reflections accumulating, so some build up while others die away. This sustain is then followed by a more rapid fade-out, which is what I think I heard in the original YouTube version. Total reverb time is typically 5-6 seconds to -60dB. I might be able to replicate that with that plugin.

Something that puzzles me greatly is that in the 2nd video I can hear a vague pitch modulation in the tail. Can you hear this or is it just me? If it’s really there how could that be accounted for in acoustics? It would need a change of the speed of sound and the only way that could happen is down to temperatures in the roof space being different to that at ground level. That would cause differences in air density but I don’t even know if that would be audible. Interestingly I found that pro reverbs can sometimes feature a pitch shifting for the tail.
Here’s a discussion about what I might be hearing:
https://www.reddit.com/r/musictheory/co ... _in_pitch/


All fascinating to think about…

Cheers

Spogg

[tulamide]

I'm pretty sure you are thinking of a convolution reverb the wrong way. It is always meant to produce the reverberation from the listener's stand! That's why you produce the sine sweep or clap at a position, a listener might be. For example, in a car, you would record the impulse response from the driver's seat, if you want the listener to have the feeling of driving the car. You wouldn't record the impulse response in the engine bay, although the sound is produced there.

Regarding pitch modulation, I do sense a wave-like behaviour of the tail, but I don't hear it as a pitch modulation. For me it is more like the sound you get, when two oscillators slowly get out of phase and then back to equal phase again. Just more subtle, of course.

I also think, that the long answer in the reddit sub is accurate.