There we go, folks! The winner of the the digital technology competition is Audio2 = Johannus Live

Audio 1 is Physis, Viscount Sonus 60

Audio 3 is sampling, Viscount Prestige 80

Audio 4, out of the competition, remains a mystery, for the moment.

First of all, I want to heartily thanks Gianluca (@ahlborn), Andreas (@Ippenstein), and Jeff (@live3TA) for taking part in the competition and being such a great team! Thanks very much!

It has been an exciting (and formative) time for me as I had to find out a common audio processing ground for all the samples, a truly challenging task! I have learned a lot and would like to share now my thoughts. But before intoxicating everyone with my considerations, I share here the original files for you to understand where we started from and how we arrived at the final results.

The same editing (*) has been applied to all the audio samples:

First reverb: LiquidSonics Seventh Heaven Professional plugin replica of the more famous Bricasti M7

Second reverb (only tails): Vienna MIRacle - Algorithmic Reverb Add-On

Audio Spectrum Enhancer: Soundtheory Gulfoss

(*) Johannus received an extra treatment, more later.

It has been very difficult to find a common reverb setup for all, given the very different sonic nature of the audio material. Physis' sample was very clear and pristine, not jammed or saturated at all, and every type of reverb fitted. Same for the Prestige. Johannus' sample, instead was particularly dark and muddy and only one reverb of Seventh Heaven could be used, all the rest was making the whole thing taste like an overcooked risotto. To be honest, even with this "best fitting" reverb, things were not shining... and here comes the little "unfair" trick I applied to Johannus, a massive boost of +6 dB (with an extensive bell) around 4Khz to bring the organ alive, and -3 dB around 120 Hz to reduce the muddiness, otherwise it was feeling like listening to the organ from outside the church. Now, +6 dB implies basically doubling the volume in that area. Although this extra treatment might sound unfair when applied only to Johannus, I decided to leave it there because, when trying to correct the Sonus and the Prestige with equalisation, I could not remove the "unnatural" components that were there, so I could not produce any improvement.

Therefore, if we want to consider the possibility to improve, with external processing the sound out of the box of these instruments, Johannus is definitely the one that allows the most to be achieved. BUT! There is a huge but.

The editing that I applied to the audio samples can be considered as a powerful make-up. It is easier to achieve a good result on a nice face, no doubt. But what if we could physically change the features of the face and improve it even more? Well, with Johannus that is not possible. The samples are set, the output is stereo (c/c#), end of the story. This is not the case for Physis were everything can be set. And the parameters that can be changed can make a huge difference, if used in the correct way.

1 - the voicing: the factory defaults are good for Halloween to scare the children. Gianluca's voicing is very good but he was competing with Aristide Cavaillé-Coll in person, not easy eh! Still, a lot of things can be improved, especially with the high-pitched stops (mutations and mixtures) that can easily introduce unnatural artifacts that can drill through the most powerful TUTTI and screw the ensemble.

2 - the audio routing: Seventh Heaven is a powerful reverb with a lush and rich sound signature. It did a good job even with the stereo output. However, the sound image is static because, like every reverb, it reproduces the sound as if it is all originating in the same spot. This is not what happens in a real organ where the sound is originating in 3D: width, height, depth. In this case the audio routing of Physis allows to achieve a lively and credible reverb, similar to the wet sample sets of Hauptwerk, without the shortcomings of the sampled acoustics. It sounds more natural and consistent. This is the system I use on my three organs.

When the above two features are used in a smart way, the final result is definitely much cleaner and pristine than anything we could achieve with sampling (also because the sound of Physis is truly dry, without the sustained portion of the reverb in the loop).

This last point, about the "dryness" of the raw sound material, can however carry some disadvantages as "dry" can turn out to be "too dry". In this case I found out that applying a pre-reverb, made of only early reflection in a small space (wooden chamber with 0.4 seconds), can enrich the sound in a consistent way, as if the pipes are sounding in a real pipe organ chamber. In this case the chain would be: pre-reverb (early reflections) + main reverb (early and late reflections) + tail reverb (only late reflections with some modulation). I leave here attached an audio file (FinaleMP3) recently recorded on my Concerto 355cc where such a system has been applied to show the final result (and you will forgive, again, my sloppy playing).

Now to the results of the poll. The public poll has definitely been influenced by a cognitive (social) bias, being the answers known to everyone before the vote. I apologise for setting up the poll in the wrong way. However, the votes collected via private messages, have somehow suprised me. For me, the clear - very clear - winner is Audio 2, Johannus. The votes I collected, however, say something else. See the score grid here below. The files have been ranked from 1 to 3, with the first choice having a score of 3, the second a score of 2, and the third a score of 1. This helps to keep things in perspective:

hidden poll results.jpg - Bild entfernt (keine Rechte)

While the audio equipment used can influence, to a certain degree, the perception of the sound, the choices operated by the participants show that there is not a clear preference between Audio 2 and Audio 1, they share somehow 50% of the votes which is a result one might obtain by a random choice. This induced me to reconsider my hearing evaluation and makes me now very interested to understand what induced some people to choose audio 1 over 2 when, for me, there was really no game there.

I wrote too much, let us know your thoughts, comments, opinion and, maybe, ideas for a future contest. At this point, for me, it would be very interesting to take again some samples of Viscount and Johannus but operate them with all the features offered by both instruments. I think that the final result could be even more suprising than the one we have unveiled now, especially for the users of both brands who would like to greatly improve the performance of their instruments.

PS: sorry, I cannot attach the raw audio file before the editing because they are .wav, a file format not supported here :(

Zitat von Soli Deo gloria im Beitrag #4

---

Als Hauptwerker mit aktuellen Sets kann ich verschiedene Perspektiven fast beliebig mischen und mir zum Üben eventuell eine andere Kombination als zum endgültigen Spiel zusammenstellen. Blitzschnell und komfortabel mit Schieberegler. Ob das auch bei obigen Instrumenten geht, weiß nicht nicht.

---

Ciao Stephan,

the bias is unavoidable when the results are known before the vote but what makes a big difference for me is the picture emerging from the ranking of the choice vs the choice alone. All private messages came in with some qualitative evaluation, that I did not mention, and people stated their indecisiveness between the audio 1 and 2 even if they finally chose one of the two. Someone even ranked audio 2 as the worst and, therefore, the final score of the private survey returns a winner in a substantiated context. This is what surprised me. For me there was not doubt that audio 2 was the best, but we cannot say the same for the other voters.

Anyway, this “competition” arised from a discussion about which technology can provide the most credible organ sound and, for me, among the files I received, Johannus’ file provided the best performance. Is that the absolute best performance, when taken out of this context? No, and the file I posted with my first post on this thread shows, in my opinion, that more refined results can be achieved. Still, it is a very good result, achieved with little effort, nothing comparable to the amount of time and money I had to invest to get - what I deem to be a better result - with physis.

Finally, I think that sampling a specific instrument is certainly the fastest way to achieve a credible (specific) pipe organ sound. The problem of providing a credible sonic context with reverb remains, as long as only a stereo output is provided. It would be interesting to test the routing system of Physis with sampling technology and see what happens. I will do such a test with Hauptwerk where there is no limitation of channels per division. For this, however, I will need some time because the routing system of Hauptwerk is a labyrinth and maybe I will ask for help in the VPO section of this forum.

descriptive statistics can be very misleading when they are not put in context. I think we pay the price of it daily in our normal lives. But this is the topic for another thread, or maybe at dinner during the weekend ;)