GrandOrgue rev 1331 with demo organ posted on SF

[L.Palo]

Hi!

At last I've been able to download your re-worked demo version and test it...

It's definitely less hiss in the notes, but there are also things lost I think. It's hard to put the finger on exactly what it is, but some things don't seem right to me. Easy to spot is some loop mistakes in the lowest notes of Contrabasse, but they are also easy to fix. More important is that something has happened with the "room" of the notes, they have less space around them now. Some of the stops has got a more electric imitation character now, a bit synthetic (like for instance the Bourdon 8) - quite possibly a side effect of lowering the samplerate.

Some more voicing is also needed for instance with the Bombarde/Trompette joint in the pedal. I definitely must play a bit more with it but I'm not at all sure if this is the right way to go. I think that the screen (panel) layout should be cleaned into different styles on different panels - if a French model is the base imitation then the combinations should be replaced with those common for that style (and a new panel created with an English/American AGO style etc). I think that one should investigate if other means of fixing the hiss could be used (different samples, other noise reduction settings etc) too without loosing too much "real organ" samples feeling.

But this is just my first thoughts about it...

Kind regards

Lars P

 

[wehtam721]

Thanks for the comments about my revisions to the demo set. You're completely right. I agree that the character of the stops has been lost a little bit by lowering the sample rate. I notice it particularly when comparing the Flute Harmonique in each version. In my opinion, I would rather have the stops sound a little bit less authentic than have the hiss. At least on my computer with my system, the hiss was quite distracting. This is just my preference. Certainly a solution that offers the more of the original character of the recordings and still reduces the hiss would be preferable and I can keep looking into this. I just fear that it would be tough to do without increasing the size of the set significantly. What I've submitted certainly isn't a perfect solution.

Regarding the panels, I tried to place some of the elements on the main panel so that somebody who had never used the software before (and these are the people we would expect to be using the demo set) would instantly be able to see what other types of controls are available without having to look through the menu. If we want to introduce different panels with different styles, perhaps we could leave some of the user settable generals on the main panel, but move the pre-configured ones to their respective panels.

Regarding what I've chosen for the preset combinations, I'm completely open to changing those. I just filled some combinations in without much thought for any particular stylings. Any thoughts on better combination presets would be welcomed.

I'm going to be busy over the next couple of weeks, but I'll give it some thought and see if I can come up with anything better for the sound (as, in my mind, that should be first priority). If I come up with anything, I'll make a post here. In the mean time, any more suggestions that anyone might have would, as always, be helpful and welcome.

Take care,
Matt

 

[L.Palo]

With some attention to the voicing I think that the 22050 version could pass as usable for a starter even though it's not ideal. But hey, a sampleset of this size will always be about making lots of compromises. You win some, you loose some. A single stop of a high quality sampleset will be larger than this demo.

Maybe you're right that we should have a more general, functional, styling of the main screen (panel) and then have the different land specific styled versions on other panels as optional screens for the user to explore later. This is possible to do nowadays thanks to Martins work!

For the patch of Kalvträsk (that I'm still working on) mentioned before I re-cut the stop noises and I think that this could be done even a bit more so that the stop noise is quicker after activation.

Kind regards

Lars P

 

[wehtam721]

Hi everyone,

I have taken another look at the demo set and have some findings and a new version to report. My original assumption that the hiss was being generated by the pitch shifting seemed to be flawed as, upon further investigation, I was able to shift things 600 cents without introducing additional noise. There must have been extra noise in the original demo files that wasn't part of the Kalvtrask files. I reworked the files to eliminate the extra noise that had been introduced and went back to the original format of two samples per octave at a sample rate of 44,100 Hz. The hiss has still been eliminated and I think that more of the original character of the pipes has been retained.

I haven't done any additional work on the set yet as I wanted to see what others thought about the samples before I moved on. If it is agreed that this is the better route to take, I'll start making some of the other improvements that we've discussed previously. There are some inconsistencies between octaves due to the samples that were chosen (I may change this going forward if it sounds better), but I wanted to get some initial comments about the quality/character of the samples when you all have some time.

Again, you'll need to be running the most recent revision of GrandOrgue (rev 1341) in order for these sets to work as they require the bug fix which was introduced in this revision.

New Version:
http://mwoznicki.yourhda.com/GrandOrgue/demo.7z
http://mwoznicki.yourhda.com/GrandOrgue/demo.zip

As before, these downloads are the same, just provided in two different formats. The 7z file is about 38 MB and the zip file is about 55 MB. I think this may be a better starting point for continued revisions than the down-sampled version I had done previously. I'm curious to see what you all think.

Take care,
Matt

P.S. If you want to directly compare this version to the previous one, I have moved the previous version downloads to the following locations:

Old Version (22,050 Hz):
http://mwoznicki.yourhda.com/GrandOrgue/22/demo.7z
http://mwoznicki.yourhda.com/GrandOrgue/22/demo.zip

 

[e9925248]

I found a little time to test your new version. The balance between the stops is very much better.

Octave 2' has some volume jumps in the bottom octave. Montre 8' volume changes around c'', Plein Jeu around c'.
There is a beating between Plein Jeu and Cornet.

Uncorrectable problems (caused by the limited sample selection/size): Lowerst octaves of Bourdon 16' sounds like a Subbass and the bottom octave of Gamba 8' speaks very slowly.

The generals from the setter support multiple banks - while normal generals exactly store one setting. If you intended to use "banked" generals on the main panel, it is OK.

In my option, it is commitable. Before doing this, I want to wait for ACKs/NACKs/options from other GO users/developers.

PS:
I have come to the conclusion, the we need to change ChurchName for this update, as the structure has changed in a incompatible way. What about "GrandOrgue demo V1"?

 

[wehtam721]

Thanks for getting back to me about the new version. I was trying to respond all day yesterday, but it seemed as though the Magle servers were down.

I'm fine with changing the ChurchName of the demo organ to "GrandOrgue demo V1" and I agree that it would be nice to get some opinions from others before committing this version.

Regarding the volume/pitch issues mentioned above, I've fixed most of them (the beating between the mixtures seems to have been improved to my ears, but I'm not certain that it's perfect). I've updated the download links I gave for the new version (post #44 above) and I've also created a patch so that if you've already downloaded the new version of the demo you don't have to download it all again. The patch gives the new ODF (changes the ChurchName and makes the pitch/volume adjustments) and it also includes replacements for the two samples for the bottom of the Gamba 8' which speak much more like the rest of the rank. You can access the patch here:

http://mwoznicki.yourhda.com/GrandOrgue/demopatch(16-8-13).zip

If you're downloading the new demo for the first time, you can just use the new version links in post #44 of this thread. You won't need the patch. The patch has already been integrated into those downloads.

My thought on the generals was that I wanted the settable generals on the main page to be the same as those that you would find in the generals panel. This way, somebody new to GrandOrgue that wasn't aware that ODF generals and Setter Generals were different wouldn't be confused when the settings they programmed on the generals panel (for example) weren't available from the main panel (or vice versa).

Thanks again for your thoughts and opinions, I appreciate you taking the time to look it over. Let me know if you find anything else and I look forward to hearing the opinions of some others in the community as well!

Take care,
Matt

 

[e9925248]

I noticed, the triggering multiple stops at once can result in a strange sound (like a "shot" - eg. Montre 8'+Prestant 4'). Maybe we should also use the multi-release capatibilties and load other stop noise too.
We could also override a little bit shifted cue-points via ODF, so that (at least for the release) the same sample is never played simultanly two times.

 

[wehtam721]

I had noticed this as well. I hadn't thought about adjusting the release markers via ODF. That's certainly a possibility. I didn't include additional stop noises to help save disk space as I know that enlarging the demo organ isn't a favorable option. I'm hoping that I'll be able to make some recordings of some stop controls as they're activated by magnets. The sounds would be shorter and would hopefully use up less disk space allowing for each stop to have it's own stop noise recordings. Also, as the current console design doesn't have images that depict the old traditional style drawstops (like those in the Kalvtrask Church organ), these alternate stop sounds may be more appropriate. I welcome thoughts on this or other possible solutions.

 

[wehtam721]

For anyone that's interested, I made the recordings of the stop tabs that I have easy access to. Here's a link to a patch with the stop tabs that I recorded tonight along with a new ODF to implement them. I haven't decided how much I do or don't like them yet, but it's at something to try at least. The patch is available at the link below.

http://mwoznicki.yourhda.com/GrandOrgue/effectspatch(31-8-13).zip

If you give this a try, drop the .wav files here into the sounds folder in the demo organ. If you want to save the old sounds (based on the Kalvtrask sounds), it might be easiest to make a sub-folder for those first so that they don't get overwritten.

 

[e9925248]

As there are already multiple stop noises, my idea was to load multipe of them as multi-attack/release in the effect stop.
The ODF changes would be minimal (Pipe999AttackCount + Pipe999Attack) - GO would then select a attack/release randomly.

 

[L.Palo]

Finally I got some time to test the new improved demo version. The samples are quite nice. Possibly the Bourdon 8 needs something for the samples from d2 (074-D) and upward as the pitchshifting has both removed the reverb tail and also changed the character a bit too much perhaps. The stop noises should be brutally cut shorter (and possibly have a short fade in) to improve response speed.

I think that the odf needs some reworking. I'm personally not happy with the current voicing - the Hautbois is much too loud and the Flute harmonique is too soft compared to especially the Bourdon 8. The pedal feels very weak to me but as it uses a lot of borrowing one needs to be careful with how it's done, but I'd like a more bass heavy sound overall in the organ. But I sure can agree to disagree - the voicing is easily changed by the user.

The generals (presets) should also have different MIDI program change numbers that activate them by default (now they all respond on 1).

I'll see if I can create a new .organ file that solves the voicing issues I personally find bothering (preferrably not creating too many others in the process) and upload it for others to test.

Kind regards

Lars P

 

[wehtam721]

As there are already multiple stop noises, my idea was to load multipe of them as multi-attack/release in the effect stop.
The ODF changes would be minimal (Pipe999AttackCount + Pipe999Attack) - GO would then select a attack/release randomly.

I have also created an ODF for this implementation (attached here). I loaded each of the 7 stop noise samples for each drawstop. With GrandOrgue randomly choosing the sample to play, however, there's still a very high chance that the same sample would be chosen to be used if multiple drawstops were activated at the same time (from the setter or a general). Selecting only four stops/controls would already give a greater than 50% chance that at least two of the noises would be the same.

The stop noises should be brutally cut shorter (and possibly have a short fade in) to improve response speed.

I can try my hand at this, but I'm not sure when I may have the time.

I'm personally not happy with the current voicing

I will yield to your decisions on this (or the input of others) entirely as I know very little about this.

Take care,
Matt

 

[e9925248]

I have commited it: https://sourceforge.net/p/ourorgan/svn/1367/

 

[JLD]

Windows versions 1367 including demo organ are on SF.
I compiled with jack=OFF since I had difficulties finding all the required header files (thread posix headers).

JL

 

[Ghekorg7 (Ret)]

Thank you e and JL !

 

[e9925248]

Windows versions 1367 including demo organ are on SF.
I compiled with jack=OFF since I had difficulties finding all the required header files (thread posix headers).

I'm using the sources (tar.gz + patch) from
https://build.opensuse.org/package/show/windows:mingw:win32/mingw32-pthreads

PS: I have found another build:
http://sourceforge.net/projects/min...sted)/pthreads/pthreads-20100604.zip/download

 

[Ghekorg7 (Ret)]

Report

1367 works ok here on win8 x64 i7 3620M at 2.4~3.2GHz 16GB ram and SSD samsung 128GB. My God how fast things are now......
The demo organ is a real beauty !

 

[e9925248]

We currently have X stops, but only fewer stop noise samples. The current implementation has the drawback, that a sample can be selected two times.
So my newest idea:
Assign only one noise sample to each effect stop. Multiple stops will share a sample, but stops usually triggered together should use a different sample (if possible).
The first instance of a sample should sound normal. The second gets Pipe001TrackerDelay=25, the third (if really necessary) 50.

 

[e9925248]

We should replace the d''/e'' sample of Hautbois 8'. it does not sound like a reed (maybe more like a celeste).

 

[L.Palo]

Listen to the Oboe of Kalvträsk extended and pick a note you like better around that region and I can prepare it for closer comparison.

For the stop noise issue above I'd also use the possibility to re-tune the samples a little (both + and -) as well as the tracker delay. Perhaps also try different amplitude levels to maximize the difference.

Kind regards

Lars P