It is currently Sat Jul 24, 2021 6:56 pm


bad sounding sample sets because too perfect

Existing and forthcoming Hauptwerk instruments, recommendations, ...
  • Author
  • Message
Offline
User avatar

Doug S.

Member

  • Posts: 487
  • Joined: Tue May 20, 2008 8:17 pm
  • Location: Massachusetts USA

Re: bad sounding sample sets because too perfect

PostFri Jun 04, 2021 1:13 pm

Thanks Martin.
Doug
Offline

Theorbe

Member

  • Posts: 70
  • Joined: Sat Jan 11, 2020 7:10 am

Re: bad sounding sample sets because too perfect

PostSat Jun 05, 2021 1:29 pm

kasterling wrote:Andy, all these reports coming in about the success of your utility with various sample sets is really exciting. I was particularly thrilled to hear the benefits to the Armley Schulze set, as it is one that I own and think very highly of. I can only hope that your efforts will eventually turn toward making your utility compatible with HW4.2. As I am unable to switch to the later versions, I am waiting with eager anticipation for the release of your utility in a version compatible with what I have. I know many others reading this feel the same thing. Three cheers to you for stirring such excitement!

Kevin


Thanks very much, Kevin. I will definitely make my app compatible with HW4.2 before too long.

General Update - Version 1.2 is now available which includes the optional limit for the random values and some other fixes. As usual, please read the ReadMe file and the updated user guide. Here is the link again:

https://1drv.ms/u/s!ArppdH7JT_uasCD2SWa ... r?e=X1FVxe

Regards

Andy
Offline

kasterling

Member

  • Posts: 19
  • Joined: Thu Feb 20, 2020 7:04 pm

Re: bad sounding sample sets because too perfect

PostSat Jun 05, 2021 6:57 pm

Thank you, Andy!
Offline

Theorbe

Member

  • Posts: 70
  • Joined: Sat Jan 11, 2020 7:10 am

bad sounding sample sets because too perfect

PostFri Jun 11, 2021 9:08 pm

Version 1.3 of my random detuning app is now available. It’s now compatible with HW 4. I decided that because the Mac version is still some while away and the changes required for HW 4 are significantly less than the work for the Mac it was worth spending a few days on the HW 4 version. When the Mac version is ready it will also support HW4.

For those who might not have been curious enough to open the “About” box, there is a donation button if you so wish. Thanks.

Here is the link again for the OneDrive folder containing the app and user guide etc:

https://1drv.ms/u/s!ArppdH7JT_uasCD2SWa ... r?e=X1FVxe

Thanks.

Regards

Andy
Offline

Wall.e

Member

  • Posts: 19
  • Joined: Sat Sep 09, 2017 1:19 pm

Re: bad sounding sample sets because too perfect

PostMon Jun 14, 2021 3:56 am

Thank you very much! That's a fantastic feature and improvement for all HW4 users! I never had better sounding trumpets.
Offline

Theorbe

Member

  • Posts: 70
  • Joined: Sat Jan 11, 2020 7:10 am

Re: bad sounding sample sets because too perfect

PostTue Jun 15, 2021 4:26 pm

Wall.e wrote:Thank you very much! That's a fantastic feature and improvement for all HW4 users! I never had better sounding trumpets.


Thanks, Wall.e. You're most welcome.

Best regards

Andy
Offline

kasterling

Member

  • Posts: 19
  • Joined: Thu Feb 20, 2020 7:04 pm

Re: bad sounding sample sets because too perfect

PostTue Jul 20, 2021 7:05 pm

To all users of Andy's HW Auto Detune utility: I suggest we regularly start posting and compiling the de-tuning parameters we have found useful for the various sample sets we have applied the utility to. Some parameters have been suggested here already for the Standard Deviation:
St. Maximin: 1.0-1.2
Armley Schulze: 2.0
Caen: 3.2-3.5.
I have been experimenting with 2 other sample sets on a friend's HW4.2 installation.. Here are my suggestions for the Standard Deviation:
Doesburg, Martinikerk Walcker Organ: 1.0
Utrecht - Dom, Bätz Organ: 1.4.
In both cases the effect of the utility was to make the organ sound more spacious and colorful. In the case of the Doesburg organ the de-tuning effect was best not applied to the mixtures and the cornets. Otherwise, they sounded too far out of tune and a little weird.
The other thing I will report here--since I was asked by my friend to voice each instrument rank by rank--is that these sample sets required rather a lot of work to get right. I had to make a lot of amplitude adjustments particularly to the bass octaves of the 16' ranks, but elsewhere as well. Generally the bass octaves were too loud. It took me about 3 hours to go through the Utrecht organ and about 5 hours to go through the Doesburg organ rank by rank. The finishing touch was to apply Andy's utility to the fully-voiced instruments.
I hope this is somewhat helpful.
Offline

1961TC4ME

Member

  • Posts: 3067
  • Joined: Mon Feb 07, 2011 1:45 pm
  • Location: Lake Minnetonka, Minnesota

Re: bad sounding sample sets because too perfect

PostWed Jul 21, 2021 10:49 am

I agree with your suggestion for the settings for St. Max, right around 1.0 seems about right, anything more does it no favors. I surmised earlier it's probably due to its meantone tuning. As for Armley Schulze I've found 1.2 to 1.4 to be about as far as I care go. I've tried 2.0 to 2.5 with Armley Schulze and although it does make for interesting results, certain notes are way out of tune (Like in a Trumpet Tune as an example) and they pretty much jump out at you, it's just too far out of tune to my ears.

One thing I recently thought of and was going to ask is: If we're using a utility like this, would it be advisable to then disable the native de-tuning in Hauptwerk? As of now I have not so far disabled it, so am using both. Perhaps by disabling the native feature this is where some can get away with the higher de-tuning figures?

Marc
Offline

Theorbe

Member

  • Posts: 70
  • Joined: Sat Jan 11, 2020 7:10 am

bad sounding sample sets because too perfect

PostWed Jul 21, 2021 5:56 pm

Hi Marc and All

I definitely recommend setting HW’s random pipe detuning to 0% to ensure it is not having any additional effect.

In v1.2 of the app I added the detuning limit to prevent pipes sounding unpleasantly out of tune. Given that this limit value can make quite a difference to the result, you might want to include it in your recommendations / comparisons.

Although I don't necessarily recommend it, you can achieve an almost Uniform Distribution of random numbers (instead of the more natural Normal Distribution) by using the maximum “Standard Deviation” value but with a low limit value.

Personally, I think the result can sound a bit artificial but you might like to try it out and see what you think. Actually, the result sounds to me very much like the random detuning built in to the new / updated sample sets but perhaps that's just my (biased) imagination.

Thanks.

Andy
Offline

kasterling

Member

  • Posts: 19
  • Joined: Thu Feb 20, 2020 7:04 pm

Re: bad sounding sample sets because too perfect

PostWed Jul 21, 2021 10:29 pm

Andy, I am very curious how your utility manages to apply the same de-tuning parameters to multiple perspectives of the same rank; for example, to the Chancel, Nave, and Surround perspectives of an 8' Principal. That is, if the user selects the "Natural" de-tuning pattern when running the utility. The different sample set makers do not necessarily refer to the multiple perspectives using the same terms. What one calls "Chancel" another one calls "Close." How on earth did you manage to sort out these different terms and enable the utility to collate all the various perspectives (where they exist) of each rank? This seems to me a very significant accomplishment and one that the native de-tuning function in Hauptwerk cannot as of yet duplicate. I know you prefer the added chorus effect of de-tuning the different perspectives differently, but many users do not share that preference (myself included). So we are delighted to be able to keep different perspectives in tune with each other.

I suspect this is one of the reasons you recommend setting Hauptwerk's built-in randomized de-tuning to zero. Otherwise, it will put the different perspectives out of tune with each other. I imagine it will also affect the de-tuning results of your utility in unpredictable ways. So there would seem to be no advantage to allow it to operate. One could also disable the wind modeling and randomizations associated with it, but as Martin has made clear in this thread, the wind modeling really has little effect on the tuning.
Offline

Theorbe

Member

  • Posts: 70
  • Joined: Sat Jan 11, 2020 7:10 am

bad sounding sample sets because too perfect

PostThu Jul 22, 2021 7:39 am

Hi Kevin

Absolutely spot on regarding HW’s built in random detuning. I always disable the wind model anyway as I find the CPU cost too high.

Regarding how I manage to sync detuning across the perspectives, that was indeed a challenge especially as it appears to be a free-for-all in terms of how ranks are listed in HW’s voicing dialog. After trying different methods, I finally settled on using a list of prefix names (which can be added to in the app’s settings) which I move to a suffix instead. Additionally, some sample sets like Nancy have a numeric prefix which I have to remove to get the required result - the rank list sorted by rank name - and hence the perspectives grouped together.

If the sample set rank list has the perspective name as a suffix, or is just stereo, then I don’t need to re-order.

If applicable, I can then group all rank perspectives together where the rank name up to the first open bracket (if present) is the same.

Thanks.

Best regards

Andy
Previous

Return to Hauptwerk instruments

Who is online

Users browsing this forum: No registered users and 21 guests