Hauptwerk VI, 96kHz - Very loud popping/cracking noises.

[mdyde]

Hello Wouter,

The issue I mentioned with the harmonic-shaping filters is specific to 96 kHz. The high-definition pitch-shifting option wouldn't have any direct bearing on that.

Do you mean that under the same circumstances (stand-alone, ASIO, 96kHz, 1x1024) your 'audio pooping' issue still occurs at 48 kHz if the high-definition pitch-shifting option is enabled, but definitely doesn't occur at 48 kHz if the high-definition pitch-shifting option is disabled (with no other settings being changed, and even when just a few pipes are sounding)?

If so, how many GB does Hauptwerk display as 'free' on its 'Audio, MIDI and Performance' large control panel? (The high-definition pitch-shifting option uses a little more memory, and also stresses CPU caches a bit more, compared to not having it enabled.)

Hello Martin,
Thank you for your patience

Here is a list:

- Processor: AMD Ryzen 7 3700X 64GB DDR4

- Memory FREE !: 4GB

- Audio buffer HW and audio interface: 1024/1 STAND ALONE!

- 96 KHZ enabled: Audio In red sometimes / CPU 20%, polyphony 2000

-High Quality Pitch Shifting Enabled: Audio in red sometimes / CPU 20% polyphony 2000

- WINDMODEL DISABLED, HIGH QUALITY PITCH ENABLED, 96 KHZ ENABLED,
POLYPHONY 7000:
NO PROBLEMS!

[mdyde]

Thanks, Wouter.

(If I understand your reply correctly, you mean that you do still have the problem at 48 Khz if the high-quality pitch-shifting option is enabled.)

Please try:

1) Keep the sample rate set to 48 kHz, with stand-alone ASIO output and a 1x 1024 buffer.

2) Via the 'Organ | Load organ, adjusting rank/memory options' screen, load the smallest sample set that exhibits the problem, and temporarily disable all of its ranks except for one (e.g. the main 8' diapason on the main division).

3) Keep the wind model enabled for it for now.

4) Enable the high-definition pitch-shifting option.

5) Turn on just the one non-disabled stop/rank and see whether the problem occurs playing and holding a single key (e.g. middle C) on that stop. (If the problem doesn't occur with just a single key, instead try to reproduce it by normal playing, using just that one stop.) Assuming the problem does occur, use Hauptwerk's built-in audio recorder to record an example.

6) Also create a diagnostic file ('Help | Create a diagnostic file') at that point in time.

7) Now disable the high-definition pitch-shifting option. Turn on that same stop and verify that the problem definitely no longer occurs when playing and holding that same key (or playing the same piece). Also use Hauptwerk's built-in audio recorder to record an example (so that we can compare them if needed).

8) Turn the high-definition pitch-shifting option on again, but this time disable the harmonic-shaping filters. Turn on that same stop and see whether the problem still occurs when playing and holding that same key (or playing the same piece). Either way, also use Hauptwerk's built-in audio recorder to record an example (so that we can compare them if needed).

Please then:

- Let us know whether you were able to reproduce the problem in each of steps 5, 7 and 8 above.
- Send your three audio recordings and the diagnostic file to Francois at support [at] hauptwerk.com .
- Let us know which stop/rank you used for the tests.

If needed, we'll then see whether we can get a copy of the relevant sample set from its maker so that we can try the same tests here to see whether we can reproduce the issue with it. Thanks.

Thanks Martin!

Sorry for the confusion, but it only occurs with the 96 kHz setting. Apologies !!!

Kind regards,
Wouter

[mdyde]

Thanks very much, Wouter.

No problem at all.

I've been working on the problem with the harmonic-shaping filters at 96 kHz today, and found out that it was actually much more minor than I had first thought. A work-around is simply to re-load the organ after changing the sample rate setting. I've edited my previous post to clarify that:

It turns out that there is a bug in Hauptwerk v6.0.0/6.0.1 in that the harmonic-shaping ('brightness') filters used by the wind model (and tremulants, and 'brightness' adjustments via the voicing screen and organ definition) were not being recalculated if, with an organ loaded, you then changed the sample rate without re-loading the organ, potentially resulting in some frequencies being emphasized a bit more more or less strongly when those filters are used prominently (and perhaps giving strange sounds, especially for very low or very high notes). (The filters will, however, be correct after loading any organ.) My apologies for that bug, which will be fixed for the next Hauptwerk v6 patch.

The work-around, if you have an organ loaded and wish to change the sample rate, is simply to ensure that you then re-load the organ.

(I will also improve the response of the harmonic-shaping filters for the the lowest notes on a 32' rank when used at 96 kHz.)

Does that work-around (re-load the organ after changing the sample rate setting) eliminate your popping/cracking problem too?

[mdyde]

Hello again Wouter/Ozykz,

A further update:

(Irrespective of VST/AU) and aside from currently (v6.0.1) needing to ensure that the organ is reloaded after changing the sample rate setting (which is fixed for the forthcoming v6.0.2), I have actually also now been able to reproduce some instabilities/distortion/popping in the pipe speech of very low notes (e.g. 16' C and below) at 96 kHz. It occurs if the harmonic-shaping/brightness filters are used strongly for those very low-frequency pipes (e.g. by having the brightness voicing control, or corresponding parameter in the organ definition file, turned up or down a lot, or if the wind model or tremulants in the organ definition are set to modulate brightness strongly for the pipes) to an extent that depends on the frequencies present in the pipes (including any remaining noise/artefacts in the samples) and the pipes' speech stability.

We'll need to think about how best to address that for v6.0.2, but if you have the problem the work-arounds would be either:

a) Simplest: stick to using 48 kHz for now. Or:

b) If using 96 kHz, for affected sample sets, via the 'Organ settings | Organ preferences | Audio engine' screen tab, tick 'Disable harmonic-shaping filters'. (However, then the wind model and modelled tremulants will sound less realistic for all of the pipes, and brightness voicing controls will no longer work). Or:

b) If using 96 kHz, via the voicing screen, for any affected pipes lower than about 8' C (e.g. for the bottom octave on 16' ranks, or the bottom two octaves on 32' ranks), turn:

- 'All perspectives: wind supply mod: brightness (pct)' down to zero.
- 'All perspectives: tremulant mod: brightness (pct)' down to zero.
- If necessary, also set 'All perspectives: overall: brightness (dB)' to zero (or below or above zero if needed, since the sample set producer might have turned the brightness up or down with the organ definition for those pipes).

My apologies that the issue didn't get spotted during development or testing. (It turns out that it can actually also occur at 48 kHz, e.g. also in Hauptwerk v2-v5, but it's then only potentially applicable for pipes below 16' C, and thus unlikely to be an issue in practice at 48 kHz.)

Hi Martin,

Thank you for your hard work on this.
Your explanation makes sense because I recall running into the issue when using a quiet 32' stop and some strings and flutes, which was nowhere near full organ, and quite a low polyphony.

I've done a lot of similar testing to Wouter, but not since I made this post since the organ is at a customer's location, and many of Wouter's experiences are similar to mine,

For example, the computer has approximately 30GB of free ram, so I can rule that out.
I have had buffers at 1, and buffer sizes of 2048 across the board.
I confirmed running as admin did not make any change.
I tried playing with the thread priority and behavior settings in Reaper, with no change.

I haven't had the chance to test any other sample sets.
Can confirm VST is 64bit
Did not yet test turning off wind model,

I did not do any extensive voicing, but the sampleset has many extensions, of which there are a number of borrows, so possibly those borrows have had major revoicing done in the ODF

The organ is running at 48kHz at the moment so this problem isn't critical for me. I will make another trip to my customer sometime in the next few months, and will do more testing then. Hopefully we'll have the fix by then and I will confirm for sure whether or not the problem is resolved.

Hello again Wouter/Ozykz,

A further update:

(Irrespective of VST/AU) and aside from currently (v6.0.1) needing to ensure that the organ is reloaded after changing the sample rate setting (which is fixed for the forthcoming v6.0.2), I have actually also now been able to reproduce some instabilities/distortion/popping in the pipe speech of very low notes (e.g. 16' C and below) at 96 kHz. It occurs if the harmonic-shaping/brightness filters are used strongly for those very low-frequency pipes (e.g. by having the brightness voicing control, or corresponding parameter in the organ definition file, turned up or down a lot, or if the wind model or tremulants in the organ definition are set to modulate brightness strongly for the pipes) to an extent that depends on the frequencies present in the pipes (including any remaining noise/artefacts in the samples) and the pipes' speech stability.

We'll need to think about how best to address that for v6.0.2, but if you have the problem the work-arounds would be either:

a) Simplest: stick to using 48 kHz for now. Or:

b) If using 96 kHz, for affected sample sets, via the 'Organ settings | Organ preferences | Audio engine' screen tab, tick 'Disable harmonic-shaping filters'. (However, then the wind model and modelled tremulants will sound less realistic for all of the pipes, and brightness voicing controls will no longer work). Or:

b) If using 96 kHz, via the voicing screen, for any affected pipes lower than about 8' C (e.g. for the bottom octave on 16' ranks, or the bottom two octaves on 32' ranks), turn:

- 'All perspectives: wind supply mod: brightness (pct)' down to zero.
- 'All perspectives: tremulant mod: brightness (pct)' down to zero.
- If necessary, also set 'All perspectives: overall: brightness (dB)' to zero (or below or above zero if needed, since the sample set producer might have turned the brightness up or down with the organ definition for those pipes).

My apologies that the issue didn't get spotted during development or testing. (It turns out that it can actually also occur at 48 kHz, e.g. also in Hauptwerk v2-v5, but it's then only potentially applicable for pipes below 16' C, and thus unlikely to be an issue in practice at 48 kHz.)

[mdyde]

Thanks very much, Ozykz.

Yes -- probably the organ definition you're using has the voicing brighness adjusted within it for the 32' octave, which was making the problem occur (and/or perhaps its wind model was configured to modulate the brightness of those pipes strongly).

For v6.0.2 I'm working on tweaking the response of the filters for very low pipes (particularly for 96 kHz) so as to make sure that it can't happen.

Hi - I had the same problem. It was on the Obervellach sample set with only a few stops on. I had just upgraded to VI and enabled the 96khz option. The hardware has 128gb of RAM - thus no shortage of memory. After reading this post I put it back to 48khz and so far no popping sounds.

This problem is very good to see and hear at Richard MC Veigh's You Tube Channel and site " I upgrade Hauptwerk to version 6, not entirely smoothly!! | My Hauptwerk Setup". It comes when he goes to 96khz.

[mdyde]

Hello JRP/münsterorganist,

I'm still working on it, but in the meantime if you do want to use 96 kHz and encounter the problem with some ranks [it only potentially affects the bottom octave of some 16' ranks, and the bottom two octaves of some 32' ranks, depending on the samples, settings in the organ definition, and voicing), it can be worked around via one of the methods I suggested, e.g. this is sufficient and has negligible disadvantages (especially since tremulants hardly affect pipes' 32' and 16' octaves anyway):

b) If using 96 kHz, via the voicing screen, for any affected pipes lower than about 8' C (e.g. for the bottom octave on 16' ranks, or the bottom two octaves on 32' ranks), turn:

- 'All perspectives: wind supply mod: brightness (pct)' down to zero.
- 'All perspectives: tremulant mod: brightness (pct)' down to zero.

Hi Martin, I'm not sure if the following will help track down this 96kHz problem, but the following is occuring on my system. I noticed that making the 96kHz adjustment under the Organ Preferences, Audio tab makes no difference to the 'crackling'. If, however I make the adjustment from 96 to 48 under the General/Audio Devices tab, the the crackling goes away.

I have also noticed that I have the 'crackling' coming from just 1 division (my Choir division) regardless of the organ I have loaded. Further, I can have just 1 stop drawn (at any pitch including 8' and 4'), and play just 1 note, and the cracking is there. At first I thought that I might have an issue with my audio device, but the 'crackling' is only on 2 of the 6 stereo channels, and it follows the cyclical 'Bus Allocation Algorithm' which adjusts every octave, so that tells me it is not my Audio Device.

Hope this helps narrow the issue down, as sound is very nice with the 96 activated. Thanks for all your work on making HW better for us.

Regards,
Mel Piercey.

[mdyde]

Thanks very much, Mel, but no need for any additional diagnostic information at this point -- I know why it's happening and am working on a solution.

For any affected ranks (generally only the bottom octave of some 32' ranks), you can work around it at 96 kHz in the way I mentioned previously:

in the meantime if you do want to use 96 kHz and encounter the problem with some ranks [it only potentially affects the bottom octave of some 16' ranks, and the bottom two octaves of some 32' ranks, depending on the samples, settings in the organ definition, and voicing), it can be worked around via one of the methods I suggested, e.g. this is sufficient and has negligible disadvantages (especially since tremulants hardly affect pipes' 32' and 16' octaves anyway):

b) If using 96 kHz, via the voicing screen, for any affected pipes lower than about 8' C (e.g. for the bottom octave on 16' ranks, or the bottom two octaves on 32' ranks), turn:

- 'All perspectives: wind supply mod: brightness (pct)' down to zero.
- 'All perspectives: tremulant mod: brightness (pct)' down to zero.

Also, just to clarify:

Hi Martin, I'm not sure if the following will help track down this 96kHz problem, but the following is occuring on my system. I noticed that making the 96kHz adjustment under the Organ Preferences, Audio tab makes no difference to the 'crackling'. If, however I make the adjustment from 96 to 48 under the General/Audio Devices tab, the the crackling goes away.

... in Hauptwerk v6 there's only actually one place that you can change the sample rate (48 vs. 96 kHz), which is on the 'General settings | Audio device ...' screen. The 'higher definition pitch-shifting' option on the 'Organ settings | Organ preferences | Audio engine' screen is independent, and doesn't affect the sample rate, and isn't relevant to the crackling/popping problem.

Thanks for this temporary solution Martin! The popping and cracking is no more!

[mdyde]

Thanks, Wouter. Excellent.