It is currently Mon Jan 18, 2021 5:50 pm


A Rule?

Hauptwerk software technical support only. Please make sure you have read the manual, tutorials and FAQ pages before requesting support.
  • Author
  • Message
Offline
User avatar

Doug S.

Member

  • Posts: 468
  • Joined: Tue May 20, 2008 8:17 pm
  • Location: Massachusetts USA

A Rule?

PostWed Dec 09, 2020 2:23 pm

When one goes from 16, 24, 48 to 96 bit resolution, is there a rule regarding the corresponding increase in ram needed?
Thanks,
Doug
Doug
Offline

mnailor

Member

  • Posts: 747
  • Joined: Sat Nov 30, 2013 5:57 pm
  • Location: Atlanta, GA

Re: A Rule?

PostWed Dec 09, 2020 2:40 pm

When loading ranks at 20 or 24 bit sample resolution, the memory used for the samples is doubled (32 bits used) compared to 16 bit resolution. Please see page 77 of the current 6.0.1 user guide for that topic.

Unrelated to the sample resolution, a sample *rate* of 96 kHz uses more CPU time than 48 or 44 kHz. I would assume some more memory for computation and buffering, but probably not as important as the extra CPU load.
Offline
User avatar

mdyde

Moderator

  • Posts: 12865
  • Joined: Fri Mar 14, 2003 1:19 pm
  • Location: UK

Re: A Rule?

PostThu Dec 10, 2020 4:34 am

Hello Doug,

To confirm and add to Mark's reply (thanks, Mark):

- Any difference in memory requirements between 48 kHz and 96 kHz is likely to be negligible (but it does roughly double CPU overheads, and thus halve the polyphony your computer will be likely to be able to achieve).

- Also note that having the 'high-definition pitch-shifting' organ preference enabled used a few bytes more memory for each voice generator than not having it enabled. Hence if your polyphony limit setting is set to a large value (which requires a correspondingly large number of voice generators) then it may add up to a small but possibly material difference in total memory requirements. (Having that option enabled also roughly doubles CPU overheads again, thus further halving the polyphony your computer will be likely to be able to achieve.)
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.
Offline

Ozykz

Member

  • Posts: 7
  • Joined: Tue Mar 21, 2017 1:16 pm

Re: A Rule?

PostFri Dec 11, 2020 7:00 pm

Martin,

Does Hauptwerk VI technically support native 96kHz samples? That is, sample files that are recorded in 96kHz, and played back in HW at 96kHz? Is there even a use case for that, or is it better to just downsample sample recordings to 48kHz.
Offline
User avatar

mdyde

Moderator

  • Posts: 12865
  • Joined: Fri Mar 14, 2003 1:19 pm
  • Location: UK

Re: A Rule?

PostSat Dec 12, 2020 4:13 am

Hello Ozykz,

Hauptwerk (v2+) does technically support 96 kHz sample sets. (I think only one was ever released, and that was many years ago, and long discontinued.) Please see this thread for my detailed reply on 96 kHz sample sets:

viewtopic.php?f=1&t=19209&start=15#p144339

mdyde wrote:
positive wrote:Would there be a benefit if the sets would be released at 96kHz ?


For Hauptwerk v6+, basically, no.

In more detail:

- Although 96 kHz could theoretically reduce interpolation distortion a little further, with Hauptwerk v6's new 'higher definition' pitch shifting option selected (which uses extremely high-quality interpolation), interpolation distortion should be well below the threshold that human hearing could detect anyway, even with very larger numbers of pipes sounding.

- Hence there would be no audible reduction in interpolation distortion compared to just using the 'higher definition' pitch shifting option with 48 kHz samples. (The interpolation used by the 'higher definition' pitch shifting setting was specifically designed to make further reductions in interpolation distortion so small as to be inaudible/unnecessary.)

- Also, using 96 kHz samples would roughly double RAM requirements, and would massively increase CPU demand further (probably more so than just using the 'higher definition' pitch shifting option).

- If the sample set producer were to use 96 kHz samples but *didn't* apply an anti-aliasing filter low enough (e.g. at around 24 kHz) within the sample files themselves, then you would effectively be re-introducing aliasing distortion for upward pitch-shifts. I.e. you would be undoing at least some of the main benefit of running the audio engine at 96 kHz.

- On the other hand, if the sample set producer were to use 96 kHz samples but *did* apply a high-quality off-line anti-aliasing filter low enough (e.g. at around 24 kHz) then you might as well just use 48 kHz samples instead (which could comfortably retain all frequencies within the human hearing range, and high-quality off-line anti-aliasing filters can be very good indeed and zero-phase, avoiding adverse effects on nearby frequencies within the human hearing range).

Hence, in summary, using high-quality 48 kHz samples (which have a had a high-quality anti-aliasing filter applied by their producer), together with the Hauptwerk 'higher definition' pitch shifting option and the Hauptwerk 96 kHz audio engine/output option should effectively give you all of the audible advantages, with none of the disadvantages of using 96 kHz samples (2x RAM, even more CPU demands, possible risk of re-introducing aliasing distortion).

Hope that helps!

[Edit: P.S. For Hauptwerk v2-v5, using 96 kHz samples could indeed have made an audible difference to perceived clarity due to the resulting, albeit fairly small, reduction likely in interpolation distortion. But for v6+, the new 'higher definition' pitch shifting option gives a massively greater reduction in distortion than using 96 kHz samples would, and all for about the same CPU overhead trade-off, and without the 2x RAM disadvantage.]
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.

Return to Technical support

Who is online

Users browsing this forum: No registered users and 2 guests