HW Advanced Voicing Parameters

[Jcvtpo]

I tried this over in the Tech Support forum and wasn't able to find an answer, perhaps this will get more eyeballs on this forum:

I am trying to better understand how the following functions in the voicing menu affect the sound of any given pipe. I've read the Hauptwerk user guide but I struggle to understand the effect of these parameters, and I've also tried the "adjust and see what happens" approach. Could someone possibly provide a simpler explanation of each of these parameters?

 Lo/Hi EQ: transition frequency (kHz).
 Lo/Hi EQ: transition width (pct of trans freq).
 Lo/Hi EQ: high frequency boost (dB).

Thank You!

[RichardW]

I admit that I didn't know and have never used that feature but looking at the manual I have gleaned the following and will try to explain it.

Hauptwerk has a filter installed that will affect the low frequency part of the pipe's spectrum differently than the high frequency end.

The High Frequency Boost will amplify the high frequency end if the numbers are positive and (guessing because I am not near a HW machine) negative numbers will also be possible that will attenuate the high frequency end.

Now, where does the high frequency end of the spectrum start and where does the low frequency part end? The basic setting is determined by the Transition Frequency setting. So you could adjust that so that all the harmonics will be boosted or just the higher ones.

The final part is the Transition Width. If the Transition width were zero (NB Not recommended) then the filter characteristic would look like a step. One amplitude to the left (i.e. the low frequency part of the spectrum) and another to the right (i.e. the high frequency part of the spectrum). However, the Transition Width will typically not be a step but a ramp. The change will be gradual over a range of frequencies. The final control adjusts how steep that ramp is.

I hope that helps.

Regards,

[adrianw]

Well, two of them are fairly straightforward.

"Transition frequency": the cut-off frequency. Nominally, frequencies above this are affected (cut or boosted) by the other two parameters. (I suspect this may actually be the centre frequency of the filter slope).

"High frequency boost": the decibel boost (or if negative, attenuation) of frequencies above transition frequency.

I expect you have struggled (as I did) with the third. The "transition width" seems to me to actually work reciprocally to what is implied by the description "(pct of trans freq)". Whilst RichardW describes what one might expect to see, it is actually the opposite. On the voicing screen it is calibrated 0-200 and determines steepness of filter cutoff. 200 is actually the steepest possible, 0 is flat.

You can get a good idea of what is going on by setting an entire rank to a transition frequency of 0.25 (roughly middle C, 261Hz) and the boost or cut to +/-24dB. Then alter the transition width to get a feel for how it affects low notes vs high notes.

Edit: If you tell us what you are trying to achieve, we may be able to be more helpful. However, as a rule of thumb altering the brightness setting (which affects the relative harmonics of each pipe) is much the easiest way to alter pipe timbre. These filter settings are much less frequently used, but can be useful for example to correct undesirable microphone, acoustic or enclosure characteristics or to minimise extraneous wind or action noises.

[ppytprs]

I admit that I didn't know and have never used that feature but looking at the manual I have gleaned the following and will try to explain it.

Hauptwerk has a filter installed that will affect the low frequency part of the pipe's spectrum differently than the high frequency end.

The High Frequency Boost will amplify the high frequency end if the numbers are positive and (guessing because I am not near a HW machine) negative numbers will also be possible that will attenuate the high frequency end.

Now, where does the high frequency end of the spectrum start and where does the low frequency part end? The basic setting is determined by the Transition Frequency setting. So you could adjust that so that all the harmonics will be boosted or just the higher ones.

The final part is the Transition Width. If the Transition width were zero (NB Not recommended) then the filter characteristic would look like a step. One amplitude to the left (i.e. the low frequency part of the spectrum) and another to the right (i.e. the high frequency part of the spectrum). However, the Transition Width will typically not be a step but a ramp. The change will be gradual over a range of frequencies. The final control adjusts how steep that ramp is.

This is not quite right. These controls make up a parametric equaliser. If you imagine a graph of frequency vs amplitude, it creates a hump, or conversely a dip. The boost controls the size of the hump, the frequency, the frequency of the centre of the hump, and the width, the width of the hump (also known as Q in equalisers) so that you can make broad adjustments, or very fine ones.

[NickNelson]

This is not quite right. These controls make up a parametric equaliser. If you imagine a graph of frequency vs amplitude, it creates a hump, or conversely a dip. The boost controls the size of the hump, the frequency, the frequency of the centre of the hump, and the width, the width of the hump (also known as Q in equalisers) so that you can make broad adjustments, or very fine ones.

Although I have not used these adjustments either, the HW user guide seems pretty unambiguous about the functions of the sliders under discussion:

"The Lo/Hi EQ adjustments allow you to apply a simple low-pass or high-pass parametric EQ (filter) to the pipe, adjusting its balance between bass and treble. The transition frequency specifies the center-point between the bass and treble bands. The transition width determines the steepness of the frequency response curve in the transition region between the bass and treble, and is specified as a percentage of the transition frequency."

I would agree with the first replies.

Nick

[adrianw]

There are no "humps" in the HW Lo/Hi EQ filter. It is not that sort of parametric equaliser.

The OP had read the manual and I expect he understood perfectly well what it said. I don't think he was asking others, who have not used this setting, to read and explain the manual to him. He was probably confused, as I was, by the fact that, although unambiguous it is (unusually for the HW manual) actually incorrect in one detail of its description of what he would have discovered the software actually does, which is what I tried to explain.

If you need to use this setting - and frankly most users will never need to - I repeat my advice to do the experiment first . Select an 8' flute rank, which has a good approximation to a pure fundamental. Adjust the entire rank to a 0.25kHz transition frequency, the entire rank to a huge boost or cut (so you can actually hear the difference easily!) and then alter entire rank's transition width. Experiment with different transition widths by doing a glissando up and down the full length of the keyboard.

[ppytprs]

The following image shows Hauptwerk's output displayed on the spectrum analyser in Logic. No voicing controls applied.



In the next one, with 24dB boost at 2k. See the hump...



With the transition width high, wider hump:



With the transition width low, narrow hump:



What did surprise me was that as I made the width lower, pretty close to the bottom, the note got louder, which wouldn't happen with a normal EQ. Not sure what's going on there. But on the whole, it behaves as one would expect a parametric equaliser to behave.

[RichardW]

The following image shows ...

Thanks for the spectra. They does show pretty well what I expected. There is no "hump" the filter has just raised the high frequency end. The fact that it rolls off is not the filter characteristic but the pipe characteristic.

They also confirm what adrianw said about the "width" control working counter-intuitively.

The "transition width" seems to me to actually work reciprocally to what is implied by the description "(pct of trans freq)".

I just wished my mind-reading was as good as his.

Regards,

[ppytprs]

The following image shows ...

Thanks for the spectra. They does show pretty well what I expected. There is no "hump" the filter has just raised the high frequency end. The fact that it rolls off is not the filter characteristic but the pipe characteristic.

They also confirm what adrianw said about the "width" control working counter-intuitively.

No, the filter has absolutely not just raised the high frequency end. It's raised 2kHz (the set frequency). How wide the section it raises is then defined by the width. And the bigger the width number, the wider the band raised (image 3), the lower the width number, the narrower the band raised (image 4). Doesn't seem counter intuitive to me.

I'd have illustrated with a lower frequency, but the representation wasn't as smooth in that area of the graph, so more difficult to see. Logic's not really the perfect tool for this!

[jkinkennon]

To put things simply, the transition frequency is the center of the hump in a parametric equalizer. The transition width is the Q of the filter which makes the hump (or valley) a narrow or wide hump. The frequency boost (or cut) is the height of the hump (or valley).

However, the title Lo/Hi EQ suggests that this intended to be a shelf filter and this is likely the case. Some graphs with a swept input (or noise) would clarify the matter. I guess one could replace a pipe sample with a white noise sample and repeat the test. In this case the transition frequency would be at the center of the shelf's rise.

[NickNelson]

I guess one could replace a pipe sample with a white noise sample and repeat the test.

You would think so, but I tried to do just this earlier today and failed. HW refused to load the .wav (actually it was brown noise). I suspect that there is some subtlety to the wav files (looping maybe) that I haven't fathomed.

Nick

[Eric Sagmuller]

I had unsuccessfully tried to use these controls in the past to boost the fundamental of an 8' principal. I for one was confused about their specific functions.

Eric