Accidose. Sorry if this isn't the right place but I just started dicking around with the latest alpha after years of just using the last stable release, and I have to thank you for the 6db bandpass filter option, which really opens Xhip up for drum synth sounds.
Particularly, filter impulse trigger (808 style) kicks sound great on this thing, and it's surprising how few synths (even commercial) offer this functionality. In fact this is the only freeware VST that I know of that allows a spike to be fed into a modulated high-Q 6db bandpass filter, to make proper 808 kicks. The envelope curvature happens to be perfect for it, too.
There's even a keyboard-follow function on the envelopes to keep the decay in proportion, when doing to make subbass basslines. You've made me a happy man.
Xhip vs Analog Hardware
-
- KVRAF
- 10588 posts since 13 Jun, 2004 from Alberto Balsam
-
- KVRAF
- 12555 posts since 7 Dec, 2004
That's actually been a feature since before Xhip was released as a VST, even way back when it was a synth inside a tracker around 1999.
The filter is a KHN filter often referred to as "state-variable". The structure itself is called "KHN" after the authors of a paper on the structure much like "Sallen-Key", "Tow-Thomas" and so on.
"State-variable" is a somewhat generic mathematical term which applies to (almost?) every type of filter and so using it as a name doesn't lend much distinction to the different structures.
The "trigger" input is actually sort of like a "filter sync" similar to "osc sync" rather than just a signal input to the filter.
You should read the Xhip manual:
http://xhip.net/synth/releases/xhip_8_user_manual.pdf
The manual is 90% completed and still needs a bit of fine-tuning or a little detail added here and there but is otherwise likely to answer almost every question you might have an fill in a lot of gaps too.
The latest version of the alpha has a lot of small changes, mostly code clean-up. One new feature is the ability to load/save "state" files which are like a .fxb.
6 dB band-pass is actually very common. The Sallen-Key and KHN and Tow-Thomas band-pass modes are all only 6 dB slope. The Sallen-Key high-pass is also only 6 dB slope although the KHN high-pass is 12 dB.
I don't think Mutant minds us using his thread for random talk about Xhip. There isn't really any other "official" place to do so unless you wanted to PM or email me.
Let me know if you run into any issues with the alpha. I'm trying to finish up the missing menu implementation stuff (add CC, save/load CC map, replace PCM, remove PCM, ...) and I'll release another alpha once that's done.
edit: update manual link again.
The filter is a KHN filter often referred to as "state-variable". The structure itself is called "KHN" after the authors of a paper on the structure much like "Sallen-Key", "Tow-Thomas" and so on.
"State-variable" is a somewhat generic mathematical term which applies to (almost?) every type of filter and so using it as a name doesn't lend much distinction to the different structures.
The "trigger" input is actually sort of like a "filter sync" similar to "osc sync" rather than just a signal input to the filter.
You should read the Xhip manual:
http://xhip.net/synth/releases/xhip_8_user_manual.pdf
The manual is 90% completed and still needs a bit of fine-tuning or a little detail added here and there but is otherwise likely to answer almost every question you might have an fill in a lot of gaps too.
The latest version of the alpha has a lot of small changes, mostly code clean-up. One new feature is the ability to load/save "state" files which are like a .fxb.
6 dB band-pass is actually very common. The Sallen-Key and KHN and Tow-Thomas band-pass modes are all only 6 dB slope. The Sallen-Key high-pass is also only 6 dB slope although the KHN high-pass is 12 dB.
I don't think Mutant minds us using his thread for random talk about Xhip. There isn't really any other "official" place to do so unless you wanted to PM or email me.
Let me know if you run into any issues with the alpha. I'm trying to finish up the missing menu implementation stuff (add CC, save/load CC map, replace PCM, remove PCM, ...) and I'll release another alpha once that's done.
edit: update manual link again.
Last edited by aciddose on Mon May 11, 2020 7:08 am, edited 2 times in total.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
-
- KVRAF
- Topic Starter
- 3496 posts since 9 Oct, 2004 from Poland
Thanks for the 18.Oct.16 update.aciddose wrote:The latest version of the alpha
It is OK.aciddose wrote:I don't think Mutant minds us using his thread for random talk about Xhip.
[====[\\\\\\\\]>------,
Ay caramba !
Ay caramba !
-
- KVRAF
- 12555 posts since 7 Dec, 2004
Actually I should probably post this here in case people check this thread for updates:
Anyone who had issues loading the alpha in certain hosts in the past should check the latest version. I've fixed a few issues (three or four I think) which could have led to that problem.
If it still doesn't load, no need to post about it unless it's a host I'm not aware of. I can add it to the list if so (so just PM or email me) but otherwise I'll just have to test such hosts directly during the final RC bug-fixing stage.
Anyone who had issues loading the alpha in certain hosts in the past should check the latest version. I've fixed a few issues (three or four I think) which could have led to that problem.
If it still doesn't load, no need to post about it unless it's a host I'm not aware of. I can add it to the list if so (so just PM or email me) but otherwise I'll just have to test such hosts directly during the final RC bug-fixing stage.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
-
- KVRAF
- 10588 posts since 13 Jun, 2004 from Alberto Balsam
Have I really been missing the -6db/octave bandpass filter this whole time? Is it hidden somewhere? Even the manual you linked only lists 12 and 24.aciddose wrote:That's actually been a feature since before Xhip was released as a VST, even way back when it was a synth inside a tracker around 1999.
-
- KVRAF
- 12555 posts since 7 Dec, 2004
It (version 7) calls the single stage KHN filter "12 dB", but in reality only the low-pass and high-pass outputs are 12 dB.
Each 6 dB (20 dB per decade) slope requires one capacitor / integrator.
The KHN filter has two integrators.
Therefore it can produce any filter with a combination of two "6 dB" slopes. A band is low + high, a notch is high - low, low is 2*low, high is 2*high.
The filter then isn't actually "12 dB" or "24 dB" but "KHN" or "KHN 2x". This is why I've changed the labels in the alpha version.
Technically these are nothing but labels and the filter is exactly the same.
You can measure the slope of filters easily by looking at a spectrum (taking the impulse response and applying the Fourier transform to it.)
Each 6 dB (20 dB per decade) slope requires one capacitor / integrator.
The KHN filter has two integrators.
Therefore it can produce any filter with a combination of two "6 dB" slopes. A band is low + high, a notch is high - low, low is 2*low, high is 2*high.
The filter then isn't actually "12 dB" or "24 dB" but "KHN" or "KHN 2x". This is why I've changed the labels in the alpha version.
Technically these are nothing but labels and the filter is exactly the same.
You can measure the slope of filters easily by looking at a spectrum (taking the impulse response and applying the Fourier transform to it.)
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
-
- KVRAF
- 10588 posts since 13 Jun, 2004 from Alberto Balsam
-
- KVRAF
- Topic Starter
- 3496 posts since 9 Oct, 2004 from Poland
Elo.
This time it is not analog vs digital, because it could be any sound and the effect would be the same.
It is more like your eyes vs your ears.
https://app.box.com/s/u1fjbbjxxb3pps6m88l6o8af800cjbgo
Sound in the OB-8 demo video by Katsunori Ujiie replaced with Xhip, TAL Chorus LX, VOS NastyDLA mkII, VOS epicVerb.
Sounds "more analog" when you are watching Ujiie san play the same notes, right ?
This time it is not analog vs digital, because it could be any sound and the effect would be the same.
It is more like your eyes vs your ears.
https://app.box.com/s/u1fjbbjxxb3pps6m88l6o8af800cjbgo
Sound in the OB-8 demo video by Katsunori Ujiie replaced with Xhip, TAL Chorus LX, VOS NastyDLA mkII, VOS epicVerb.
Sounds "more analog" when you are watching Ujiie san play the same notes, right ?
Last edited by Mutant on Sat Jun 10, 2017 6:21 pm, edited 1 time in total.
[====[\\\\\\\\]>------,
Ay caramba !
Ay caramba !
-
- KVRAF
- Topic Starter
- 3496 posts since 9 Oct, 2004 from Poland
Hi.
Today i saw someone on GS trying to match a nice Prophet 5 high resonance sweep with his Prophet 6.
https://www.gearslutz.com/board/electro ... st12058947
IDK if non-members can see attachments, so you will have to believe me that for whatever reason (maybe P6 just can't do it) he didn't get it very close.
The original Prophet5_27_2.wav file can be downloaded from the bottom of this page:
http://www.siliconbreakdown.com/prophet5.html
Apart from that i like the sound a lot, what made me take the challenge was him writing "vst-like sound" obviously meaning "inferior sound".
Well, so i took Xhip and checked just how inferior VSTi synths really are.
First mp3 is Xhip on the left and P5 on the right, second mp3 is mono, first p5 and then Xhip.
https://app.box.com/s/5559b1efduwdja2qlsxx5npq9fle61tl
https://app.box.com/s/umfa183wocarwzajacyiowxjni9z0b78
No EQ, just tape and tube saturation plugins i always have on the master and a limiter and then some volume envelope in REAPER.
If i substract lining up MIDI notes and few times rendering the audio, i think it took about 30 to 50 passes (tweak, listen, tweak etc.) to get it this close, could be even better with some more time and effort.
The biggest difference i hear now is how the resonance of the higher notes goes down through frequency peaks.
Today i saw someone on GS trying to match a nice Prophet 5 high resonance sweep with his Prophet 6.
https://www.gearslutz.com/board/electro ... st12058947
IDK if non-members can see attachments, so you will have to believe me that for whatever reason (maybe P6 just can't do it) he didn't get it very close.
The original Prophet5_27_2.wav file can be downloaded from the bottom of this page:
http://www.siliconbreakdown.com/prophet5.html
Apart from that i like the sound a lot, what made me take the challenge was him writing "vst-like sound" obviously meaning "inferior sound".
Well, so i took Xhip and checked just how inferior VSTi synths really are.
First mp3 is Xhip on the left and P5 on the right, second mp3 is mono, first p5 and then Xhip.
https://app.box.com/s/5559b1efduwdja2qlsxx5npq9fle61tl
https://app.box.com/s/umfa183wocarwzajacyiowxjni9z0b78
No EQ, just tape and tube saturation plugins i always have on the master and a limiter and then some volume envelope in REAPER.
If i substract lining up MIDI notes and few times rendering the audio, i think it took about 30 to 50 passes (tweak, listen, tweak etc.) to get it this close, could be even better with some more time and effort.
The biggest difference i hear now is how the resonance of the higher notes goes down through frequency peaks.
[====[\\\\\\\\]>------,
Ay caramba !
Ay caramba !
-
- KVRAF
- 12555 posts since 7 Dec, 2004
The amount of variation in peak level as the filter phase lines up with harmonics of the input signal depends in part upon the input level, feedback (Q) and saturation of the filter.
You can create a much smoother effect by increasing the drive (increase input level in the mixer) and apply light saturation (<20%) with lower Q settings.
As Q increases toward self-oscillation it becomes more difficult although it is important to note that in the majority of use cases Q is actually set quite low despite our tendency to crank it up to max to get full-on self oscillation.
Due to the extremely high levels of noise in analog circuits most analog filters tend to be driven as much as possible to improve the SNR. This has the consequence of reducing the amount of amplitude variation due to saturation of the resonance/oscillation. Depending upon the type of saturation (Xhip = hard clip + parabolic) different harmonic spectra result from this. So the amplitude variation becomes a harmonic variation and due to various effects leads to the creation of side-bands around the harmonic peaks of the signal.
The combination of these factors is what makes one filter sound different than another although the vast majority of people are completely oblivious to these properties of course.
Unfortunately Xhip isn't designed to emulate analog circuits and instead uses parabolic (second-order) and hard-clipping with a limit on feedback to unity (1.0) or less.
This helps to avoid aliasing because a second-order non-linearity only produces a single harmonic at twice the frequency of the input as in ring-mod. This means 2x oversampling can capture the complete band.
Hard-clipping of the feedback path with feedback less than or equal to unity means that initially as oscillation starts the level may build up and result in some clipping, although any continuous oscillation will eventually settle at exactly the clipping point without producing any harmonics at all. Again, this reduces aliasing. The consequence however is that oscillation can't start all on its own from noise and must be triggered by an input signal.
The type of saturation taking place in a filter like used in the prophet-5 is much higher order and as a result harmonics are generated far beyond the audio range (>20kHz.) This doesn't have much audible effect on its own but is rather a mostly inconsequential side-effect. In software all those extra useless harmonics would become aliasing noise!
You can create a much smoother effect by increasing the drive (increase input level in the mixer) and apply light saturation (<20%) with lower Q settings.
As Q increases toward self-oscillation it becomes more difficult although it is important to note that in the majority of use cases Q is actually set quite low despite our tendency to crank it up to max to get full-on self oscillation.
Due to the extremely high levels of noise in analog circuits most analog filters tend to be driven as much as possible to improve the SNR. This has the consequence of reducing the amount of amplitude variation due to saturation of the resonance/oscillation. Depending upon the type of saturation (Xhip = hard clip + parabolic) different harmonic spectra result from this. So the amplitude variation becomes a harmonic variation and due to various effects leads to the creation of side-bands around the harmonic peaks of the signal.
The combination of these factors is what makes one filter sound different than another although the vast majority of people are completely oblivious to these properties of course.
Unfortunately Xhip isn't designed to emulate analog circuits and instead uses parabolic (second-order) and hard-clipping with a limit on feedback to unity (1.0) or less.
This helps to avoid aliasing because a second-order non-linearity only produces a single harmonic at twice the frequency of the input as in ring-mod. This means 2x oversampling can capture the complete band.
Hard-clipping of the feedback path with feedback less than or equal to unity means that initially as oscillation starts the level may build up and result in some clipping, although any continuous oscillation will eventually settle at exactly the clipping point without producing any harmonics at all. Again, this reduces aliasing. The consequence however is that oscillation can't start all on its own from noise and must be triggered by an input signal.
The type of saturation taking place in a filter like used in the prophet-5 is much higher order and as a result harmonics are generated far beyond the audio range (>20kHz.) This doesn't have much audible effect on its own but is rather a mostly inconsequential side-effect. In software all those extra useless harmonics would become aliasing noise!
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
-
- KVRAF
- Topic Starter
- 3496 posts since 9 Oct, 2004 from Poland
I'll also try disabling the external saturation plugins today, when i have some time to tweak the sound again.
Apart from turning up the internal saturation knob which was already at about 10%, i'll experiment with automating the resonance knob during the decay phase.
Apart from turning up the internal saturation knob which was already at about 10%, i'll experiment with automating the resonance knob during the decay phase.
[====[\\\\\\\\]>------,
Ay caramba !
Ay caramba !
-
- KVRAF
- 12555 posts since 7 Dec, 2004
The only thing you'd want to turn up is the oscillator levels in the mixer. 10% saturation should be more than enough to emulate the p5's filter. The main volume then needs to be reduced by the same amount.
The +12 dB boost is a bit limiting although you can sync two of the same oscillator to get another +6 dB to push really hard. (I should test whether this limit still makes sense.)
alpha presets: (need to upload the latest alpha where I changed the preset extension to .xhippreset)
http://xhip.net/temp/mp3/linear.mp3
http://xhip.net/temp/mp3/saturated.mp3
http://xhip.net/temp/Linear_Filter.adxi
http://xhip.net/temp/Saturated_Filter.adxi
You can hear in the saturated clip that the harmonics of the filter are changing rather than the peak amplitude. The amplitude remains very flat. The type of harmonic spectra depends upon the type of saturation, which is about the furthest from ideal possible in Xhip's filter for this type of sound. Nonetheless it can still approximate the effect well enough and do so at 1/10th the cost of better filters.
The +12 dB boost is a bit limiting although you can sync two of the same oscillator to get another +6 dB to push really hard. (I should test whether this limit still makes sense.)
alpha presets: (need to upload the latest alpha where I changed the preset extension to .xhippreset)
http://xhip.net/temp/mp3/linear.mp3
http://xhip.net/temp/mp3/saturated.mp3
http://xhip.net/temp/Linear_Filter.adxi
http://xhip.net/temp/Saturated_Filter.adxi
You can hear in the saturated clip that the harmonics of the filter are changing rather than the peak amplitude. The amplitude remains very flat. The type of harmonic spectra depends upon the type of saturation, which is about the furthest from ideal possible in Xhip's filter for this type of sound. Nonetheless it can still approximate the effect well enough and do so at 1/10th the cost of better filters.
Last edited by aciddose on Mon May 11, 2020 7:22 am, edited 1 time in total.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
-
- KVRAF
- Topic Starter
- 3496 posts since 9 Oct, 2004 from Poland
Admirallo asked me in a PM for REAPER effect chains for my always-WIP-and-too-lazy-to-finish Shine on You Crazy Diamond cover.
I too think the Xhip presets i posted earlier are not necessarily useful without all the effect and MIDI files for different parts of the main pad sound, so instead of just the chains, i'll post the project file.
Cleaned up the project a bit (it is still a mess, not even the correct tempo because i just used ears and visualization plugins to detect note starts and ends) removed all MIDI except the pad (because i still want it to be "my cover" once i finish it some day).
https://app.box.com/s/ocp3ssydcxyec1rjnwxdbn50dll6ekoc
I don't remember how REAPER handles missing plugins, so if it misbehaves, you will have to open the .RPP in some text editor and search for ".dll" to build a list of free plugins i used.
Once you add the solo and some of the sound effects it should sound about like this:
https://app.box.com/s/mbpxke6bdpq66cq1pn6cvmz4rz9d1haz
I too think the Xhip presets i posted earlier are not necessarily useful without all the effect and MIDI files for different parts of the main pad sound, so instead of just the chains, i'll post the project file.
Cleaned up the project a bit (it is still a mess, not even the correct tempo because i just used ears and visualization plugins to detect note starts and ends) removed all MIDI except the pad (because i still want it to be "my cover" once i finish it some day).
https://app.box.com/s/ocp3ssydcxyec1rjnwxdbn50dll6ekoc
I don't remember how REAPER handles missing plugins, so if it misbehaves, you will have to open the .RPP in some text editor and search for ".dll" to build a list of free plugins i used.
Once you add the solo and some of the sound effects it should sound about like this:
https://app.box.com/s/mbpxke6bdpq66cq1pn6cvmz4rz9d1haz
[====[\\\\\\\\]>------,
Ay caramba !
Ay caramba !
-
- KVRAF
- 12555 posts since 7 Dec, 2004
If you switch to using the RC I've posted on the alpha page now you can avoid the issue with the different plug-in IDs between the alpha and RC by using the save-state feature from the last alpha version, replacing the plug-in with the RC and loading the state you saved.
It's a bit of a pain to fix alpha projects that way but unfortunately VST doesn't provide any other way to ensure compatibility between v7 and the alpha at the same time.
You must use the xhipstate files and can't use FXB because the host checks the ID and refuses to load the FXB.
Just saying since if anyone downloads the RC now they should find it won't load alpha projects. So you need to keep the latest alpha version around until you manually convert all those projects.
It's a bit of a pain to fix alpha projects that way but unfortunately VST doesn't provide any other way to ensure compatibility between v7 and the alpha at the same time.
You must use the xhipstate files and can't use FXB because the host checks the ID and refuses to load the FXB.
Just saying since if anyone downloads the RC now they should find it won't load alpha projects. So you need to keep the latest alpha version around until you manually convert all those projects.
Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
The coder's credo: We believe our work is neither clever nor difficult; it is done because we thought it would be easy.
Work less; get more done.
-
- KVRAF
- Topic Starter
- 3496 posts since 9 Oct, 2004 from Poland
[====[\\\\\\\\]>------,
Ay caramba !
Ay caramba !
