Snex_node Needs to be wrapped into a compileable DSP Network

Matt_SF

@Christoph-Hart Noted thanks ! Yes this was not meant to be anything, I was just testing something.

Christoph Hart

@Matt_SF said in Snex_node Needs to be wrapped into a compileable DSP Network:

Yes this was not meant to be anything, I was just testing something.

Yeah, it would require some Steve Jobs level marketing to monetize this :)

ustk

@Christoph-Hart said in Snex_node Needs to be wrapped into a compileable DSP Network:

I noticed you're using the stk.DelayA node, this is deprecated (I removed the entire stk module). Use the jdsp.jdelay node for a interpolating delay line instead.

But jdelay isn't meant to be modulated like DelayA was, right?

Christoph Hart

sure, just don't modulate the limit. The core.fix_delay is not the most suitable candidate for modulation as it fades between two delay buffers to avoid pitch shifting, but that's what you want for all chorus, phaser, etc effects.

ustk

@Christoph-Hart ok so I have to test again, because what I just tried gave me nothing but artifacts and glitches…

Christoph Hart

Well it‘s the delay class from JUCE which is used in their chorus and phaser implementations so it‘s supposed to work...

ustk

@Christoph-Hart Unless I'm doing it wrong:

HiseSnippet 1447.3oc6X8zaaTDEeVaOlXmTTAZEbbUEGRkJQ1lxej3PbSbbIh5Vq3Pa4TY5tSrG75YV1cbZLHNwG.NCm3C.GPb.ItwAthDR7EfSHN1uAva1Yc1Yh2XmXQaEU3KIy6Mu27ad+adusajviFGKhPNU1eRHE4rFt2Dtbv1CHLNZ2VHmWD2gDKoQtZRaMIjDGS8QNNEuohfSkRnjeOdysHADtGMiDBcWAyidK1HlLiZ2le.KHnMwmtOajwtudyc8D7sEAhw.dJhqgBIdCI8o2ln1VALxo7N9LoHpmjHowHmRaI7mzaf3Qb89uKKl8v.pZQcTOPQZxsEA9JDq9ez1CXA9cmduiQHGb2LqPQsU3R3NLe1wzyrFWLggalDl1CmByCd0MgWsbgW8bfGx.ckznC.gveb.QZiLk6HkASX6o1kKo7Xlbho65YJbeIbWlzaP93sPN3EbTOowapy+B3cN3.pmLCrkvsu+x5oOGAhNyZ6ph64EwBkv4q.xZoKcae+5V.RSF47c3c38Yb5FdQT.ZshCuMU9HQzv0uhOMfLAr2W4puW0CFy8TFc2vHZHIhtunKvb8Xxnv.5dffWy8gABug8XeF8pU+7pUp9EUcMDJAlao1w5dCHbNMHd1soxk4xHQv57widHM5ZtGRBFmouYRmKd1rh43D2SLVx386PjQrifL5aOdTOnJhGc6TzAzbJfThlrtlZsJlpGk6mr3ugeoLqqV6jxr9TlH0kgxk.8TSZhKK8+g+S4ephmZkUrD9Jr2l3A2wIcIxApvdnDmDhqnQa3cb7kgX1QYJUnOFstTa9EvwgAP8zSUuZ14qnhVJRifE.v7UTAshJog+HgeN55RF553c.tFSmhxNOWnVRG2Kh8fGMT0PlEuQzMLXuvCXagJJ+nVDIQYZaw.CFYxVigjdcZnEEkmk6SO533ftjH38HodukOdEvuCieWUDtYQqNjiNAsK1rmjFpxrLKFqtkkUEJonYzAx3PJadHeyWq98SaN6g75m7P9V095Mj9Hs0KkZyubzulbzqhaGQ+zwTt2Djo5KpvFZ9WxGmy4W6TujqkbRt6oJ5ilQSnyu87wad51yaBEVl+gT5rdH5LltCHwKvCctzX4jWsML4+h62G8W+vlHcvOVsmJ3XV+FPBjcj+KhGA+cio7VPX+zmdbJ.OyXFA6r7FCM4YMFFItkwgTxPafupNkMgwhRVKl0fv9jn9TYBjOIQ3bTm4tfLqf+jjZoYdDE0UwsTDS52bGNAdTIwTb1tteTyYxa93Wo0uuI5ewBI1KLLfPQ5.J4DE8VS650bdV4kyAwIOqrh9YkFK26JNVZJ0WZooK.TiC2HkkkVruyEN20RZ9UmhqVaIzywLiHmZ44yeHktTbVr5MFKEfqNMZclndnklE3EpuTdg4DZZs3C2cZreZyQPSYgzHIS0MmSK5gvve59Xq.Wp3gRQnUC2WVybU78HGRSl6LocnKmr9.QzH2aRAnpdVu9oM44O+m+1OdVm7L7LO44c7j.D1OhviCEwz5lZ9D7ZXxqEUNlaucMIqcotdsgqmw9Vo4ThMLI1kvsTFr1RScTs6lsTWaqGEtM92IqkHib5tiCho2i4KGXn2+vfbCSxuOIxG7KdVigUbol8wd5Ti8myfPk++IxWz.Q0l6.QViRVYJF6wTuWsC+PZ.jllfwWFxKOfLNPNkpcdVGAWDNPvYVQ.6Qgwr52mZEXk6E5FRIwaXFkK0bOJ7nUrUmp2BJGQhRZ8c4rE0O61h44udMrFttphCtOe7kTJ9bzWR4IQFJT+7+Nd5WEC0za3l3tce16ueZ7QWdZbFiHdQhG3o+VUpBFqjPAt27jO7bEbG0Z25nC0cwgw01nFZD7ryC77TwzuAXexWlFKgLu4RHy0WBYdqkPl2dIj4cVBYd24Ji5E+zViUw+.gt6jzIpiS1zcEQ+CjC2c6.

Matt_SF

@ustk @Christoph-Hart and I noticed the same glitches appears if you modulate the simple gain Delay parameter :

HiseSnippet 1079.3oc6X0rjaSDDdjsmTqKfTg7SAG0AtQQJKuKAxIbVu1KtvN1rZqDn3PpAo1VSEoYDRiVv23MfWMdD3NW3M.lQxRZjWgWGE1JYSgO3xc2yOeS+e6EQbGHNlGgL5d95P.Y79X60Lg2POBkglbBx313YjXADYlw530gj3XvEYXz9TECitcPoe9qu5XhOg4.krPnmwoNvTZ.UTxcwfug56Ol3BmSCzV8QCl3vYC497DIdZi6gBINujrBdJQsrVXjwsF4RE7HaAQ.wHiNGycWa6w+YV15eFMl9i9fhvBYKOnL1i49tJDq9MZnG02cQ96NFgLvKJ0BsyzB2GOi5RK3WpMtSp.yxcnqOLZsK3YoCud0BOqZfGRCccxPmDDb2DehnJxTliMBn7pVpILAvhoh05lq2nv8CwKnBGu5waqZvqzPcci2MF+O.OZ4RvQTB1N3weWSszuBNhstLV5hOimH.40mozTTT1JyYDQD8WrpBp1ulfpV8ylaL6Bk.5oIA1x.TGPpdXLvWdQXCkMZCcOEsxbYCL2ThVkBsTzFaDlRztTXeE8eK+nDVPrQ3g5BKHzsdGTj.iFD5CmVl.Kigohi0NRfgPm.9jbWte8O+16M34TWgW4J99A0jhaB6BHRrf6Sh11gcKCR68yfTimQm8NpR8jLuwjE3t3TU90EfeEh8LzP7syPLFOc77M4Vk+xr.+VWMHGGA+TBvbzXdv.UIuIkNbC9sAOmbArjGErIXuX6SgUxKRmy4PPHWVZ1Q2AyNfyEdxnypERu+fobd3HFQpFb0OjEdjXX9xkwfP+XFlDK3A4XIsRqs.BOgHHHie.+nidn7yxOU88iGJ+5v4+2PeoL5MO.o0d5vcuTKYgISYSeKODICwE9S2.P7cwooMu4jE5N3Ej2FRZ1nRs40D6sUASKcgVaIredc4+0Rss2Hr1RsOHSY9dXUNiz4ER0iO.mmCw7TfAQWNWY0Bt+99NwP3dOwvbGgDBmGQXwg7XvR+j2RV+pU9EIrpKOiUkUoddikOOKcegbl80S0KcnpbXR5JmzrTCbAYVtZaP9ZbmG6HerJcWkT2I9wPZbk149GZr6qy9qIQtaWrngsGVcpBs0WSV3a8+SRcU0w5s65X5sgzMGiY8uNR1moOWxIuwokjDeQN2pwYy3LdnGmQq3AbFHSrrZETwwp1GzSDB4n25cTbF3CxlGJY8IClRY.IRpmfFpKr1ecwtrWeLNCtlpjCluaLAb62gl.95HBUl+7lik9ivxb58MSM2lu4s2a0lAt11LZsqI52tMiJSzm1rvq++ZvUdG5sxf1UqLFWpUl.hSD+ExhshHtuJUyAobjZLV5e0XW7LEsoE5BheR5nf8dXOTfrf0KbbTQCelTyV+d52f8bXC1yQMXOedC1yiZvd9hFrmubm6Q0qvSRjSolE4HYrXT17oFEy4J6Y8e3jtzy.

Christoph Hart

Unless I'm doing it wrong:

Scriptnode 101: if modulating sounds bad, put it in a frame block (or fix8, fix16, whatever works).

ustk

@Christoph-Hart Of course... Silly me, I used an oversample chain....

Matt_SF

@Christoph-Hart said in Snex_node Needs to be wrapped into a compileable DSP Network:

Unless I'm doing it wrong:

Scriptnode 101: if modulating sounds bad, put it in a frame block (or fix8, fix16, whatever works).

Mmmhh, we agree that we are talking about putting the jdsp.delay node into a container.frame2_block or a container.fix8_block right ? Because it not working here, so I'm guessing I, now am doing it wrong

ustk

@Matt_SF Put the whole thing in it, not just the delay node ;)

Christoph Hart

This post is deleted!

ustk

@Christoph-Hart What is the main block size decision? (8-16-32..)
Is it the computing charge of the graph, like the more you have to compute, the larger the block should be?

Christoph Hart

Yes, at the end it's all about CPU usage, but be aware that these kind of applications are supposed to be compiled into a C++ node, so the performance difference between the different block sizes might not be so huge in the end.

However, if you don't hear artifacts with higher block sizes for the intended use cases, there's no reason to go lower, so the general rule of thumb would be "try the highest block size that doesn't sound bad".

Matt_SF

@ustk @Christoph-Hart yahooo it works !!! : :party_popper:
Thank you, thank you, thank you both of you !! I've been failing for weeks, on and off, at finding a solution. I know I still have a lot to learn but this is a problem solved !!
I can finally have a good night sleep... and find another wall to break down :beaming_face_with_smiling_eyes:

Brian

@Christoph-Hart I think we are going to need some video tutorials from you on this insanely amazing DSP powerhouse you have created ;)
And is the PC SNEX workbench working till you sort out the Mac version?

Matt_SF

This post is deleted!

Matt_SF

Hi again, scriptnodes enthusiasts !
I'm facing another problem. Here's where I am :
I successfully compiled math node and integrated into a project (although the compiled plugin doesn't output any sound but I have to dig deeper into this before complaining :) ).

I have now a whole nodeNetwork with only nodes, which is quite CPU hungry so I want to compile the network.

In the workbench, I've successfully wrapped a test-Network into a DSP network, but when compiling I'm getting this :

e:\applications\programmation\hise-develop\hi_dsp_library\dsp_nodes\routingnodes.h(391): error C2039: 'frameData' : n'e
st pas membre de 'scriptnode::cable::block<2,1>' (compilation du fichier source ..\..\Source\Main.cpp) [E:\OneDrive - S
oundFingers\-- HISE SCRIPTNODES PROJECTS --\FX_NodeExportTest\DspNetworks\Binaries\Builds\VisualStudio2017\FX_NodeExpor
tTest_DynamicLibrary.vcxproj]

Using last develop branch version + windows 10.
And this is the WB project snippet (just a simple delay) :

ScriptNode554.3oc6U0zaSCDDccR2jTZEBDbGIti0t1tj5SPIUE0CMJhf5UXwdRhUs8FsdMsgeabn+WPh+Fkci+H1VMQ.MUkC3KVimwiey68FuC49.xnyoGiL5hSlGFHQmv7jbwhQL4LjwSwd7XIKHFDlYoGlFMXFKNFBSPFswVngpVLfGxSE5XBZ.OJBhkHiVXz6VLmkj.9YYLZORvmCBY.jfPKeQ0ciVCWAhdXuYpul85Qwx7aOTXXjGsPUpFA6iOM4bcVOVH5bVXJneM5J3tSE3tKV.dPvWgl3UvSkAwSMKReWvaqZr1HlfEARPjA9hHk9cVPbNdwXhIAcF6pxXpJdrDlON3ak4GeAbYFrKpPOROBeB.9eg4cQwzu6NpO6m+t7Gu4nTIOhIyPVNkjyD6gmDb0m7gP1h5bwSTZm.LWk8dhJZ86QE8vTBgTiMdtpD8C0WTGWBk9ZZMxQWfMgRrUE35dniaeBEkMzGqGoOFDAqHqat4EgS94j21jrpfuNaFeVNMTK5lTKlOTE.cwGPsTYqJNcvIPrecc4wkdzk41ZpxsrPMfq5qmLfGWxRk6M0DwpXtKdpZQmVGz6kYlldW+EvF7QsuceTO7qnjlaU+g9nmgOvzwxxpOoeemCcn1DWTl77d8.U5f5gtFaa+xq+6cPaQG93HNWNS4SJvWGrkt6qAM2m7z93O.IfLqs0Z9ZcQ4mmX8O74IF0VTWZseH87cdH77MUw+av2nAuVvu.b1184N

@Christoph-Hart Am I missing something ? or maybe the whole functionnality in not entirely ready ?
Cheers !

Christoph Hart

Yes probably something wrong with the send / receive codegen (from the error it mixes up frame and block processing). I‘ll take a look later.

Snex_node Needs to be wrapped into a compileable DSP Network

20

1.8k

10.4k

90.8k