Collaborative project!

[hisefilo]

Morning guys! I'm back. Unplugged for the weekend.
I've jus pushed a new version with leakage and re-built the main timbre. @ustk how can I do to copy Leslie to the new .xml I've created??? Or.... if you can just plug it once again??

[hisefilo]

@hisefilo @ustk Nevermind, I did it :)

[ustk]

@hisefilo Yep I wanted to ask, haven't you pull/fetch the git? If you did, no problem

[hisefilo]

@ustk :) Just pull again. Pushed a new one Need some opinion about timbre, leakage. Also Leslie speed can be done???

[hisefilo]

@ustk the guy is Hammond-B3-002.xml

[ustk]

@hisefilo Cool :) I'm on the speed thing, probably for tomorrow ;)

[hisefilo]

@ustk awesome!!!

[hisefilo]

@ustk https://ccrma.stanford.edu/~jos/doppler/doppler.pdf super-interesting!

[ustk]

@hisefilo that's the one I have (partly) read. But honestly the maths are too hard for me so I just guessed and applied the main principles

[hisefilo]

@ustk awesome!!!!!

[ustk]

@hisefilo what I don't like in my Leslie attempt is the strong tone change it applies on the signal. Well in real life I guess it does the same, but I don't know if it's a good thing with your simulation. We have to take care not to apply the same correction twice
The second thing it the CPU, I can't find a way to remove or just apply the frame node only where it's needed so there's no glitch

[hisefilo]

@ustk said in Collaborative project!:

what I don't like in my Leslie attempt is the strong tone change it applies on the signal.

I agree. Is doing something that changes it too much. It enhances lower harmonics in some way. Will do a comparison on Apple's Leslie to see dif

[ustk]

@hisefilo And cut too much of the highs...

[hisefilo]

@ustk Well. Apple's does the same. Or kind of. Sound changes a lot on rotor stopped.

Original timbre (last commit)

Apple's rotor on (but no speed)

https://www.youtube.com/watch?v=f0VItpjyz-c&feature=youtu.be

[hisefilo]

@ustk trying some things on scriptnode. Any idea how to get rid of artifacts?? I'm modulating stk.delay_a and cannot figure out how to avoid them?

HiseSnippet 1606.3oc6Z0zaaaCFlzNpI1MMcaHXXXnGDJ5gTfVC6j9I1A6FG6BiE2ZDk9wNUvHwXSXIRMI5j5VLfMfcX+E1trS61vtu8iXX.6Wv10cI+AF1HEssjrkcscQRCJruDw2WRoG990C4KRCOlI12m4AfY1uqKF.WUynKk2pbKDgBpsC.tlVcjOG6oqDscWWjuO1B.goenT.LyRffemTbajMhZhCEA.OkQLw6RbH7PoMJ84Da6pHK79DmHy9VkpYxnkY1rNB7jVKOvEY1F0D+HjbZoz.vKTwhvYdFbDG6CfKsMypqQK1wT07eJwmbfMVNn.vP7hThqxrsjHV9LnbKhsUi96ae..tbiPqPZkUXcs5DKx.4gViOHPgd3JhZOfolD7JDEd4SDdER.dfHnaIE5DffY0wFwiiLo6nmBBKtmpFkio9Dd2ntq2ov8C0ZP3lsRFuoR.uBG0oMd647ujVkCODaxCA6RZUe975omg.Q3n1trZFldDWt36KAxp8FpW84EhAHkX.7m0pPaRn3bldXAz1w28QX9wLu1abUKrMp6tBcW85e1gcnlRittqG1E4g2m0PnbCejiqMdOwBug9A1Ly1FjWgud1WmMS1uJqdjEE.yskyXCyVHJEa6O5zj4xTtGydCZGmCvd2P+Hjcmv22HoyKOcVQFsFkverKt2zhsG5KVT7HBJCkN.TghRHnXOVGNg1rNh6QdI.p8nNNFhpRl3x81sBYvT.4RCFmWNVFiZfoVAC9OwudJKHGC6orPek.INvTtPdOWTPHPumEOI82WTafWSpiYI9yJUQlBiV2FHdKYdjnlIWDnh8xYNHfc0v0U.H.+vvdWgMIlXXZ4KuesW0doLywQBPYk2dO+Lhk7ipAugXFp3u8DwLx5308iVIGlR99T6n.TurDVKq4zwlS.icGnTeZAX0ZjfJUOSoDTYT1pWfhCqKo4yamqupgfD7LwFlNr539HulXdP8JjmfNTPHqpXzejnTQM5QXOtpdPcB8oxjsv225kpid4PxzKsK6Xr2P7y.vSbcGQ50JYvwtxBBQmoQa7wJyVTlEkudGosCLzmD.FdSIq2KqBK9jbhz6.hDwjUygYohqi4cVOZPS+YbZ4jBiaVJRbypZLeSwAYj7ZiFQ6gyEQ86jTvjCehXnCHbjiDzroUoBOwGKOfSeuFLciNGXSLkOEOxKcjHuLSNxS3xGIxakYHxakDh7No33i7tf77F3QC7.SOjOo3nP9KRDx2pTBP969yXP9G9d4ueMFjypkO2lad+6m+1.Ug9pd3urClZFIio4gxeWuzaIvymHvOoXR157ynsd0.bqumLJCLBV.wJy5SZtoHWMdtxZZNh+lqutyQIJiqN6bDseRwoOZWXsi3Ad85+xC+qW8sEmTcV0GaTieB0TWQke3hQsi6FtnpjUfhym9fwtcxp46vX7V3gpBeY0VZfxyMaKHbTJvUke2ZVQOMxfMuTbe1zG6N3dAXtxkWghDGRt20ll0HyqM2m.3lkux+7224SGSjYkWJNUtuu7dcR61rR6jZbzNoN+lHtlvEcHRb.1DyGmI5mDfdoeb58Th4NUzOaUXq6du.5mKqYDjlHtySPOYhSAs12TZBohSolmTaGDGI898t3SjoBg6fOhXhUw1YDFR+1blarKmulR4E0LDG5KnGUAH+iCFq+LzQX8GhEGGTddqBSnMU+wz1lJ2otMUO1jK9766gn9tL+3mOF6P1mQw9wbY9XIqohzLh7xLjWhpNoXUwlLw0Xf3c7BJk7.GVGZrPCX54pKEw6iTj4mPKKVZQuyBfa5oCtugl9joOFMHxFwTQTjvVjjDfwOpe4k9RiGjWmQYtsXTYQyvHf8vbORyl33EqRZC8.NGY1N50E2CaiQ9w4JDQgHufSWOe1hBSusXR9qOQSAWcYlo96G87L86Q87bFfxYQm1lSVjMGGKh9Q+9u8tlE4JkVvhrfE4smEYyErHKXQVvhb5whr03XQ95+8Y+zYHKh35rKtKxBVjSGVjsVvhrfEYAKxTvhbV7MbPldrWXp9+bPFKuRfDw9lF7eRUFs5xw5E.Go5ymlV9b4ANhJhuvzTZtuov9j7Z1bNVyVywZt0brlaOGq4NywZt6brl6Mw0HIidPGNyQkJJDznRP6Rgvv16mF7+XInVhA

[ustk]

@hisefilo This is where you need to use a frame and to oversample, and watch, inexorably, relentlessly, your CPU touching the sky at first, and then fading slowly, disappear, in the celestial vault... Well... With a bit more nodes at least

HiseSnippet 1726.3oc6Z8zaZbDEeVvSrgPbZqrphpxgUQ4fiTJBvN+S8fIFiirpIAYb9SOYMdYLLhc2Y6tC1lDUodnG5Wg1K8TuU06seHppT+DzdsW7Wfp14sCvtKrPvjX6nTPJJLuYer+d+Yd+dyLtpK2f54wcQZo1siCEokAWqisnYolDlMZqMPZKhqP7DTWckn063P77n0QZZIeDHPK0bH+Omr15DShsAMPDB8LNyftMyhIBjVs3myLM2jTmtKyJzSuZwsL31k3l71R7jDmC4PLZQZPeLAdrDXj1kJWmI3t0DDA0CoM2575cp0jejs54eFyisuIEFjGUS9CoDuI2rNfXPJpTSlY8p8raODRa9pAdgjJuvR3Jr5r9xC7Fef+D5AZD1enkXbvKeX3kahgGJD5lSgNIH30aaRDQQFDN5NAiGMRsksfZ6wDcBGttPg6GhqxDFMiGuIhAux.0YMd6F7uBt7AGPMDAfcN7lu3rORqoMruKMtlgKyQHe+.Pxzcn9luHeD.oDiz9IbY6FLaZVCWpDZa347Xp3HtaqkuQcpIoy1x4tws9rCZaa.NccGWpCwktKupbxk8HVNlzcjJda88M4FspwdI8VoeU5To+pz5gTxGlqCOwxFMI11TSugeLXsrsvkatrcaq8ot2V+PhY6feugVNmbzdwvKmiIHtCusfY2nBQ3xNFogebaqZxpHFzRcQmTlVBDnp+3bvXHmpF0tt+f+U9o6j4gwZcmLeuIQfwPsER4cco9grteW9MH9bYbeuLLGuNTdaShgzH6TkHZB48xZbBYhE0MqQ+DrLA5kGIA+fvdaomNhXsjvOduZkJaoD2xB.HTor62eNqN7RwZ2V9Dp7kcjwXntaEuvUdilbC+1Jqy2BlWsf8HWhC7Oq1lBV97nH100BYW7CotpToUO9LyZhGkYvAHbz.7.WIiRg87SwO6.Xx3.37XOGSIa3HSJTSe1CpDJPs..pqp7Z759Yj4QulL1KhDzfvchPdyz3dfNJlWJDl6+DC.as299z4B4Syf4dFx9b.Zugcntzrgl9BwglLf6cWhaCpnKanC0Uvn83ifQRV3jfIkB+TOJz+C5YPgb.SUauuIy.9FpJAVUI6TDTMY+QR01xVVPPnJtTgY2U4.h8JjiGP1BE2leD0cfFGQnm53LjzEJVSPc.lpvM2TqE8HkWOrTvJtDzNBEMDJPSNjOYsgg7WDKjWsXLP9a+iHP96+N3yuDAxow4xVnvCdPt6fT7Ja5R+x1TaiNCB7LEeCAdtXA9IqEmuN2ozWmwG256.YYngvBpKAoJ2xi0nfbsZz0JKhsj+e1dycgSMF+hlvY9ZuYY9mr1jm4K87ghFuZoe9Q+4K+lwDMlGqdYCGHBurGEpH1kvNTRqngjKqJe4Ow6PEtBECFo4jF6Yw4hlzApHeUkI0ex2YLKsDCJTtlBduaUGVx32z3dj.iGDOOdCPL5IN82BAUnB4ksIx9o6tCqwkYtTwgyL0isNwRwUfS9rgyL+zRW+u+q69IiHyr7wxMg34AaAD7anwXwo6Yw4e+xjOkrtIFEqah2cq8rnLDc.Q1WdrkfNUruw.8h+vj2vf7YmH12Uxux8tOR0dbM+JCxcX5ehU8e2Jc0J9Zq93Sws.1vjRbGn85LJFN+otPJ7z6PIzR7dI61fAjHgpPakXQ+8.0u9xH2Bj+dKOG1Ua38mEToOBrtB1SzJauoNq1dyaFWbjLlzusJAGeFyMio6TDZbYLJZigSbGhBRFgGcSEwmzLK5bgEchVNdj0lmzYBO3oasAQP.OW2yCLjdZZaPOjYPUMgjRheuVBtSjyXdQ0jWFWSVMw+pV7oX9X+w5OmbHU+QTYcF3bAxOlaa42mzaawYhuskmXHju9ccI1dNbunQIpEaWtM0KRP1iB6tSs4tPxKwItwN0IqsozHiUmZDQaW+X5Cs3ssijLokbhOr8Sws9L2ra840cn64F6gtG45JR0Ci0Xv49VVVIwTttvGieTuV+5IMZdcEtM2oI2FZnMHnuCU3xZzfFsrPbFzCEBhQqv0o1gJalJbB7MKBmrNwshpYrowWjex8EiKdcMrBt5vhQ82Otstj+u715NWtyooj3nvnHNzO7290KZhiqWbFwwLhioh3nvLhiYDGyHNN6HNVYTDGe8+77e7bj33j0lsiiYDGu8HNVYFwwLhiYDGS.ww4w6vhX3x2yP8GGHjKufuDoca6+W5aJbEXrddzgpy9CiykMGxRVQbOCCvc+oR+S75TXJzYkoPmUmBctyTnycmBct2Tny8GqN.YzCaK3VpkhRAUK6eNnZZAWvZRz+AzksAkA

[hisefilo]

@ustk wow!!!! CPU killer. Newer Hise solved that?? Im on scriptnode oct/2019

[ustk]

@hisefilo I know @Christoph-Hart is rewriting everything around scriptnode and SNEX, so we might get lucky soon enough.
Imagine the CPU with a bigger graph (with delays)...
In fact, the only way to reduce CPU would be to use frame and oversample only where we need them, but it's not currently possible.

At the moment, all my attempts (not to have a global frame) led to a crash, or just didn't work at all (I'm speaking about stk delay only, if you don't use them everything works fine without frame or oversampling).
You have to embed the whole graph in these containers
I don't know if it's normal behaviour or a bug (the fact that it's not working when inserted only where we need)

My actual project is very big and still light on CPU because I'm not using delay, so consequently no frame or oversampling. And with maybe more than 300 nodes, the CPU maxes out at 3%... And half of the 3% is due to 4 limiters only...

But the next projects will be delay-based graphs, so it's not looking good for now... Finger crossed!

[Christoph Hart]

The framex_block nodes are not supposed to be used in an actual project (at the end), they are just useful during development, that's why the C++ code generator is part of scriptnode. This snippet should run about 10x times faster if it's compiled as custom node.

However Greg is right, I am currently rewriting SNEX to be able to compile the C++ generated code from a scriptnode patch, so the workflow for using frame blocks in almost-native performance might be vastly improved.

If you don't want to wait (it's a really big TODO list here), you might get the CPU down to sensitive levels by using a fix16_block instead of a framex_block.

[ustk]

@Christoph-Hart Cool, that's a thing I have probably read somewhere, but now it makes sense :)
If that can help, it is apparently not possible at the moment to make a custom node with a JIT node inside. XC throws an error when recompiling Hise...