In this video, we will be exploring saturators and their efficient oversampling. Saturation is perhaps the most popular audio effect, and in a digital environment, it is most commonly produced using some form of sigmoid function.
When using distorters, new harmonics are generated. If our signal contains higher frequency components, the end result may include digital alias noise. Contrary to popular belief, this isn't always something to worry about-it's useful to monitor the spectrum analyzer. If our base signal contains audible frequencies close to 10,000 Hz, then we have reason to be concerned about aliasing, and it's good to use a distorter that employs efficient oversampling.
Infinite linear oversampling, also known as antiderivative oversampling, requires slightly more resources than 2x linear oversampling, but the results are significantly better than any linear or even multi-point oversampling. It can be made even more effective when used in combination with other oversampling methods.
In this video, we will review Efflam Le Bivic's documentation on ILO oversampling, then build the hyperbolic tangent ILO version, and finally, we will write it in Lua and test the LuaJIT compiler that is currently under testing in Alpha Forever.
Timestamps:
00:00 Intro
01:00 Naive TanH saturator
01:35 2x linear oversampling
03:26 Testing the linear oversampler
04:00 ILO oversampling
04:12 Efflam Le Bivic's paper (found it here: www.native-instruments.com/en/reaktor-community/reaktor-user-library/entry/show/10381/ )
07:36 Building the ILO saturator
09:40 How to calculate the integral of an NL function using Wolfram Alpha (ChatGPT is also fine)
10:56 Testing the ILO saturator
11:40 Lua code (using VSCode and Codium AI)
12:04 Creating input and output plugs
12:17 Required functions for the LUA node (idx is the index of the polyphonic channel)
12:38 Lua has no classes, but tables can emulate them
12:50 Creating ILOTanH objects (one for each polyphonic channel)
12:15 Function for initializing variables for the ILOTanH object
14:05 Init function is called whenever a polyphonic channel is created (or when effect mode is entered)
14:37 DSP function (will be called on every sample): this is the DSP code, that could be interesting for Audulus users as well
16:15 No need for e: math.log uses e as a base when only one argument is used
18:40 Calling the DSP function inside the process function (on every sample)
19:27 Testing the ILOTanH code in the Lua node
20:00 Final thoughts, patch can be downloaded from our Discord
23:03 Bye
Our Discord server: discord.gg/QXQyQhRfQV
We are developing Alpha Forever Modular, a modular VST plugin, and sound design tool.
My latest release made using Alpha Forever sounds only: czcams.com/video/Hks0Gat8XnY/video.html
Follow me at:
Facebook: 9b0music
Spotify: open.spotify.com/artist/3LWp7Y5CCVs1HN9BG06YAw?si=X48ukA7FSVOTf9dWssMNgg
Soundcloud: soundcloud.com/9b0
Bandcamp: 9b0music.bandcamp.com/follow_me
Discord: discord.gg/QXQyQhRfQV
Know more about Alpha Forever Modular:
afmodular.com