Ok, ok, ok… thx a lot!!!
The sole and only developer of SL since v1… wow!
Ok, great program! Great idea and ingenious implementation!
I quit now
Some 30 year ago, I dreamed about this kind of working with soundfiles, but had neither the mathematical background in depth, nor the programming skills necessary to create something like SL. Than, some years ago, I heart about spectral editing and that arouse my interest in it, but did not track further.
Recently I bought SL for a special purpose, i. e. frequency shift of selected areas of a spectrum. Fine, it does it.
And with this, I got interested in this subject again. So, yes, there is FFT going on and other stuff, I remember some of those things from the 1990s on.
Here is a concrete wish: When we work visually in the spectrum, we should be able to make a simple increase or decrease in contrast - visually, like in any paint-program.
Once the contrast in the visual spectrum is done, do the backtransformation and hear the results.
It should be possible to do this. Or maybe it is already implemented and I just have not found yet.
(I wonder how it is done from a programmers point of view, that we work visually and than access the actually sound data through FFT. The visual interface is so coarse in comparison to the sound data, I guess.)
Also:
-Would it be possible to make use of parallel algorithms for CPU and GPU?
The modern PC today can have lots of RAM (64GB is nothing unusual today, 128GB oder 256GB are easily possible). So SL should use the available RAM more thorough.
Spectral Layers is one of those programs, that changed the way people think about the world.
(Mathematica or Photoshop are similar in this way.)
It relies heavily on computing power and memory. Today, it is normal to work with 192 kHz and 32-Bit recording, creating GB-sized files with ease and we work with it. And the FFT in acoustics is just one-dimensional…
Amazing!
Well, right now there is at least an eraser and an amplifier tool.
But a contrast tool could indeed be interesting too, in the meaning of making loud frequencies louder and at same time quiet frequencies even more quiet (or the other way around to reduce contrast), depending on an adjustable threshold (maybe even auto-adjustable based on the minimum and maximum level currently found within a tool brush area).
… If I think more about it, I guess you can already achieve something like that with Unmix > Levels and then set the gain for each layer accordingly. I imagine a tool with adjustable brush like the eraser/amplifier would still be a nice addition for some more quicker and/or precise edits.
I sort of kind of agree but not in the context of the terminology you used (like “contrast”). I believe gui viewing options should remain separate from processes and editing and tools (and for good reason). However I do believe that processes(like transformation and gain) could be improved so that they can be done in real-time (instead of offline processing). For example using the retangle selections tool to select an area and then processing the gain in real-time (bypassing the entire processing process).
@Sunnyman Most computations are parallelized in SL : the rendering of the spectrogram (when you change a setting, or zoom/pan) use all CPU cores at once, and all AI processes use all CPU or GPU cores (depending on the AI Processing Device setting in Preferences > System).
Standard processes (non-AI) aren’t fully multithreaded yet though.
@Sunnyman@Laturec That’s quite an interesting idea, as a tool. I would say in the context of audio processing this could be called something like Compress/Decompress, and I can see a couple uses for that, as a mean to apply artistict compression and decompression locally, or to reduce or enhance surrounding noise or surrounding reverb locally, something like that. Adding this to the wishlist for a future SL version (although SL11 is already pretty much packed with lots of features already!)
to add more to the wishlist for a next version of SL:
Mirror the frequencies on the middle horizontal axess of the chosen spectrum.
Play backwards (mirror on the middle of vertical axes of the spectrum).
Spectral change for a selected region adjustable like the envelope for loudness. (for example: start with 1000Hz downshift at the beginning and end with 2000Hz downshift at the end of the selection in a continues (linear, or even adjustable) way.
