I know at least MeldaProduction’s MSpectralDynamics has the capability to analyze a signal’s spectrum and apply a dynamic processing very much like a compressor or expander (or even a custom shape). I have no extensive experience with it, but from some quick testings it seems quite powerful if you know how to handle it.
Not sure if you could achieve the same “contrast” with it as with an image in an image editor though, during my testing there was quite some “volume ramping”/“pre-ringing” to loud sharp transients (like a piano key) - unsure if this can be tweaked with certain settings, or if it is just a usual limitation with the FFT stuff (resolution of time VS resolution of frequency).
1 Like
I believe Sonible and Techivation both have dynamic processors that work in the spectral domain rather than with waveforms. There are probably others but those were the ones mentioned when this matter was discussed in August, IIRC.
1 Like
Sounds interesting
1 Like
I understand what @Sunnyman says that image is…yet that image in the contrasted PiP doesn’t look like what he has in the spectrograph…FFT size of each image looked different to me …but I’ve looked over the images again…it seems to be the same image
I was implying you should code one 
SL does work visually on the spectrograph
Pls quote at least the complete sentence:
“Since SL is supposed to work visually on the spectrum, the basic tools of picture editing should be accessible on a sound bitplane accordingly, imho.”
1 Like
your wish is my command 
I edited the previous quote which you have quoted
ok, Marc
what will all these photo tools do to the the audio?
Sharpen, polarize, grade color, rasterize, yadda yadda yadda do to the audio? Add noise?
anyway, I wonder what it is you are hoping to get out of it? sound manipulation like you’ve never heard before? Accidents that end up being game changers?
Again! I agree with you, however tbh, you’re thinking too small (and I mean that in a good way).
To put things into perspective (and you’re probably not going to understand this) but I find it ironic that Pro-tools (the leading industry standard in both music and film industry) recently desperately decided to team up with Spectralayers. For Avid to do something like that it means they know that Spectralayers has so much potential and are going the safe route with teaming up with Spectralayers.
Me personally! If I were the developer of Spectralayers, I wouldn’t have teamed up with Avid because I know my worth. I say all of that to make the point that you @Sunnyman are right but (as we all know) there are many agendas and there’s a lot if ulterior motives behind-the-scenes that we don’t see nor understand. I believe that most of the features that members are requesting hinders the developer and doesn’t bring any value (in-terms of actually being useful in the real world) whereas a feature like contrast enhancements can be useful.
Here’s my two-scents: A feature like contrast enhancement can easily be implemented. As a matter of fact I can already foresee how this can be implemented by taking images and converting it to sound (because at the end-of-the-day it’s all images of sound/audio waves), however the way you all are requesting features can become a hindrance to the main developer. Some of you are requesting more separation algorithms like multiple voices and (me personally) I dont necessarily want the developer to focus on that multiple voice feature (nor improve it) because it hinders the developer on inplementing other features like contrast enhancements. The point that I’m trying to make is you all have to make up your minds on the features you truly want implemented and you have to decide which features you feel is important (because it takes an incredible amount of time and resources to implement such features). You have to decide do you want better multiple voice algorithms or better tools (like contrast enhancements)? I vote for more useful features like contrast enhancements.
You have to keep in mind that the development cycles of Spectralayers is extremely slow and demanding features like this can be delayed until Spectralayers 15. That is why it is important to decide what is truly needed now versus what is a want.
1 Like
Sure, I agree the contrast tool can certainly help with selections and other potential uses
it is interesting to see zero votes on this request, tho
Hi, yes, my wishes are small, but will have a great impact on the overall usability of Spectral Layers. Since SL is such a magnificent product already, more powerful features just make it even better. My suggestions about an implementation of contrast enhancement and the power-slicing of layers into delta-db are somewhat of standalone features of auditory spectral analysis, imho. These suggestions came natural to me after some working with SL and seeing the need to have such features. Since SL is a visual editor, that mimiks the workflow of photoshop (or paint) to some extent, those seem very natural to me.
And yes, it seems true, that those wishes are small and somewhat timeconsuming for the developer, i. e. Robin. I guess, that Robin easily understands those wishes and might also have a programmatic solution at hand already, but can not implement them easily, because of all the other tasks, that are in the pipeline of development.
I think Spectral Layers is an absolute masterpiece of software engineering. It’s concept is so clear and well formulated - it amazes me every time, I work with it.
Some decades ago, I talked about this kind of working with sound data with some friends in a pub, while trying to understand the deep implications of
fouriertransformations and its numerical intricacies. We didn’t have the understanding nor the necessary programming skills at that time to build something up. So, it was just a dreamy idea. Now, we have it, there is Spectral Layers and I can not thank Robin enough for bringing this to life.
So, now I am back - the ideas of visual spectral manipulation are endless. And we have the means to do this today!
2 Likes
https://www.mathworks.com/help/images/contrast-enhancement-techniques.html
https://ietresearch.onlinelibrary.wiley.com/doi/full/10.1049/ipr2.12450
Couldn’t this be done on the FFT-Data?
https://docs.opencv.org/4.x/d3/dc1/tutorial_basic_linear_transform.html
It would be so helpful to have this.
The contrast enhancement would make sound extraction so much easier - even without the heavy use of AI - and far more predictable.
In combination with AI afterwards, it might be even better.
1 Like
This plugins does the trick in principle.
But, well, the sound quality could be better, imho.
It’s a nice plugin, but I believe, the algorithms could be improved a lot.
So, maybe the inversion of the dB-leveling would be really cool feature in SL.
To clarify, see here:
Before
After transformation (silence remains silent below -90dB):
Btw, anyone know of a spectral delay with very high quality? Even a true (analog) hardware-version would be cool. (That this might not be the cheapest kind of equipment is clear.)
Adendum: This is not spectral contrast enhancement, but rather spectral inversion. But anyway, it goes into the same spectral direction. Spectral contrast enhancement is neither inversion, nor spectral compression in generell.
1 Like
This can already be achieved with a technique called convolution reverb. The problem here is that VST 3 implementation within Spectralayers has some serious improvements to be made in order for that to happen.
Thx, it seems, that I am not the only one, who considers the VST3-implementation with much potential for improvement.
The lagging and instability renders the VST-interface, in it’s current state, as unusable for my needs.
Workaround: Using Wavelab with moving spectrogram in SuperVision, which is kind of suboptimal.
I’ll check for convolution reverb.
Hm, " Bricasti Design M7", looks interesting.
But I guess, a mere spectral delay is a different usecase for convolution. I am not even sure, if convolution and delays are the same, in principle. Maybe they are. So any delay is convolution?