Since the spectrogram is a graphical representation of a sound file, it would come in handy to have a contrast enhancement function like in Photoshop or other photoediting software.
Here paint.net for instance
Here is a usecase for this:
It would be easy to use contrast like in a photoediting software
photoshop? that’s ms paint!
I can foresee usefulness in this request…it could be especially helpful when unmixing levels I should think…particularly if contrast could be set per layer rather than overall…that might eat a lot of processing bandwidth IINM
I can’t say I understand your example, tho… @Sunnyman care to explain what your example img portrays?
It’s just some repeating noise
change to graylevel (8-bit) first. In case of 32-bit float audio, accordingly
Concerning about processing power: I actually don’t care, because the graphics card should be fast enough 
This?
and what would a contrast adjustment help you with in that case?
ambient “real” noise/ room tone / field noise I can see the use for…is your noise from a real location or synth generated / draw in?
Here you have an example: Speech in noise.
Found this article:
" Spectral contrast enhancement of speech in noise for listeners with sensorineural hearing impairment: effects on intelligibility, quality, and response times"
“This paper describes a series of experiments evaluating the effects of digital processing of speech in noise so as to enhance spectral contrast…”
There are lots of other uses for this and I think it sould be possible in Spectral Layers, since it is is rather “basic” feature in picture editing.
To do this in the spectrum is mathematically a bit more demanding, but it can be done.
For instance in forensics, as seen here:
No!
This is only for visual representation (“spectral display”).
The spectral contrast enhancement concerns the actual datapoints, which is mathematically far more expensive.
Ok, think of the following:
Take a picture and imagine you create the sound data from this picture.
Now take this picture and do the contrast enhancement with a picture editing programm, like photoshop or paint or whatever.
Now take this new, contrast enhanced, picture and create its sound data from this.
The title of your feature request does, literally, call for “contrast enhancement like in Photoshop”, which certainly suggested to me (and I’m sure, others) that it only had to do with visual representation.
The PubMed article summary you linked to doesn’t appear to describe anything SpectraLayers can’t already do; and when that was written in 1993, I doubt the authors would ever have imagined what’s possible with today’s software.
What you then go on to describe …
… is what Photosounder does, and the link to the Earshot website just looks like a screenshot from RX.
indeed, that was my interpretation
your text isn’t getting processed in my brain…you gonna need to knock up some example images to support your text to help me to better understand
iiiinterrrressssting
Danny Brown excerpts?..I watch video…now, to do that?
Wow! another dev francais! Pretty cool stuff
Ok, to clarify with pictures again.
Here the original
To illustrate use paint.net
Now chose some part of the spectum
…and increase its contrast
The selected area could be arbitrary points, not limited to a rectangle.
It would also nice, to use some aliasing or something like gaussian blur in the picture → and than see (i. e. listen) how it would look/sound like.
With all features or tools, that can manipulate pictures, this than I would call truly working in the spectrum.
As I’ve written some weeks ago, spectral editing seems to me just in its infants, compared to the sophistication we have in picture editing today.
Compare spectral editing of sound today to picture editing in the late 80s/early 90s of the last century.
(But SL is one of the frontiers of spectral editing, though. If SL would not be as good as it today already, we would not see this discussion, imho.)
See all the adjustments (don’t even look at the effects to the right…)
Almost all this could be done with the spectrum (some parts do not make sense in the spectrum of one layer, since sound does not consist of an RGB-vector. Sound in the spectrum is “just” an 8/16/32-bitplane.). That would be even more awesome, than SL is today.
Wow! Thx for mentioning Photosounder. That does a lot of those tricks already. This should be included in Spectral Layers!
It would be interesting to capture a spectrum of SL and paste it into Photosounder, just to see/listen, what is happening.
Guess, that will show us the dependence on the FFT-parameters (size and window in particular, also physical resolution of screen).
Okay! Not sure if anybody is listening to anything I’ve said or reading anything I’ve mentioned. The problem here is that Spectralayers edits waves files as-oppossed-to IMAGE FILES. There are no standards for audio for IMAGE FILES. I agree that Steinberg has the opportunity to create a standard audio IMAGE FILE (just like Steinberg created a standard for midi and VST’S) so that recording engineers can begin tracking vocals from to IMAGE FILE’s and begin mixing IMAGES rather than a wave file.
I read your every word, sir…at least the posts here at SL forum…(but not before SL10)
OK, marc, this is a bit clearer to me now you’ve kindly taken the effort to provide some images. Are those images of actual SL spectrograph or is your contrasted PIP from a different audio file?
so, as @Joey_Kapish is talking about…Marc, you are expecting the audio to follow the contrast adjustment, or are we talking purely visual aid? Because if audio was to follow the contrast setting, wouldn’t we simply observe the loss of all low level sounds as black level decreases? Then what about the other way where contrast is decreased? maybe we get a lesser difference in amplitude, blurring the sound…BUT at the lower levels, SL would have to induce noise? Based on previous selection?
As a purely visual aid, this idea would help me with manual editing, for sure…and I can see the point of contrast for audio editing too…yet, I don’t see this as part of SL in its current form…maybe a VST?
Further, (as @Joey_Kapish mentioned somewhat) vision and audio are completely different things. Physically speaking, audio is eleven times more complex than visual light…that is a lot computing power required…and we are scraping for more power in SL11 as it is.
Anyway, this is a very interesting suggestion IMO; especially in the vision only context. I think as far as audible “contrast”, a great deal can already be achieved with current SL modules, VSTs and manual tools…altho, using those would not be “as fast” as a straight up contrast tool in SL.
It is a screenshot from inside the image editor paint.net, just as a visual example what @Sunnyman wants to achieve as an actual modification with a tool or process within SpectraLayers.
In terms of “contrast” in the image world, you can think of it as “dynamics” or an “compressor” or “expander” in the audio world (to lower or heighten “contrast”, respectively).
I actually made a feature recommendation for this some time ago, but for a brush based tool (like the eraser or amplifier), so you could locally adjust the “contrast” of the spectrum (maybe for example to amplify some harmonics while reducing background noise). Of course, there must at least be either a mechanism to automatically detect the “low” level and the “high” level, or a manual setting to set the “center” threshold level from which the tool should adjust the levels accordingly.
I know there are already compressor/expander/dynamics plug-ins (VST, etc.), that work on a spectral foundation, but I haven’t really used them yet.
I don’t know of any VST-plugin, that does an arbitrary contrast enhancement of the spectrum.
If there is one, please let me know.
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.
(Which in this case, as I said earlier, is mathematically a bit more expensive, because of working with the FFT-data.
Sound editing of the spectrum is some decades behind picture editing, because of the necessary calculation power and it is more complicated and somewhat ambiguous)
