How adjust softsynth low/high keys level
-
- KVRer
- Topic Starter
- 28 posts since 4 Feb, 2019
"TAL-UNO-LX V2" is known for perfect raw oscillator shapes.
Following pictures show a raw saw osc first playing note C1 and then note C5 in 96kHz/24bit. As you can see the same situation here.
(The previous frequency spectrums pictures (see above) were down in 48kHz/24bit.)
Following pictures show a raw saw osc first playing note C1 and then note C5 in 96kHz/24bit. As you can see the same situation here.
(The previous frequency spectrums pictures (see above) were down in 48kHz/24bit.)
You do not have the required permissions to view the files attached to this post.
-
- KVRian
- 626 posts since 15 Jun, 2015
Well, I'm out of ideas.
I've tried this test on three different softsynths and three hardware synths, and I can't reproduce what you're seeing. Tried it with square and saw, but all my results look like the screenshots I shared.
What you're seeing is not normal. Hopefully someone else will come with something.
I've tried this test on three different softsynths and three hardware synths, and I can't reproduce what you're seeing. Tried it with square and saw, but all my results look like the screenshots I shared.
What you're seeing is not normal. Hopefully someone else will come with something.
-
- KVRer
- Topic Starter
- 28 posts since 4 Feb, 2019
@RichieWitch
Finally I found out why your spectrum analyser and mine differ. It has to do with the "tilt" or "slope" setting most spectrum analysers have. Fabfilter Pro Q3 manual says about this:
"The Tilt setting tilts the measured spectrum around 1 kHz with a specified slope, expressed in dB peroctave. The default setting of 4.5 dB/oct results in a naturallooking spectrum, resembling best how loudness is perceived by the human ear."
So after I set the tilt setting to 0.0 dB the spectrum analyses looked the same as yours. Meaning the analyser curves got the same maximum level for notes C1, C3 and C5.
Before I was testing different spectrum analysers and wondered why they all differ. And then realized that some of them got the slope option enabled others not. I tested the best free spectrum analysers "Voxengo SPAN", "MeldaProduction MAnalyser" and the build in "Spectro Meter" in Studio One.
So all my Fabfilter Pro Q3 spectrum analyser pictures I provided above are done with the default 4.5 dB tilt setting that provides: "a naturallooking spectrum, resembling best how loudness is perceived by the human ear," as Fabfilter Pro Q3 manual says.
The 4.5 dB tilt /slope setting is the default setting that Fabfilter Pro Q3 comes with Voxengo SPAN spectrum analyser got the same default setting. Its manual says:
"Note that by default SPAN uses 4.5 dB per octaveslope for the spectrum display which makes it look considerably “elevated” towards the higher frequencies in comparison to most other spectrum analyzers available on the market. This setting can be changed in the “Spectrum Mode Editor” window."
I think both respected products do that for good reason: Man is not a robot. Our ears don't have linear frequency perception. Therefore Fabfilter and Voxengo default settings are tuned to our ears.
No wonder that my ears (that are quite good) just hear what the spectrum analyser of Fabfilter Pro Q3 is showing with that default tilt/slope setting.
So the problem stays the same. My ears hear that soft synths play low notes too quite and high notes are too loud. But at least now I know why.
Thanks to all of you who helped me on that way!
Finally I found out why your spectrum analyser and mine differ. It has to do with the "tilt" or "slope" setting most spectrum analysers have. Fabfilter Pro Q3 manual says about this:
"The Tilt setting tilts the measured spectrum around 1 kHz with a specified slope, expressed in dB peroctave. The default setting of 4.5 dB/oct results in a naturallooking spectrum, resembling best how loudness is perceived by the human ear."
So after I set the tilt setting to 0.0 dB the spectrum analyses looked the same as yours. Meaning the analyser curves got the same maximum level for notes C1, C3 and C5.
Before I was testing different spectrum analysers and wondered why they all differ. And then realized that some of them got the slope option enabled others not. I tested the best free spectrum analysers "Voxengo SPAN", "MeldaProduction MAnalyser" and the build in "Spectro Meter" in Studio One.
So all my Fabfilter Pro Q3 spectrum analyser pictures I provided above are done with the default 4.5 dB tilt setting that provides: "a naturallooking spectrum, resembling best how loudness is perceived by the human ear," as Fabfilter Pro Q3 manual says.
The 4.5 dB tilt /slope setting is the default setting that Fabfilter Pro Q3 comes with Voxengo SPAN spectrum analyser got the same default setting. Its manual says:
"Note that by default SPAN uses 4.5 dB per octaveslope for the spectrum display which makes it look considerably “elevated” towards the higher frequencies in comparison to most other spectrum analyzers available on the market. This setting can be changed in the “Spectrum Mode Editor” window."
I think both respected products do that for good reason: Man is not a robot. Our ears don't have linear frequency perception. Therefore Fabfilter and Voxengo default settings are tuned to our ears.
No wonder that my ears (that are quite good) just hear what the spectrum analyser of Fabfilter Pro Q3 is showing with that default tilt/slope setting.
So the problem stays the same. My ears hear that soft synths play low notes too quite and high notes are too loud. But at least now I know why.
Thanks to all of you who helped me on that way!
Last edited by mireiner on Thu Mar 28, 2019 1:01 am, edited 4 times in total.
-
- KVRian
- 626 posts since 15 Jun, 2015
-
- KVRian
- 629 posts since 15 Jun, 2017
Haha, physics versus perception
Fletcher & Munson did some fundamental research on loudness perception, almost a 100 years ago.....
https://en.wikipedia.org/wiki/Fletcher% ... son_curves
When tuning is equal temparament each next note frequency is a factor 2^(1/12) = 1,05946309 Hz higher then the previous. Each octave doubles the frequency.
Notes & Frequencies (on a piano):
https://en.wikipedia.org/wiki/Piano_key_frequencies
C1 = 32.70 Hz (where the human ear is not so sensitive)
C3 = 130.8 Hz (where the human ear is more sensitive)
C5 = 523.2 Hz (where the human ear is even more sensitive)
This continues into the thousands of Hz....
C8 = 4186 Hz (whereafter the sensitivity starts to drops off again)
Apparently human ears are conveniently the most sensitive to the most relevant frequencies (stuff like speech lives here).
Fletcher & Munson did some fundamental research on loudness perception, almost a 100 years ago.....
https://en.wikipedia.org/wiki/Fletcher% ... son_curves
When tuning is equal temparament each next note frequency is a factor 2^(1/12) = 1,05946309 Hz higher then the previous. Each octave doubles the frequency.
Notes & Frequencies (on a piano):
https://en.wikipedia.org/wiki/Piano_key_frequencies
C1 = 32.70 Hz (where the human ear is not so sensitive)
C3 = 130.8 Hz (where the human ear is more sensitive)
C5 = 523.2 Hz (where the human ear is even more sensitive)
This continues into the thousands of Hz....
C8 = 4186 Hz (whereafter the sensitivity starts to drops off again)
Apparently human ears are conveniently the most sensitive to the most relevant frequencies (stuff like speech lives here).
Last edited by Kwurqx on Thu Mar 28, 2019 4:08 pm, edited 1 time in total.
-
- KVRAF
- 4589 posts since 7 Jun, 2012 from Warsaw
Which ones exactly? As far as I know, every softsynth I own has key tracking as a modulation source. This is so common feature I can't quite see how you can have any issues with that.Internal virtual instrument option: "volume key tracking"
Drawback: I was told that some hardware synthesizers have this option. But none of my softsynths provide it.
Spectrum analysers often use pink noise normalization. Because they are intended for musical purposes in a studio, not electrical measurements.Finally I found out why your spectrum analyser and mine differ. It has to do with the "tilt" or "slope" setting most spectrum analysers have. Fabfilter Pro Q3 manual says about this:
That's why there are different tools to measure amplitude, such as oscilloscoper or plain simple bar meters
Blog ------------- YouTube channel
Tricky-Loops wrote: (...)someone like Armin van Buuren who claims to make a track in half an hour and all his songs sound somewhat boring(...)
Tricky-Loops wrote: (...)someone like Armin van Buuren who claims to make a track in half an hour and all his songs sound somewhat boring(...)
-
- KVRian
- 629 posts since 15 Jun, 2017
O...BTW,
You could do a slow clean sine pitch sweep through an analyzer and see how that translates to representaion of energy levels / amplitudes.
Alternatively: send true white noise through an analyzer. All frequencies should be present at equal levels (so a straight horizontal "frequency curve" as you called it).
You can then be sure that the actual amplitude/energy level (of the sine/signal) is constant. This would therefore eliminate the "note" / keytracking / patch / programming variables.
Also, make sure there are no processors like filters, EQ, compressors etcetera in the signal path (which could alter the actual output levels during the sweep).
You could do a slow clean sine pitch sweep through an analyzer and see how that translates to representaion of energy levels / amplitudes.
Alternatively: send true white noise through an analyzer. All frequencies should be present at equal levels (so a straight horizontal "frequency curve" as you called it).
You can then be sure that the actual amplitude/energy level (of the sine/signal) is constant. This would therefore eliminate the "note" / keytracking / patch / programming variables.
Also, make sure there are no processors like filters, EQ, compressors etcetera in the signal path (which could alter the actual output levels during the sweep).
-
- KVRian
- 629 posts since 15 Jun, 2017
More nerdy stuff:DJ Warmonger wrote: Thu Mar 28, 2019 2:50 pm
Spectrum analysers often use pink noise normalization. Because they are intended for musical purposes in a studio, not electrical measurements.
Pink noise is like white noise, but with a -3 dB per octave downward tilt towards the higher frequencies. Somehow a "natural" energy distribution.
So it's sort of like the amplitude drop off of a true saw or square spectrum....but with ALL frequencies present instead of just multiples of the root partial's frequency (all or odd).