I think my point number 2 should be the most impotant thing in an anylizer even a free one.
"2. Real time ballistics with no fall time and no averaging at all. Period. Like the fastest ballistics in your curve eq plugin version 1.9c - after that your fastest time was still slow. This helps to 'see' sound, like kick drums. eg. In Span a kik pattern of 8th notes reads louder than the same kik with 4th notes."
Music is real time. As humans we recieve the information as sound presure waves in real time. So I prefer to see the information that represents the truth. (see my point above regarding 4th notes and 8th notes - try it).
Whilst I'm here blabing away, do you think gliss eq could have an inverse dynamic, different to what it does now? Instead of turning up a frequency and then squashing it, what about turning a freqency up by expanding its dynamics ( an uncompressor ). You can do this in Soniformer already.
Pboy, I may add a new 'inf' speed option later. However, I agree with Andrew that non-averaged spectrum is not what can be useful in general.
I'm also planning to experiment with the 'dynamic' thing you've mentioned. Possibly, I can make an 'average dB per second' dynamics indication so that spectrum areas which are dynamic or static can be seen.
Have you seen the historgrams in Izotopes stuff and TC electronic limiter? The goal would be to represent a histogram for each frequency band. Instead of a 2d graph for the signal level history, a thin bar (1d) with color to represent values, hence it becomes 2d again. Then you can play with the balistics and what have you and you should end up with a 2d plane like curve eq etc but with another dimension, ie - history of peaks values. Please send my royalty cheques to P.O. Box,,,
Auditory filter stuff
Gammatone filter stuff
NB equation 4 is wrong (it should be 4.37/1000 not 4.37)
However I think substituting a gaussian curve to "spread" the data in the frequency domain would be a pretty reasonable Gammatone substitute, and maybe easier to implement. If you do this then the standard deviation of the Gaussian curve needs to be set so it corresponds to the correct Equivalent Rectangular Bandwidth (ERB) for each given frequency. If you want to persue it then let me know and I can give you the correct formula.
However, please don't sacrifice development of IM etc. for the sake of a Spectral Analyser ;-)
Thanks for those URLs.
Andrew, please send the relevant formula to firstname.lastname@example.org. Gaussian filtering seems like approach I'm looking for, at least I wanted to implement some kind of averaging into spectrum drawing code.
This topic was last updated 180 days ago, and thus it was archived. Replying is disabled for this topic.