version 4.5.0

[hvianna]

What's new

Added:

• fadePeaks - peaks fade out instead of falling down - thanks @TF3RDL for this idea;
• gravity - customize peak drop acceleration;
• peakFadeTime - customize peak fade-out time;
• peakHoldTime - customize peak hold time;
• UMD module export - allows using the module via require() and provides a global variable for easier usage in the browser - thanks @shahkashani for the suggestion (#71 and #72); see the Usage section for more details.

The new gravity equation restores the original peak drop behavior and decay curve (using the default value) of versions 1.0.0 through 3.4.0.
Check out the peak drop analysis tool for more information on the evolution of the equation and to compare different values of gravity.

As usual, you can interact with all the new properties in the fluid demo.

---

This discussion was created from the release version 4.5.0.

[TF3RDL]

thanks for bringing the idea of fade out peaks from analog-style spectrum analyzer/sliding DFT visualization project, which I've added it into my aforementioned CodePen project because I saw an audio visualizer thing (basically a peakmeter) in Geometry Dash 2.2

BTW, I think ditching the AnalyserNode's built-in FFT entirely (in favor of custom FFT and other audio analysis functions that don't actually rely on WebAudio at all) per #30 would makes audioMotion-analyzer more feasible for non-realtime analysis and even analysis (both realtime and non-realtime) of non-audio signals in the absence of Web Audio API since the AnalyserNode is designed for realtime analysis and it is a part of Web Audio API, and in-addition, would use the full potential of AnalyserNode.getFloatTimeDomainData and AudioWorklet for realtime audio analysis

[hvianna]

Yes, I think it would be great to have these alternate analysis functions in addition to what WebAudio provides.

The first thing I need to work now is on separating the audio data processing from the graphics rendering in the _draw() function, so it will be easier to integrate different analysis methods.

I'll be doing this in a new v5 branch, for the next major version. Would you be willing to help with PRs to bring in your implementation of the custom transforms?

[TF3RDL]

True as separating the audio data processing from drawing function can help with integrating custom analysis functions and maybe even makes it easier to add infinite average spectrum in-addition to main bar peaks (much like foo_enhanced_spectrum_analyzer) as in this updated CodePen project about "analog-style" spectrum analysis right?

BTW, a linear peak decay function as in this CodePen project linked above (and foobar2000's built-in Spectrum and foo_enhanced_spectrum_analyzer visualization) would be great addition to audioMotion-analyzer