3.0.1 Well Tested

With the advance of automated tests, I thought it prudent to create a new release.

Preliminary listening seems to make it seem as if this is the best performing release so far,

Note that the sliders for phase width are now exponential, so they range from -5 to 5 now, not unlike the strength

There should be less popping frequency artifacts :) 🥇

v2.0.3-beta1

Upload latest dev branch as a DLL compiled with VisualCodeBase

v2.0.2

small bugfix:

process the Nyquist band, so that Nyquist noise doesn't leak through when both mix sliders are set to 0

v2.0.1

Minor changes to v2.0:

Bugfixes:

• fixed input and output corrections - should fix panning

Changes:

• Uses MLT (sine) window instead of Hahn window (sine)

Known issues:

• CPU usage is higher with 2.0, perhaps try to optimize, or release a low-fidelity low usage version
• silence is not true silence: I suspect that either the lowest coefficient or highest fft coefficient leaks through, despite setting both mixes to 0.

STFT Release

Change the internal FFT to use the STFT template from https://forum.cockos.com/showthread.php?t=225955
Should result in better sound quality (less transient smearing / mushiness).

Main release

With the improvements from this version that fixed the main clicking issues, I feel this is now ready for 1.0.0:

Change-log:

• changed default slider values
  • will get no sound when first loading the effect due to it defaulting mixes to 0
• add experimental phase2 - controls the difference in phase that the algorithm looks for
  • almost always should stay as 0, (note for self: it might also make sense to implement this as an input correction instead)
• add second window
  • should fix issue with clicking on certain tracks with certain settings
• add input buffering
  • as a last resort to try to fix clicking, except only implemented for input as of yet
  • not intended to be changed while the plugin is running

One more window function

added triangle window - default window is still Hann,
potentially is the window with the least amount of artifacts with two overlapping tiles (50%) overlap, which is how the plugin currently works
(Probably should have added this window function with the other ones.)

Small fix

0.4.1:
Make sure not to process band 0, even if low cut is set to 0, since it can cause clicking (discontinuity in the waveform)

0.4.0:
Windowing: add ability to switch between one of five different fft window types, (4 are new. Selecting the hann window - which is the default option - works exactly the same as the plugin before)

Windowing

Add four new fft window functions and a new selection list to switch between them.

Minor fixes after new algorithm

0.3.1 changes

fix issue with larger fft sizes (have to change in code, not exposed to a slider)

0.3.0 changes

• algorithm rewrite:

  • different phases get attenuated in center signal by a linear amount when strength is 0. When in phase, it will not be attenuated. When out of phase by phase width or more it will be attenuated to 0 (can set phase width linearly over spectrum).
  • phase width control should be more consistent in cases where amplitude difference is great.
  • new slider allows for controlling whether a difference in amplitude will cause attenuation. The value of this slider can interpolate between 0 and 1.

• Allow outputting with initial phase adjustment undone, or left as is. See example usage in readme.md

  • has no effect when "Phase (degrees)" is equal to 0.