You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Turns out, it is possible to have non-integer k on CQ-sDFT algorithm, which means analysis frequency can exactly corresponds to musical notes instead of approximate one (as mentioned in #3) and it also makes it easier to convert into variable-Q transform like this, it just need one additional factor, namely fiddles
The text was updated successfully, but these errors were encountered:
Hey, this is amazing! Thanks for sharing!
And big thanks for doing the implementation :)
May I link your gist from the references section of my project?
May I link your gist from the references section of my project?
Yes, for somebody besides you and me who make foobar2000 components (by using my JS project as a reference and re-implement the same algorithm in C++ for use in foobar2000 visualization components), especially the fact a lot of components like Musical Spectrum are no longer developed (thus no 64-bit version for use in foobar2000 x64) and are only usable within 32-bit version of foobar2000
to add to this, I've made this CodePen project demonstrating the sliding DFT with noninteger K allowing variable-Q transform with smoother low frequency peaks and also a faster IIR cousin, the sliding windowed infinite Fourier transform (SWIFT) and even analog-style filter bank (which some professional RTAs including Spectralissime uses it over FFT) and a YouTube video demoing the good-old algorithm not based on FFT
Turns out, it is possible to have non-integer k on CQ-sDFT algorithm, which means analysis frequency can exactly corresponds to musical notes instead of approximate one (as mentioned in #3) and it also makes it easier to convert into variable-Q transform like this, it just need one additional factor, namely fiddles
The text was updated successfully, but these errors were encountered: