Skip to content

Latest commit

 

History

History
44 lines (35 loc) · 1.61 KB

TODO.md

File metadata and controls

44 lines (35 loc) · 1.61 KB

toilet kitchen cabs mirrors microwave palish appliances/clean air fryer trash maintain bathroom counter/sink tub sweep/vac mop put away dishes maintain

Maybe todo: contour pitch mean should be magnitude weighted ir generation - https://dsp.stackexchange.com/questions/41696/calculating-the-inverse-filter-for-the-exponential-sine-sweep-method/41700#41700 Audio/PV conversions can compute all non-overlapping windows at the same time using this http://fftw.org/fftw3_doc/Advanced-Complex-DFTs.html perturb sucks currently but might have potential using something like simplex noise, see comments on process it also could be placing frequencies outside bins that can hold them, use the bin shift strategy as in PV::time_extrapolate PV::time_extrapolate could use a Func2x1 instead of interp, not sure the best way to set it up though

Process ideas: Audio::limiter look-ahead option in compressor would take care of this Audio::split_at_frequencies for multiple splits PV::align_harmonics PV::chorus might work, based on expanding partials Per frame we should be able to detect partials and expand them The main issue I see is partial continuity

Task: Something's in the Lake at Bagel Woods

in prism, this: "bound_bin( frequency_to_bin( freq ) - 10 );" and the following line use 10 arbitraryly
	This seems bad to use without testing, but it might be fine
	notes_close may also be a bad metric

factor get_pointers to... somewhere

ohhhh maaaaaan time_to_frame always rounding down might be causing extremely small issues all over

explodes:
	synth.filter_1pole_multinotch( 5, []( Second t ){ return 10000*t; }, .9 );