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
One thing: would it not be better to calculate the weighing factor only once and not in every computation ?
Or am i wrong ?
In the Windowing Function it says:
this->_vReal[i] *= weighingFactor;
and then factor is computed above.
If then factors where computed once (in the construnctor ??) one could use the factor, but it would not be neccessary to compute the factor each time, it would cost memory, ok, but one could perjhaps choose to do it that way or the other.
Today (31.01.2023) i measured the conversions:
with Windowing in every loop i get 20 FFTs per second in my code
in a test i filled the array with 1.0, d#started FFT.windowing() and stored the values in another array. then i multiplied the incoming values with this stored factors. now i get 30 FFTs per second.
2nd edit:
Still in the develop-branch the hann-windowing is wrong:
instead of
weighingFactor = 0.54 * (1.0 - cos(twoPi * ratio));
it hast to be
weighingFactor = 0.5 * (1.0 - cos(twoPi * ratio));
see https://en.wikipedia.org/wiki/Window_function
Thnx for this lib, great work.
One thing: would it not be better to calculate the weighing factor only once and not in every computation ?
Or am i wrong ?
In the Windowing Function it says:
this->_vReal[i] *= weighingFactor;
and then factor is computed above.
If then factors where computed once (in the construnctor ??) one could use the factor, but it would not be neccessary to compute the factor each time, it would cost memory, ok, but one could perjhaps choose to do it that way or the other.
Today (31.01.2023) i measured the conversions:
2nd edit:
Still in the develop-branch the hann-windowing is wrong:
instead of
weighingFactor = 0.54 * (1.0 - cos(twoPi * ratio));
it hast to be
weighingFactor = 0.5 * (1.0 - cos(twoPi * ratio));
see https://en.wikipedia.org/wiki/Window_function
or just use https://www.desmos.com/calculator?lang=de and put in both variants, you will see the difference:
with x from 0 to 1, which would represent the ratio.
The text was updated successfully, but these errors were encountered: