Introduction

Time Buffer

The calculus time of an analysis varies depending on the analysis parameters.

Window Size and Buffering Time

Calculus

The window size determines how much time takes to the FFT Buffer to get full and generate a complete spectrum

  • Filling Time Buffer = (Window Size/SR) * signal duration

  • Filling Time Buffer = (1/FR) * signal duration

This is proportional to the period of the frequency resolution, so to speak.

The higher the frequency resolution of the analysis, the longer the filling time buffer is.

Example

Let's take a 10 sec. signal for instance, with a 1024 FFT Size and 44100 SR. We get :

FTB= (1024/44100) * 10 = 0,23 s

FTB = 1/(44100/1024) * 10 = (1/43) * 10 = 0,23

With a 4096 FFT Size and 44100 SR. We get :

FTB= (4096/44100) *10 = 0,93

FTB = (1/(44100/1024)) * 10= 1/43 = 0,93

FFT Size and Buffering Time

Oversampling Increase

We have seen that the size of the FFT – number of bins per window – could be increased proportionnally.

Buffering Time

The buffering time increases linearly with the oversampling rate. If the oversampling rate is multplied by 4, the buffering time will be also multiplied by 4.

Window Step and Buffering Time

Oversampling Increase

The window step increases the number of times a window is calculated.

Buffering Time

The overlap factor increases the precision of the calcucus, and increases the calculus time proportionnally.

  • Time Buffer
A propos...IRCAMRéalisé avec Scenari