Basics about `pyvoimooo` processing
===================================

Signals
------

All input signals are supposed to be arrays of `float` values in [-1.0,+1].
You can use :func:`pyvoimooo.readwav` to read any lossless audio file to ensure this property.

Delays
------

All processing should be time synchronous. I.e. A modified waveform is **not** delayed/advanced compared to the original file.
If this happens, please send us a bug report (at rd@alta-voce.tech) with a script reproducing this unexpected behavior.


Pitch adaptive Overlap-Add (POLA) processes
-------------------------------------------

Some processing are using a POLA procedure.
This means that the :math:`f_0` value is estimated for each frame, and the window length is then adapted accordingly. In most processing, the window duration is set to 3 times :math:`T_0` (:math:`=3/f_0`).