Input value range for data?

[Franky1]

Wave audiofiles are often created with int16 audio data, but pyloudnorm seems to require float arrays as input for the data arguments.

How do the float values have to be scaled respectively what is the maximum numerical range for the full scale so that the calculations by pyloudnorm make sense?

Is it [+1.0; -1.0] in float?

[iver56]

Yes, the -1 to +1 range is common for float32 audio arrays. It is common to process audio in float32 even though the storage format is something different (e.g. int16 wav)