[Freepats] Pipe organ example

[Roberto Gordo Saez]

On 8/12/06, Mark Constable <markc at renta.net> wrote:
> > - 96KHz, 32 bit floating point
> >   File size: 448M (wav), flac does not support 32 bit samples
>
> I'm not sure if the above has any sonic merit over the 24bit
> version. Would anything take advantage of 32bit floating point ?

For our purposes I think that 24 or 16 bit will be ok. 32 bit file is
too large, can't be compressed with flac and quality improvement
shouldn't be noticeable while listening.

The advantage of floating point is while editing or processing,
because of non-linearity and open range. Floating point does not clip;
for example, you can record most time at 0.0 - 200.0 range, but if
there is a loud peak or unexpected amplitude increase in middle of
recording session, it will be happily recorded using higher values
(500.0, 60000.0, 10^8...). There is an enormous dynamic range. And
there are always detail for sounds at very low amplitude. For example,
between values 0.0 and 0.01 there are still lots of decimal values. As
another advantage, less quantization noise is added when processing or
mixing.