WAV compression code 3

I’m trying to use a program called Endless WAV to embed loop points in some files and I’m noticing some bizarre behavior that I’ve traced back to Audacity. It seems that Endless WAV doesn’t like any of the files I’ve created recently with Audacity and as far as I can tell it’s because when I get info on them in Endless WAV it says that they have “compression code: 3”. These are all 44.1k 32-bit WAV files. They appear elsewhere to be indistinguishable from other 44.1k 32-bit WAV files but when I load others into Endless WAV they show as having “compression code: 1.” I don’t even know what this means - nowhere in the export options for Audacity does it ever give me an option to select a compression type for that file format, and in fact I thought that this was an uncompressed format. I can even import other 44.1k 32-bit WAV files into Audacity and then re-export them and they will show up as compression code 3 in Endless WAV. So somewhere something about my Audacity exports is wrong, and I can’t figure out where this setting would even be.

Codec ID 3 is (32-bit) float format.
Codec ID 1 is integer format.

Thanks, I just figured out that the problem is that Audacity has been lying to me about the format of certain files. If drag 16-bit WAVs into Audacity, they show up as 32-bit float. So the ones I thought were 32-bit were actually not. So that’s a separate problem, but at least now it makes sense why I was seeing code 3 on those other files which are actually 32-bit.

Audacity can produce different types of 32-bit audio file …
https ://manual.audacityteam.org/man/other_uncompressed_files_export_options.html
The “float” ones are “compression code: 3”.
two-types.gif

Audacity isn’t lying. By default, imported files are converted to 32-bit float format. It is highly recommended to NOT change that setting.

Why??

The default 32-bit float resolution gives the highest quality, especially when processing audio, works natively with IEEE floating point representation, has sufficient dynamic range for even the most demanding audio work, and most importantly does not clip.

The only downside is that it requires more disk space than 16 or 24 bit integer formats. Given that the cost of disk space has dropped from $10000+ per GB in the early 1990, to less than 3c per GB in the 2020’s, there’s very little reason to compromise on quality, safety or convenience when selecting the default sample format - just leave it at the default.