There should be conversion routines for the 24-bit floats (f24). I think maybe: * [x] ~~f24 <== string~~ * [x] ~~f24 ==> single ("single" being this lbrary's bastardized format)~~ * [x] ~~f24 <==> binary32~~ * [ ] f24 <==> binary16 * [x] ~f24 <==> u8~ * [x] ~f24 <==> u16~ * [x] ~f24 <==> i8~ * [x] ~f24 <==> i16~