32bit float vs 24bit...

I remember watching a Steinberg mixing tutorial video years ago where the presenter stated in no uncertain terms that projects should be set to 32bit floating point depth because this allows you to record with lots of headroom left over and also allows you to mix in any way you want and there will be no internal clipping.

Ever since then I’ve mixed every song in 32bit float.

However, I’ve since heard over and over (most recently in The Home Recording Show podcast) that all those same things apply to 24bit recording and that all DAWs internally use 32bit floating point precision for mixing. So that there’s no need to record at 32bit float as it just uses unnecessary disk space.

So what’s the truth? Am I just wasting disk space?

The headroom on recording / AD conversion is the same for 16, 24 and 32 FP bit.
32Bit FP can be useful when recording with VST Fx.
Setting the project setup to 32 bit FP is useful prior to offline processing, though that’ s independent of the recording format. More info in the manual and the many already existing threads on the forum.
And yes - at least Cubase uses 32bit. FP internally anyway.

Yes, you are. Recording with anything over 24 bits is waste, since there are no A/D converters capable of giving more resolution. Internal VSTis may use higher resolution, but IMO it’s doubtful you’ll hear a difference here between a 24 bits and 32fp bits audio track. All internal DAW calculations are indeed done with 32fp bits, so it may be recommendable to mix to a 32fp bits file, for later mastering with the highest quality starting point.

::::To quote the all mighty Split:::

I bounce a lot, especially after all editing is complete and I am ready to mix. Oh, I also record in 32BitFP and have been for years.

So why does Steinberg’s own tutorial video recommend emphatically setting your project to 32bit float?

because it’s possible to break the headroom by rendering…