From pro-level field recorders to consumer lavs and camera add-ons, it's one of the first specs you see listed on ads, videos, and spec sheets. However, the bit depth is only one piece of the puzzle when it comes to high-quality recordings, and there are plenty of recorders with 32bit float capabilities that won't solve your problems. It takes a combination of good hardware and technique to get your recordings to sound their best. A lot of pro-sumer recorders that boast 32bit float as a selling factor are really utilizing the auto-mix functions that go hand and hand with the higher bitrate. For example, recorders like the Zoom F3,the AV-FR2, and PR-2 that do not have dual A/V converters make use of auto mix to keep your recordings at a good level and 32bit float as a backup feature to save your ass in unortunate scenarios.
Bit depth refers to the amount of information captured in each sample of audio collected - the higher the bit depth the higher the dynamic range. Recording at higher bit depths also comes with more power required and higher file sizes. Compared to video recording, these files sizes are still extremely small, so I wouldn't worry about that. You don't even need the fastest cards in your recorders to capture 32bit float audio. The human ear can only hear about up to 120 db, which is painfully loud. We can't discern between 24bit and 32bit audio, so you can't argue that 32bit sounds better than 24bit.
There are lots of other technical factors that determine the quality of audio recorded - such as sample rate and bit rate, as well as the quality of the hardware recording the audio. 32bit float allows us to capture more information than 24bit integer by using floating point data that simultaneously captures fixedi ntegers and exponents, allowing for much larger numbers to be represented. Technically, this does mean that the higher the number, the less accurate it is. I won't get into the math and science of it, but know that just because we are catpuring more data does not necessarily mean we are clip-free. Higher-end recorders use dual A/D converters that specialize in quieter and louder sounds. They then blend those signals together. The different recorder and the quality of the converters, pre-amps and input hardware still matter. The signal can clip before it hits the digital stage, and there's no brining it back from that. If you notice that the audio you capture from a 32bit float file still sounds clipped or degraded, it's probably because the clipping occurred before the conversion to digital. Sounds like wind noise, scratches from lav movement, etc. can absolutely cause this.
If I were to buy a 32bit recorder as my main device, I would make sure that it has dual A/V Converters to get the best out of the 32bit float feature. When it comes to features that you are looking for, I would consider other aspects of the recorder before worrying about 32bit float capability. The cleanest pre-amps and best converters are the most important thing to me, as well as a good form-factor, robust design, and interface that makes sense. For example, the Sony M10 and legendary D100 have fantastic pre-amps even though they lack more modern features like 32bit recording. In most scenarios, I don't mind this at all because my main focus is the quality of my recording when using these recorders. Maybe I'm out collecting nature sounds for a sample pack or film I am working on, and quick turnaround isn't my main worry. If I were on set of a documentary running through the forest with an athlete, I might be more inclined to pick a more modern recorder with a faster UI and 32bit float capabilities to save me from minor mistakes and deal with less-perfect pre-amps. Just like all aspects of field recording and audio for video, there will always be trade-offs at hand and the best audio designers are able to think quick on their feet about what the best decision is in front of them.
