32-bit float: Love. Hate.

I’m not really an audio guy, but I do know that good audio is super important; and bad audio can literally ruin your video.

As a solo video creator, a lot of times I’m focused on capturing the visual action that is happening.

I do set up my audio equipment, but I just rely on what I can capture with that set up while the camera is rolling. A lot of times the audio is beyond my control and I’ve ended up with audio that was too soft to be heard clearly, or too loud and completely distorted.

That’s why I was excited when Panasonic announced the XLR2 microphone adapter alongside their new flagship micro four thirds Lumix GH7 camera.

This is the first camera to record 32-bit float internally and I think this could be a game changer for many independent filmmakers.

I wanted to share some of the things I’ve found after using it on a few projects now.

Compared to the version 1 adapter, the new XLR2 is larger and comes with a microphone clamp that you can remove if you don’t need it. There’s also a new 3.5mm microphone jack as Input #3, that I’ll talk about a bit later.

The construction is still plastic, but sturdy. And it’s pretty lightweight at just 191 grams.

As of this post, the XLR2 works with the Lumix GH7, and also with the GH6 via a firmware update. Although, on the GH6, you cannot take advantage of the new adapter’s biggest strength which is 32-bit float audio.

On other Lumix current cameras, the XLR2 will not even be recognized, but I expect Panasonic will fix this in future firmware updates. Although, which other cameras might be able to take advantage of 32-bit float remains to be seen.

So, what does 32-bit float audio mean for content creators and videographers?

32-bit float gives you much wider audio dynamic range than the typical 24-bit recording most cameras use. It allows both very quiet and very loud sounds to be captured without distortion.

This means that as a creator, you have much more flexibility in editing, to boost quiet sounds, or bring down loud sounds without losing quality.

Standalone audio recorders started offering 32-bit float audio recording a few years ago.

To use them with your video, you’d need to synchronize the separate audio file during editing.

However, the XLR2 with the GH7 captures 32-bit audio directly in Quicktime movie files, eliminating the need for manual synchronization.

Side note, 32-bit float audio is only available when recording in Quicktime; not MP4 and not Apple ProRes.

Here are 3 things I love about 32-bit float and using the XLR2 adapter.

First, it’s the flexibility in editing that it unlocks.

If you’re a photographer and shoot in RAW, 32-bit float is like the audio version of that capability.

With RAW photos, you can often recover details in the highlights or shadows and bring out the best image your camera can generate, rather than trying to edit a Jpeg and ending up with poor results.

Here’s a quick demonstration of the difference between 24-bit and 32-bit float recordings. This is a situation I run into quite frequently when I am recording cooking videos. Probably a lot of wedding, documentary, and event videographers also have to deal with sounds that are too quiet, or can suddenly get really loud and start to clip in your recording.

First is the original version 1 XLR adapter on my GH7, recording in normal 24-bit audio.

When the hand mixer starts, it completely overloads audio.

Even if I lower the level in software, you can see that there is a hard cut off, and everything outside is completely gone from the recording. There isn’t much you can do here except to try to re-record, or just accept the bad audio.

Now attaching the new XLR 2 adapter, you can see a new option in the menu to use just the XLR adapter, rather than XLR + Camera.

Selecting this, you can now record in 32-bit.

Notice that the audio level meters no longer show the orange and red zones, and the gain dials on the adapter are simply disabled because you don’t need them.

When the mixer starts, the audio is not totally clipped like the 24-bit recording. You can lower the levels and keep good audio fidelity.

In theory you can also boost quiet audio more cleanly in 32-bit float. But unless you’re recording in a perfectly quiet sound studio, so far I’ve found this to be pretty limited in real world usefulness.

I’m using Rode NT5 small diaphragm condenser microphones, which don’t have a -huge- amount of self noise, but together with room sounds like the refrigerator to my side, this studio is not perfectly quiet.

I recorded some knife sounds, but boosting the 24-bit and then the 32-bit float recordings, I really couldn’t get any more useful results out of the 32-bit recording.

So, your mileage may vary for quiet recording.

Another nice thing about the XLR2 adapter is the extra 3.5mm jack that acts as Input #3 and records to audio channels 3 and 4, while also recording in 32-bit float.

A number of newer Lumix cameras have given you access to 4 channel recording if you attach the older XLR1 adapter, but you needed to use the camera’s own microphone jack to record to channels 3 and 4, and of course you were limited to 24-bit audio for everything. With the XLR2 adapter, as long as you connect everything to the adapter, you can record in 32-bit float across all 4 channels.

While you could even use a wireless system and plug it into the 3.5mm jack, keep in mind that the wireless system will have its own preamp and gain settings. The XLR2 will be downstream from that, so you probably won’t get any benefit from 32-bit float on that input #3.

But even with that limitation, you -will- get 32-bit float on any XLR mics you connect to the adapter, so it may still be worth plugging a wireless system into the 3.5mm jack on the XLR2 rather than the camera itself.

Sometimes I need to record conferences and presentation that go for an hour, or two, or even all day. I’ve used separate audio recorders in the past, and unless you have a timecode system to lock the signals together, things can drift out of sync over time.

The nice thing about the XLR2 adapter is that I can record in 32-bit float over long durations, and have the audio stay in sync with the video.

Okay, but of course it’s not all wine and roses.

Here are 2 things I hate about working with 32-bit float on the XLR2 adapter.

Maybe other editing software handles it better, but I use Adobe Premiere and while it handles 32-bit float audio, there are a couple of things that I find to be troublesome.

First is that when you drop a clip with 32-bit float onto your timeline in Premiere, the audio plays back about 12 decibels lower than what I’m used to hearing when I edit a 24-bit project.

I could turn up the volume on my headphones, but then of course everything outside of Premiere gets that much louder… i.e. too loud.

Also, I’m still in a mixed audio environment, where my GH7 footage is now recorded in 32-bit float, but clips from other cameras are recorded in 24-bit. When I mix 32-bit and 24-bit on the timeline, because the 32-bit clips play back more softly, the audio levels jump from clip to clip.

What I’ve taken to doing is keeping 32-bit clips on one audio track and boosting that track’s level by 12 decibels; then I put 24-bit clips on separate track. This allows me to at least hear things at a consistent volume while I’m editing.

However! I always do my audio cleanup in Adobe Audition for its better audio controls than Premiere.

Until now I would just mix everything down into a single audio file, take it into Audition and clean up there. But with 32-bit float, I first need to lower the level in Premiere and get rid of the 12 decibel boost I gave it, then export just the 32-bit track as a wave file for any eq, compression, and clean up in Audition. But I’m left with my 24-bit audio on a separate track that needs to be handled separately, so my audio workflow has sort of doubled with having to clean up 2 separate tracks now that I’ve included 32-bit float.

If anyone knows a better way of doing this, or maybe if DaVinci Resolve is better for audio and can avoid the extra steps, please let me know in the comments. I’d love to hear from you.

When using 32-bit float on the GH7, you lose the option for realtime audio monitoring. The camera is locked on “recorded audio” which lets you hear exactly what’s being recorded.

This sounds nice in theory, but because I’m a solo shooter, I often need to give directions for the shot.

With recorded audio, there’s a split second delay between when I speak, and then hearing myself through the monitoring headphones.

This is really jarring and my brain isn’t used to processing my own voice with a delay. For one word directions, it’s not so bad. But if I need to say more than a word or two, I now have to put the camera down, and take off the headphones so I’m not distracted by my own delayed voice, before giving directions.

I’m not sure if this is anything that can be fixed with a firmware update, but Panasonic, if you see this, it would be very helpful to keep realtime monitoring as an option when using 32-bit float.

So that’s it. Those are the 3 things I love, and 2 things I hate about working with 32-bit float on Panasonic’s new Lumix GH7. I hope this was helpful for anyone thinking about getting the XLR2 adapter, and if anyone has advice on things I can do to remedy the two things I hate about it, please let me know in the comments.