What is 32-bit floating point?A better way to encode digital audio

Encoding is always part of digital audio. The lossless format encodes the entire recording. However, if the treble of the original recording is clipped, or if the gain is too low to pick up the quiet sound, you can’t encode anything that isn’t there.

But what if the recording level wasn’t important? What if you can forget about the problems caused by automatic gain control (AGC)? What if you start with a soft sound like the flapping of a butterfly wing, end with a jet engine afterburner scream, and you don’t need to adjust the gain? Catch details in the full range of all sounds?

Now you can do so. Think of it as a high dynamic range for your ears.

32-bit floating point

Audio is encoded into files at different bit depths. The fewer bits there are, the less digital audio can accurately represent analog sound waves, and the more digital noise is generated.

16-bit audio encodes sound to one of 65,536 levels. The maximum number is the maximum volume. Its bit depth allows for a dynamic range of approximately 96 decibels.

24-bit audio (the most popular professional format today) supports over 16 million levels of sound. This corresponds to a dynamic range of 144.5 dB. Better, but still well below the maximum sound pressure at sea level of about 210db.

Both 16-bit and 24-bit audio are encoded as fixed-point numbers.New format called 32-bit floating point The audio circle encodes audio in the IEEE-754 standard single precision format. 1 bit for positive or negative. 8-bit exponent; and 23-bit decimal part.

Converted to decibels, you get a range of over 1500 dB. It’s a much wider range than you’ve ever needed.

Moreover, unlike fixed-point encoding, there is no specified maximum sound level. Therefore, there is no loud clipping.

Fully increase the gain of 32-bit floating point recorders! The sound level can later be adjusted on a 32-bit digital audio workstation (DAW). Today, most professional DAWs handle 32-bit floats.

As always, there is no free lunch. 32-bit files are about one-third that of 24-bit files. But given today’s multi-terabyte hard drives and SSDs, that’s hardly a problem.


The bigger problem is that few audio recorders support the 32-bit floating point format.But they are there-check Sound device And zoom -And they cost hundreds of dollars. I hope to see more options soon.

Once you get a 32-bit float sound from your audio interface, there is a lot of software that can use it.Free open source Boldness –Available on macOS, Windows, and Linux-Edit 32-bit floats.Do so Protools And Logic Pro..

Here are some caveats. If distortion occurs before recording (mic capsule overload, power line hum), 32-bit floats are useless.

Also, 32-bit floating point digital signal processors (DSPs) are more expensive than fixed point DSPs. When packaged into a high quality audio interface, the difference isn’t that big, but it’s not the cheapest recorder today.


32-bit floats aren’t a panacea for all audio problems, but they eliminate the need for recording engineers to monitor clipping during recording. She can focus on mic selection and placement, room tones, and making sure AGC is off.

Audio is usually the most important and difficult part of a video project. Removing audio level issues from your recordings will make the hard work easier.

comment? If you are using a 32-bit float, let us know what you think.I’m planning an audio project and wondering if I need to upgrade ApogeeOne interface.. What is 32-bit floating point?A better way to encode digital audio

Back to top button