Yes, 64-bit float exists
64-bit float is the double-precision format defined in IEEE 754. Each number has a sign bit, 11 exponent bits and 52 stored fraction bits, giving about 53 bits of precision. Single-precision 32-bit float gives 24 bits of precision. Double precision gives roughly 15 to 17 significant decimal digits, compared with about 6 to 9 for single precision.
In audio, that extra precision is for calculation. It isn’t for storing a recording.
Where 64-bit float is used in audio
DAWs do their mixing with floating-point math. Each addition or multiplication can introduce a very small rounding error. Summing many tracks, or running long plugin chains, lets those errors build up. A 64-bit accumulator keeps them far smaller.
Some DAWs offer 64-bit float processing or 64-bit summing as an option. Reaper processes internally at 64-bit. Cubase, Nuendo and Studio One offer it as an option in some versions. This changes between versions, so check your DAW’s current settings.
Can you record or deliver in 64-bit float?
Mostly no.
- No microphone preamp or analog-to-digital converter records 64-bit float. Portable field recorders top out at 32-bit float.
- Some DAWs can render a mix to a 64-bit float WAV, but many media players and hosts can’t open those files.
- Distributors generally want 16-bit or 24-bit WAV. Export at 24-bit or 32-bit float for mastering.
Is it worth using?
For most mixes, the audible benefit over 32-bit float processing is negligible. Rounding error at 32-bit float sits far below the noise floor of typical recordings. The cost is small: 64-bit processing can use a little more CPU and memory, depending on your DAW and plugins. If your DAW offers it, test with your own project and watch the CPU meter.
What to record at instead
Record at 24-bit, and use 32-bit float when levels are hard to predict. Export mixes at 24-bit or 32-bit float for mastering. For the full comparison, see bit depth explained.
Not the same as a 64-bit processor
64-bit float describes how numbers are stored and calculated. A 64-bit processor describes the width of its registers and memory addresses. The two are unrelated. For what a 64-bit processor does, see what is a 64-bit processor?