It is officially implemented: current development builds of Z-Maestro now do all of their mixing using 32-bit floating point arithmetic. What does this mean and how does it differ from the previous mixing engine?

Well, up until now, Z-Maestro has always done all of its audio mixing and editing using 16-bit signed integers. For output, this isn’t a huge problem. After all, CDs are all in 16-bit 44100 Hz stereo. Few people can actually claim to hear a difference between 16-bit and, say, 24-bit output. However, there are a lot of reasons to move to higher bitrates.

The biggest reason is that in the mixing stage, there is often a lot of math that goes on. This includes divisions, multiplications, and additions. With 16-bit integers, all of this up and down arithmetic can actually degrade the quality of the output a tiny bit. With particularly complicated audio effects, it can actually become noticeable. In order to combat this in the past, most audio effects up-converted their inputs to 32-bit float, did the math, and then down-converted the signal back to 16-bit integers.

Of course, doing all of this conversion is a little expensive performance-wise, so it makes sense to eliminate it. Another reason to switch was that we wanted Z-Maestro to be able to support 24-bit output. Modern computers don’t work well with raw 24-bit data, but they do work with 32-bit data.

Lastly, editing can also introduce artifacts when using 16-bit data. Now that we have switched to 32-bit floats, the audio editor will be able to perform much more accurate edits without much possibility of aliasing artifacts.