Posted on Leave a comment

Under the hood: Polar vs Complex Averaging

For transfer function measurements, there are two commonly referred-to options for temporal averaging of magnitude data: Polar and Complex.

Polar averaging calculates the decibel magnitudes of each frame and then takes a moving average of the result. It is occasionally called decibel averaging for this reason. Polar averaging tends to be the more stable and forgiving of the two, especially when wind, air currents, mechanical movement, or similar noise is present. This type of averaging lets more reverberant energy into the average, which may better represent what you hear, particularly for musical program material. 

Complex averaging, on the other hand, keeps two separate running averages of the real and imaginary parts of the complex signal and then calculates magnitude and phase from these averages on the back end. It can provide better overall noise rejection and tends to exclude more reverberant energy than polar averaging does. This may give you better clues regarding speech intelligibility than polar averaging.

In subjective terms, polar averaging may be the more “musical” of the two options, as it tends to let in more reverberant energy. Complex averaging, which shuts out reverberant energy, correlates better with subjective speech intelligibility.

Alternatively, polar averaging tends to look more like what the system sounds like, wheres complex averaging tends to show you a better representation of what the system is actually doing in the environment by excluding more late arriving energy. This option, while set globally in Smaart by default, can be set separately for each transfer function measurement engine.

You can set set the global averaging type in the Measurement Settings menu, which is Polar by default. This only controls the averaging of magnitude traces, however, as phase traces will always use complex averaging.

In Smaart v9 and newer, all data sets required for both averaging types are saved when a measurement is captured. This means that even after capture you can select the data set that best suits your needs from the Trace Info window. In Smaart v8.5 and older, averaging type selection is only available for live measurement data and trace info displays the averaging type used at the time of capture.

Smaart is excellent on its own, but most users find they get much more out of it after some structured training. That’s where our seminars come in. At TZ Audio we run practical seminars, both online and in-venue. We offer seminar-only or full “all you need packages” including software & hardware. It’s simply the fastest way to become comfortable and confident with the measuring a sound system.

If you’re in Norway, Sweden, Denmark or Iceland – or elsewhere – we offer is online seminars and traveling to Norway is a valid option too of course. We’re here if you have any questions about the software or upcoming seminars.

Thanks for reading!

Leave a Reply