Sound power is the average of the radiated power from the loudspeaker in all directions. The most common method is to take measurements in a spherical surface all around the loudspeaker, integrate the measurements over the surface of the sphere, and then divide by surface area of the sphere to obtain the average power.
For ANSI/CTA 2034-A (aka spinorama), Frequency responses (FR) are measured at the 70 points in the vertical and horizontal orbits in 10 degrees angular spacing (i.e.
32 36 points for each orbit, with 2 shared common points = total of 70 FR measurements). For each frequency, each of the 70 measurements is weighted according to the portion of the spherical surface it covers, and the 70 weighted measurements are summed to give the sound power at that frequency.
The process is repeated for all frequencies. For FR measurements with 1/20 octave frequency resolution, the total number of frequencies is about 200 (10 octaves at 20 points per octave).
The weighting values are given in Appendix C of ANSI/CTA 2034-A. Note that we can't directly sum dB SPL values. They must be converted to sound pressure (unit: N/m^2) first, obtain the weighted sum, and then reconvert back to dB SPL for reporting.
You can download the CTA 2034 standard at no cost at this link (will need an email address for registration).
This Standard describes how to determine the frequency response, directivity and maximum output capability of a residential loudspeaker.
shop.cta.tech
Unless the loudspeaker is perfectly omnidirectional, the sound power will be different from the on-axis response. Since practically all loudspeakers are more directive at higher frequencies, their on-axis response will be higher than their sound power. The difference between the on-axis response and sound power is the sound power directivity index (SPDI). For example, when the SPDI is, say, 4 dB at 2 kHz, that means at 2 kHz the on-axis SPL is 4 dB higher than the sound power SPL.
[Edit] Correct errors.