Here is something interesting, which when you think about it in electronic analysis terms, it only makes sense.
I decided to build a class AB amp, not a computer simulation, and analysed it with a signal generator, and oscilloscope.
The Av was around 8 with a 50mV input to a 400mV output unloaded, at 1 Khz. Then loaded it with an 8 ohm speaker, and it dropped to about half, so now the amplitude is around 200mV Av of around 4.
No surprise, I was expecting some loss in Av.

Then I started playing around with the frequency, and as I decreased it to around 182 Hz. the waveform doubled in amplitude, and speaker signal got louder. Av was back at 8 again.

As I further decreased the frequency, the Av dropped back to 4, so I went above 182 hz. and again the gain dropped to around 4 again. The Av peaked at around 182 Hz, well that can only mean one thing, a resonant frequency, so I put a resistive load in place of the speaker, and sure enough the Av. remained at around 4 up and down throughout the frequency range using the 182Hz as base frequency, as frequency got much lower the gain dropped due to Xc of the input and output coupling capacitors.

So in conclusion, the speaker being it is a reactive device, (XL) was showing its own resonant frequency on the scope, something I never thought about until I built this circuit, and was running it through its tests.
.

 

Yes, speakers become noticeably reactive at around their resonant frequency with consequent change in their impedance.
Typical well-designed audio amps have such a low output impedance that this change in speaker impedance does not cause any significant change in amplifier output voltage with frequency.