Signficance of duty cycle on PWM output

Using an oscilloscope and looking at the PWM output, I now understand what the duty cycle is for a PWM signal but I’m curious if anyone knows about the effect or usage of it on audio signals. For example two 2200Hz signals, one generated with 50% duty and the other 25% duty will both be 2200Hz. But the 25% one will have half the width on an oscilloscope. Why would I prefer one over the other for an audio signal? I’m using PWM to generate tones that are modulated through a radio and being decoded on the other side, I’ve been using mostly 50% so far, not knowing what else to use and it seems to work most of the time except I have problems at higher transmission rates - just trying to rule out this duty cycle of the PWM signal as a potential issue.

Since you are going to convert the square wave into sin wave to transmitt it on your radio (if you don’t, the first serial capacitor included in your modulator will do it for you…), you shall not change the duty cycle from 50% or it will result in a lot of unpredictable harmonics…

With a 50% duty cycle you will have only odd harmonics
With anything else, you will have a mix of even and odd harmocis, the even ones beeing closer (and strong enough) to the main peak and more difficult to filter out.

See sample spectrums here:

Great, thanks!