Mostly power handling. It maxes out at about 94 dB already.
No, it's thermal handling that is the issue here. It has plenty of Xmax to handle > 250 Hz, just not the power handling to match. Now music really has a fairly decent crest factor, so that will give some extra headroom, but it's still a bit tight.
This is excursion with 50W applied:
View attachment 182560
Yes, but...
1) Voices don't hit the same peaks that explosions do. So you're already... what, 10-15 dB down? Remember, 10 dB is roughly what it takes for an apparent doubling (or halving) of loudness. So that's 10% of your power relative to peak requirements, right there.
2) The bandwidth is limited. With a driver like that, maybe you're running 600 Hz - 2500 Hz or so. If I plug that into a formula I have on one of my design sheets (based on an old chart from Fane showing distribution of power vs frequency), it tells me I need 23% of the power relative to the whole speaker, when playing typical music (vs taking an overall speaker SPL reading with a meter). With voices, way more of the power will be concentrated in the midrange - but it's still not doing fullrange duty. Even conservatively, I'd say the mid would be taking 50% of the overall sound output at most. Of course, this will go up as the HP frequency goes down. At 300 Hz I'd use a factor of 75% instead.
3) I'm not including this factor, but short-term peak power handling will be more than the rated 20 W. IIRC, IEC 268-5 uses a signal with 6 dB crest factor, you're not going to be listening to a continuous 97 dB @ 1m, are you?
Basic, stupid math - 0.1 * 0.5 = 0.05 => driver power is 5% of the power you'd need for a full sine wave at your maximum SPL.
Uh, I probably could explain that better. But I hope that makes sense in some way. Long story short - I think you're fine for total output, but again, DON'T expect to do full-range sine sweeps and get clean output @ 105 dB all the way. It's not your use case.