That design from Hood looks a little off, to me. Partly because I think the 2N3906 may run out of gas given certain drive cases. It also seems to be running very hot (a bit high on the quiescent current side of things) than is strictly necessary. But these are just quick mental sanity checks and I can't be sure without building one and/or running Spice on it. And I won't bother because I'd rather just offer something that I have tested.
As far as using the 2SC5200, I don't imagine problems in the following design I'll offer, as I've included a Miller cap to roll off circuit response. Using them below should be just fine. Just keep in mind that these devices are in a class-A amplifier and will be dissipating maybe \$15\:\text{W}\$ each. So give them some serious heat-sinking capacity.
Here's what I'd suggest trying:
I've bootstrapped the \$8\:\Omega\$ speaker as this turns \$R_1\$ into a semi-decent (and very cheaply used) current source. (If you understand that \$C_{_\text{BOOT}}\$ will acquire a relatively stable voltage difference across it and accept that the \$V_{_\text{BE}}\$ of \$Q_1\$ is also relatively stable, you can see that the voltage across \$R_1\$ should be similarly relatively stable.)
\$R_1\$ will set the quiescent state of \$Q_1\$ and \$Q_2\$. If you increase its value, those two BJTs will run cooler. But depending on the voltage gain and signal input, that also could mean running out of drive current for the speaker and the resulting distortion. Lowering its value will run things hotter and provide more steam needed to avoid distortion. It's your call here. But feel free to make small adjustments to \$R_1\$, one way or another, per your needs. I've set \$R_1\$ to run things a little on the cooler end. But if you are driving the speaker hard, you may need to lower its value a little. Just FYI.
The voltage gain can also be changed by adjusting \$R_4\$. (I've get the voltage gain to about \$16\times\$ or about \$24\:\text{dB}\$ as \$R_4\$ and \$R_3\$ form up a voltage divider that reduces the output signal by that much.)
Per the instructions you see for Dr. Hood's design, you should adjust \$R_6\$ in the above schematic to get the collector node shared by \$Q_1\$ and \$Q_2\$ to about half the supplied \$V_{_\text{CC}}\$. This allows near the maximum output swing without running against the rails.