Umm, if you disconnect and then reconnect the external speaker do you get a loud "POP" from the speaker when the plug makes contact? Is there a tiny spark where the speaker plug makes contact with the jack when they touch?
Sometimes if the cord on an external speaker shorts, it will overload the amplifier chip. What you're hoping for is a sign that ONLY the capacitor between the audio chip and the speaker was damage. A 470 uF electrolytic cap is a lot cheaper than the TDA1905 audio chip. And a lot easier to change, if it shorts.
C95 is there to couple the speaker's AC audio signal from the output on pin 1. Problem is, that pin 1 on the TDA1905 has a steady DC voltage, around 6 or 7 Volts DC. The audio signal we want rides "piggyback" on this DC voltage. C95 isolates the DC dead short that the speaker's winding represents. This way, only the audio portion reaches the speaker coil, NOT the DC voltage. If C95 gets overloaded by a damaged external speaker cord, it will short inside, and permit the full DC voltage from the chip to reach the speaker coil.
Might fry the speaker. What is supposed to happen is that the overload sensing circuit inside the chip shuts it down before the overload can damage it. Makes the audio fade out in a few seconds or more.
Of course, this may have not ONE thing to do with your actual problem. Just sounds like one we've seen before.
A quick check to see if the chip is still okay is to take one speaker wire loose, and insert a new 470 uF cap between the wire and the speaker. Be sure to check the polarity before you do. You don't want to 'patch' this new cap into the circuit with the polarity reversed.
If it receives okay with the "test" capacitor clipped in line, odds are that replacing C95 will fix the problem.
If the TDA1905 is blown, the audio won't sound any better with the 'test' cap in line.
Even so, the 1905 may not be blown. There are five more electrolytic caps connected to that chip. Each one of them is statistically more likely to fail than that chip. All audio power chips like the 1905 CLAIM to contain overload protection. The overload sense/protect in that chip is better than most. The caps are more failure-prone than it is. Don't see a lot of them break down.
Or maybe the trouble is nowhere near C95.
73