Sir Tim B . . . . .
Initial update in the interim:
I now see that you need to go thru Sir Warrens pre-emptive checks, operations and signal enhancements, in case of a simple deficiency in those areas.
GRRRRREEEATTTTT . . as coming out of the throat of Tony The Tiger, as your last statement just told us that there is no deficiency in the power supply - the AF amplifier section or the speaker, as all are being shared by the AM section.
The next easist thing would for your unit to be utilizing a FM ratio detector in its design, where there is a 1-5 ufd electrolytic capacitor typically used in that FM detector curcuitry. In passing of time, a related deterioration of that units capacitance down into the .5---.1---ufd values, makes its detector output diminish appreciably, thus the accounting of a low volume on FM.
How-some-ever YOUR unit is using an FM discriminator circuit, so one might lean towards suspecting a quirky 19T8 tube or possibly FM detector alignment . . . . . . Haven't been diddling any of those slugs on the unit already, have you . . . hmmmmm ?
I can throw 3 things at you in an initial check out.
1 . . . Look at my mini thumbnail of the sets circiutry , involving only our points of interest. Locate my BLACK box markup of test point [A] which is at the 1st grid of the sets V6 / 6BH6 limiter tube .
If you place DVM metering there expect its negative voltage level to increase as an FM station is tuned in to . Be sure that you have a 5 ' or so length of wire connected to the FM antenna input to assure as strong of a signal level as is possible. Then you tune across the band and observe the voltage at [A] increase when peaked on stations. Choose the station that comes in the strongest . . .e.g. . . .with the greatest negative voltage.
Peak the tuning of that station with micro adjustment of the tuning knob..
2 . . .Move metering over to the [C] test point as still referencing against chassis ground. Is that voltage pretty close to zero volts
or at least down into in the 10's . . . to worst case . . .100's of millivolts.
Ideally it would be very close to zero. But to now further evaluate, take note of the voltage and determine if it is a negative or positive polarity of voltage .
3 . . .Then you ever so subtly, start moving the station tuning knob in the direction that causes that voltage level to rise . ever so slowly inch up the reading, taking note of its value, at one point it should peak and start diminishing in value.
Take a note of that peak value.
Now you reverse your tuning direction and work back towards the direction that you started from, meeting those familiar old voltage readings again , and saying a hearty HI ! to each one.
When you have met that initial level where you started from, you should find the voltage dropping ever lower and then thru the ZERO threshold and swinging into the opposite polarity of voltage.
Keep up the same procedure as before and be watching for that peak and then a diminisment of voltage as you then pass THIS
peak. Then you stop and take note of THIS voltage peak level. If there is some degree of normalacy in the units alignment, those peak voltage readings should be in the ball park of each other .
Thus, indicating a normalacy in the FM alignment and its overall response.
Example:
+4.5 and -4.5 . . .dat be gooooood
+4.6 and -4.4 . . . maybe
+3.4 and -5.6 . . .dat do be bad
+2 and -7 . . . . . outrageous
If they are waaaay off, we might then consider a bad 19T8, before touching anything.
If you don't have another 19T8 and are a modern electonics person you should certainly be familiar with a common fast switching diode that operates well at higher frequencies. Commonly assigned a number of 1N4148 or 1N914. Readily available from Radio Shack or scavenged from home electronics chassis.
For evaluation purposes of the sets 19T8, one could sub in two of those diodes, tacking across the bottom tube pin terminals.
One diodes anode goes to tube pin #1 , the other diodes anode goes to tube pin #2, the two diode cathodes (bands) join and go to
tube pin #3.
See if the units audio level then pops up, or go thru the same evaluation procedure as before to see if the figures are different.
Standing by for an update . . .
Schematic Thumbnail:

73's de Edd