The AC gain is 1+(R4/R5). That holds for low to medium values of gain, after that you have to consider the non-infinite open-loop gain, naturally.
That "1+" is quite important - with R4 replaced with a short, you have a unity-gain follower - aka "CFP" (
compound/complementary feedback/follower pair). Commonly done when you want something a bit better than a standard emitter follower. The configuration was used in the BBC LS5/8 crossover, for example. You also see this configuration used for audio output stages. Most often with unity gain, but you do sometimes see examples where there is modest gain (perhaps x3 or x4), which usefully enables the designer to run the front end of the power amp at lower supply voltages.
Another place where that configuration works well is video amplifiers, where you often need a gain of x2 - in which case you just make R4 and R5 equal. There's more than enough bandwidth for that application - it's surprisingly "fast" for such a simple circuit. This is something to keep in mind when experimenting - speed can bring problems.
Given that most ceramic cartridges are high output - in the ballpark of 0.5V, perhaps - I'd suspect that even x10 (20dB) is more than required.
I'm not convinced a JFET is needed here. Using a BJT instead will still result in a high input impedance. For audio circuits, I much prefer BJTs because of their higher gm and more predictable results, and mostly only use them for controlling signals (e.g. power-on mute). In really simple terms, if we assume a follower circuit made from a pair of BJTs with a hfe of 200 each, that's a combined current gain for 40,000. If there is a 1k load resistor to ground at the output of the follower, the input impedance should be 40,000 times 1k, which is 40G! Obviously, the reality won't be anything like as simple, but either way, you'll comfortably get into the low M-Ohms region.
Having said all that, some ceramic cartridges need a lower impedance load for equalisation. Syncrodyne has written a lot about this on this forum and others - for example, see his posts in this thread:
https://www.vintage-radio.net/forum/...ad.php?t=60957
For the sake of balance, I admit the JFET does simplify the biasing slightly, even if the results vary according to Vgs. If using a BJT at the input, you'd need another resistor from the rail, and that means the rail needs to be somewhat cleaner than is required as it stands. But against that, Vgs is quite variable...
To conveniently separate the DC and DC conditions, I would change the arrangement around the output side. As it stands, the load for the follower is the sum of R4 and R5 (and R6 for DC). Instead, drop a resistor from the collector of TR2 to ground - choose a value that gives a nominal 1-2mA in the collector (so for a 10V rail and 5V at TR2 "output", pick anything between 2k7 and 4k7, perhaps). Having done that, omit R6, but retain R5 and C2 to ground, and R4. R4 and R5 now set the gain, but don't alter the DC conditions, because the DC gain is unity (1+R4/infinity). Keep R4 the same - as its value sets the current in TR1 - and alter R5 as needed.