shabaz , the current source in the tail of the differential amp (orange area below):

what was your inspiration to design that sub-circuit that way?

From memory I originally didn't have any of that, and used a resistor, but needed beyond a certain value to then be able to have lower gain and low distortion (since this stage wasn't intended to have high gain; I would have been happy with just 6 dB or so) but couldn't find a combination that would also give me a good signal input range (I needed at least several hundred mV to suit the circuitry preceeding it).

Eventually I gave up and tried the BJT current sink, there may well be more optimal biasing that could have eliminated the need for the BJT sink at all, I just couldn't find it through trial-and-error (I'm not an analog designer otherwise I may have known a better technique). (By trial-and-error I mean using the KiCad simulator).

Once I went down that road although I took some time on it, I believe better results could be achieved. I likely just started with one transistor and then for some reason switched to the darlington pair, but there are better current sink topologies. For the combination of relatively large input (few hundred mV p-p), what seemed like good common signal rejection, low desired gain and low distortion in the speech bandwidth, the tweaks had reached a point where the simulator results suggested there was no need to improve further.

I was finding it took ages to find something optimal since there are two inputs and I'd have to keep changing the signals to check it was still working fine across these requirements.

"...since there are two inputs and I'd have to keep changing the signals..."

Something that's handy for these kind of situations are the controlled sources in spice. Here I've got a voltage-controlled voltage source on the second input controlled by the generator on the first. If I set the controlled source gain to -1, then the second input is simply the inverse of whatever the generator is set to, and if I set the gain to 1, both inputs are the same and the second tracks the generator.

First attempt to simulate the current source / sink:

First attempt to simulate the current source / sink:

In LTSpice, with NPX BC547C model:

... and the zoom:

I'd do it like this with Tina-TI. Here I've asked for a DC Transfer Analysis where it sweeps a generator between two voltages and plots the result. I imagine you'll be able to do something equivalent, because it's almost certain to be fundamental to the spice simulator.

The graph then shows the compliance of the source. Not very useful below 2V, but fine above, with a reasonably flat top.

Below 0.5V the current goes negative, because there's a path back through the base-collector diode from the potential divider, but you can ignore that.

This is with the vertical scale magnified so that you can see the slope better.

Keep in mind that that's coming from simplified transistor models in a simulator - so the real slope may be different, and may not even be a precise straight line like that.