Which circuits to use?
The Cathode Follower offers low output impedances, and no gain. The Plate Follower has a low output impedance and inverts the signal phase, but is awkward to use because of its fairly low input impedance, which will shunt the volume potentiometer resistance. The Plate Follower is a good bit more complex than a simple Cathode Follower and will not allow for the noise canceling trick that this circuit uses.  (If a dual polarity power supply were not available, then this circuit, along with a Cathode Follower, would meet our design objectives. )

Of the two commonly used phase splitters, the Long Tail and the Split-Load, only the Split-Load offers no voltage gain. In addition, this phase splitter has the added advantage of an extremely poor PSRR figure, which, paradoxically enough, in this circuit results in lower noise at the output.   

Split-Load phase splitters are always a bad choice--aren't they?
This circuit suffers from an undeserved bad press. The Split Load Phase Splitter is a much better circuit than most imagine. Its reputation suffers from decades of misuse, but if its operation is understood fully, one realizes that it is a useful circuit. It is criticized for not having any gain, clipping sooner than alternative phase splitters, having dissimilar output impedances on each phase leg, and for exhibiting a very poor PSRR at its plate output.

Not having any gain is a virtue in our no-gain line stage and our line stage will never be called on to produce any more than a few volts output, which this phase splitter can handle easily. As for the problem of dissimilar output impedances, it does not mater in our unbalanced design.

(In fact, the truth is that as long as each phase leg of this circuit is equally loaded both resistively and capacitively, the discrepancy in output impedances does not in practice make that much difference. This paradox can be resolved when we compare what happens in two cases. In the first case we load the positive phase leg [the cathode output] with increased capacitance--the circuit now begins to resemble a Grounded Cathode Amplifier with a bypass capacitor across its cathode resistor, i.e. its high frequency response at its plate is increased. In the second case, we load the inverting leg [the plate output] with a increased capacitance--the circuit now begins to resemble a Cathode Follower, i.e. its high frequency response at its cathode is increased, as the capacitance across its plate fixes its plate voltage at high frequencies. A wash of sorts occurs where both phase legs share the same frequency response with a balanced capacitance loading of cathode and plate outputs. Does this mean that the circuit can work into any low impedance and/or high capacitance load just because both phase legs are equally loaded? Of course not. Capacitance must be charged with current to force the voltage swings we need, which means that if the current is sufficient to charge the equal load capacitance at the slew rate needed to insure a certain bandwidth at a certain peak output voltage, this circuit should work just fine.)

As for the poor PSRR, we will see in the "Trick" section how its failure to reject noise works in our favor.

Cathode Followers sound bad, don't they?
"Not necessarily," is the short answer. Once again, a Cathode Follower need not sound bad; and, if properly designed, it can sound as good as any Grounded Cathode amplifier. A quick re-read of September's Circuit of the Month provides more detailed information on the role of transconductance and current in determining the sound quality of a Cathode Follower.