R5, R14 and R2 are trimmers.
This can be contained in a single package (TL074).
I must admit that the triangle waveform isn’t pretty on the scope. You can cleary see a glitch where the diodes switch on and off. But could it be audible when filtered ?
A bit of gain following the triangle (diode loss) to bring the triangle to the usual 10VP-P is needed, as well as for bias the triangle corectly around 0V. Obviously, the stage isn’t on the schematic, silly me.