Demo video

Notes: 100nF (first example) and 10nF (second example w/ speaker) capacitors are used. Unfortunately, there is also one major flaw in this first OP set-up (100 Ohm resistor instead of 1 kOhm resistor, i.e. 1.1 kOhm and not 2 kOhm for R2).

Schematic

As with any oscillator, no input is needed. It swings by itself. 

A 10 nF capacitor would lead to a tenfold increase of the frequency. 

Functional description

The first OP stage is an integrator (note: negative feedback loop), the second a non-inverting Schmitt trigger (note: positive feedback loop).

The wave is triangle shaped on the integrator output (IC pin 1) because the capacitor is forced to charge linearly: the current through resistor R1 is constant and the input of the OP Amp draws (virtually) no current.

The gradient x (V/ms) is calculated by the following formula (U_s would be 9V here):  

R2 and R3 determine the voltage thresholds (therefore the oscillator frequency) of the Schmitt trigger.

No reset switch for the integrator is needed, for it is part of a larger device i.e. oscillator.

Power Supply

There has to be a plus and a minus voltage (ideally symmetrical) to supply the whole thing. This isn't always readily available (though you can just use two 9V battery blocks). 

But one can built up a power supply with an OP Amp too. Below you see a quick&dirty set-up with a UA741CN OP Amp (active voltage divider) and a couple of transistors that I scrapped from electronic waste (this design has some problems, therefore I don't share the schematic).

An unsymmetrical supply voltage will lead to an unsymmetrical triangle-wave signal; the oscillator won't swing at all if it is too skewed.

Application

Not yet decided for this project. The oscillator could be part of a sound generator/synthesizer.

A triangle signal generates a very distinctive, not too unpleasant sound (triangle signals consists solely of odd harmonics: 1st, 3rd, 5th, etc.).