Acoustic white noise is useful as a masking sound to improve privacy in offices, or as a soothing sound field to help concentration and sleep. It has also been said to aid in therapy for people suffering from tinnitus (ringing in the ears). White noise is a signal containing components at all frequencies with equal amplitude. The circuit shown in the figure generates pink noise (which is white noise rolled off) sufficient to drive a pair of small speakers or earphones.
It's well known that an avalanched emitter-base junction of a junction transistor is a good noise source. In this circuit, Q1 is biased to its emitter-base junction avalanche voltage by inverter Q2, and in doing so, generates a millivolt or two of white noise. A2 and A3 provide 50× gain each, for a total of 2500×, producing volts of noise from the junction. A4 serves as a unity-gain inverting amplifier. When the headphones or speakers are connected between the output of A3 and A4 to form a full bridge output, voltage gain is doubled and the maximum drive voltage approaches 14 V p-p. A1 buffers a 4.5-V reference made from the two 470-Ω resistors. The base of Q1 isn't connected.
The full-bridge op-amp circuit can provide plenty of drive voltage for an efficient pair of speakers, but not much current without additional buffers. So if it's used to drive a pair of speakers or earphones, you'll get the greatest power output by driving them in series rather than parallel.
The noise spectrum can easily be shaped to compensate for the characteristics of a particular set of speakers or earphones by controlling the bandwidth of the amplifiers following the noise source. For example, placing a small capacitor (0.001 µF for a start) across one or both of the 20-kΩ resistors will result in a high-frequency boost of the amplitude of high-frequency components. Placing a small capacitor across the 1-MΩ resistor (47 pf for a start) will cause a decrease in the amplitude of high-frequency components.