High-quality random number sequences are indispensable in scientific research and industrial applications, from cryptography to Monte Carlo simulations. For experimental utility, the underlying RNG must not only meet stringent statistical benchmarks but also offer straightforward integration with existing hardware and software architectures.
The importance of both parameters --- quality of randomness and integrability --- often surpasses that of the generation rate. Our work presents an analysis of quasiclassical and quantum optical random number generators, considering cases with and without postprocessing through randomness extraction. The use of heavily attenuated laser light allows for rapid sequence generation, though the resulting bit streams demonstrate low-quality randomness. In contrast, a quantum source based on spontaneous parametric down-conversion produces highly random sequences, but at a much slower rate. To enhance overall performance, a hybrid approach combining light from the two sources is also examined. Additionally, the effect of randomness extractors using von Neumann and V.F. Babkin methods was evaluated for sequences for each source.