The Paulstretch algorithm is a procedure for achieving aurally pleasing extreme time-stretches while avoiding the “phasiness” issues common to phase-vocoder-based implementations. The algorithm accomplishes this by randomizing the phase information where a typical phase-vocoder approach would perform phase-unwrapping or another method to preserve phase alignment. When performed without time-stretching, phase randomization is often referred to as whisperization. This allows for more extreme time-stretches than other approaches. Where most audio time-stretching is usually used for small adjustments, Paulstretch is regularly used to stretch audio by a factor of 5, 10, 20 or much more. This effect is popular for its aesthetic timbral effects and is regularly used in soundscapes, film/television scoring, and more. When time-stretching is performed at such extremes, however, it is unclear how randomizing the phase, different FFT window sizes, the stretch factor, and other variables combine to affect the reconstructed audio output. This study employs multiple audio analysis methods to examine the spectral, timbral, and perceptual effects of the Paulstretch algorithm.