Time-stretching and pitch-shifting are decoupled operations requiring phase vocoder or WSOLA rather than sample rate change

In raw sample playback, changing playback rate simultaneously changes both pitch and duration. To decouple these, specialised algorithms are required. The two main approaches are: (1) Phase Vocoder, which analyses with STFT, modifies the rate of phase advance per bin (for time-stretching) or shifts frequency bins (for pitch-shifting), then resynthesises with overlap-add. (2) WSOLA (Waveform Similarity-based Overlap-Add), a time-domain method that overlaps-and-adds segments, selecting splice points to minimise discontinuities. Artifacts include phasiness (phase vocoder) and robotic quality. All modern DAWs use one of these approaches.

Examples

Ableton Live Complex Pro warp mode uses phase vocoder. Beats warp mode uses segment-based WSOLA. Strudel’s slow and fast functions affect only time; pitch changes use the note function.

Assessment

Why does simply slowing down audio playback change both pitch and duration? What does a phase vocoder modify in the STFT output to achieve time-stretching without pitch change?