live-coding-academy_ ⌕ search
home/ atoms/ realtime-vs-offline-dsp

Realtime DSP processes audio as it is produced; offline DSP calculates ahead of playback

Concept L1Foundations B · Sound design & synthesis

Realtime synthesis/processing happens with only milliseconds of latency between computation and listening — the CPU must finish each audio buffer before the next is needed. Offline (non-realtime) processing writes results to disk first, then plays them back. Offline is necessary when: (a) algorithms are too complex for the CPU to finish in time; (b) the algorithm needs the whole signal before it can act (e.g., sorting notes by pitch). Offline can also be faster than realtime when CPU throughput exceeds audio duration. Neither mode is universally superior — realtime enables live interaction; offline enables arbitrarily complex processing.

Examples

Realtime: playing a filtered sine oscillator live in Max. Offline: applying a convolution reverb that takes 3× real time to compute, saved to a .wav file for later playback.

Assessment

Give one scenario where offline processing is conceptually required regardless of CPU speed, and one where it is required only because of CPU limits.

Prerequisites — understand these first

Sources

“When the stream of data goes directly to the audio interface as it is processed, so that there are only few milliseconds between the processing and the listen- ing of the synthesized sound, one speaks of realtime synthesis.”
corpus · electronic-music-and-sound-design-vol-1-cipriani-and-giri-of · chunk 6

Taught in modules

Unlocks (3) — atoms that build on this