7 April 2026 | 6 min read
Sample rate conversion, explained without the maths
Resampling is not a setting, it is a calculation. Doing it twice, or at the wrong moment, is how audio drifts out of sync.
Sample rate is how many times per second the recorder measured the signal. 48 kHz means 48,000 measurements. Changing that number means recalculating every sample from scratch, which is why the quality of a resampler matters and why you should do it as rarely as possible.
The rates you meet in the field
- 48 kHz: the film, TV and broadcast standard. Default for everything on set.
- 44.1 kHz: the CD and music streaming standard.
- 96 kHz: high resolution recording, useful for heavy pitch manipulation in sound design.
- 32 kHz and below: legacy formats, phone codecs and voice memos.
Why 48 to 44.1 is the awkward one
48 and 96 are related by a simple factor of two, so conversion between them is clean. 48 to 44.1 is a ratio of 147 to 160, which forces the resampler to interpolate values that never existed. A good algorithm handles it transparently; a crude one adds aliasing that sounds like a faint high frequency grit.
The sync rule
Convert once, at a defined point in the workflow, and keep everything downstream at that rate. Mixed rates on a timeline are the classic cause of audio that starts in sync and drifts by the end of the reel, particularly when a NLE resamples on the fly instead of conforming the file.
Higher is not automatically better
Upsampling a 44.1 kHz file to 96 kHz adds no detail. It only adds file size. Resample upward only when a delivery spec demands it, and never as an attempt to improve a recording.