Splitting a long recording into pieces is a boundary problem: where the cuts go, and what happens to the remainder. Both modes here answer it explicitly.
Equal parts
Divide the file into a set number of pieces of the same length. A 48 minute recording in 4 parts gives four 12 minute files.
The last piece absorbs any rounding, so the pieces always add back up to the source exactly.
Fixed length
Cut every N seconds. A 10 minute file at 90 seconds gives six full pieces and a 60 second remainder.
This is the mode for a hard limit: a messaging app that caps uploads at 5 minutes, or a device that reads files of a fixed size.
Boundaries on the waveform
Each planned piece is drawn on the waveform before anything is written, so the cut positions are visible against the material. If a boundary lands in the middle of a word or a beat, change the count or the length until it does not.
Output
Every piece is written in the source format at a bitrate matched to the source, and numbered in order: recording-01.mp3, recording-02.mp3, and so on. They arrive as one ZIP.
Re-encoding
Splitting a lossy file re-encodes it. The pieces are one generation below the source, which is inaudible at a normal bitrate but real.
Splitting a WAV or FLAC is lossless, so use those when the pieces will be edited further.