Published by PitchChanger · Updated · How these guides are made
How to use this workflow
Choose boundaries around the music
Listen for a quiet transition and avoid cutting through a word, drum attack or held note when possible. Save the full original. Record each section’s exact start and end in source time before exporting. Use Audio Trimmer for preparation and Stem Splitter for each eligible section.
An eight-minute boundary map
For a hypothetical 480-second source, the following sections are each under 360 seconds. The shared interval is source seconds 230 through 250: twenty seconds of overlap. Section B’s local time 10 seconds corresponds to source time 240 seconds. These are planned offsets, not measured model outputs.
Swipe or scroll sideways to read all columns.
| Section | Source interval | Length | Assembly offset |
|---|---|---|---|
| A | 0–250 s | 250 s | 0 s |
| B | 230–480 s | 250 s | 230 s |
Reassemble by source position
Place each section at its recorded source offset in an external editor. Align every role from a section together. Compare the overlap while listening to the source; choose one transition location rather than blindly adding both full overlaps. Two copies of shared content can make the join louder. Independent estimates may also differ around the boundary.
Check the complete passage
Listen before, across and after the join for lost attacks, doubled events and abrupt ambience changes. Keep a source-to-local-time map with the deliverable. If the join remains distracting, choose a different boundary or use a tool that supports the full recording. Do not claim that splitting removes the model’s quality or resource limitations.
Common questions
Should I add both complete overlapping sections together?
No. That duplicates shared material. Use the overlap to choose and verify a transition in an external editor.