How to split a voiceover recording at silence

A long voiceover can be hard to review when pauses, retakes, and room noise are all in one recording. Splitting it at quiet gaps makes the takes easier to inspect and gives you a cleaner starting point for client edits.

Use the Audio Editor to split an audio recording by silence in your browser. Then listen around each cut before exporting, because a quiet breath or the start of a word can look like a pause to an automatic split.

Split the recording at silence

1. Open the Audio Editor and load your voiceover recording. Keep an untouched copy of the original file in case you want to start over.
2. Listen to the recording once, or scrub through the timeline, and note the pauses that should separate phrases or takes. Decide whether you want separate sections or a single edited voiceover with the gaps removed.
3. Use the editor's split-by-silence function on the recording. If you can select a section first, select only the part you want to process. This helps avoid splitting an intro or outro that you want to keep intact.
4. Play the result from a little before each new cut to a little after it. Check that the last word stays whole and the next phrase does not start abruptly.
5. Make any needed manual cuts, then export the finished audio as WAV or MP3. Use WAV if the client will edit or process the voiceover further. Use MP3 when a smaller, convenient file is more important.

Before you export, confirm the client's delivery request. They may want one continuous voiceover, a set of separate takes, or both. Use a clear file name and keep the original recording separate from the edited version so there is no confusion about which file is ready for review.

Choose where to split

Split at a pause between phrases

A short pause can separate sentences without making the delivery sound unnatural. Leave a little breathing room at the start and end of each phrase. Cutting right against a word can make speech feel rushed, even when the word itself is not clipped.

Split between takes

When a speaker records several versions of a line, a longer quiet gap may make a useful boundary between takes. Keep each take intact if the client needs to compare performances. You can remove the unused takes later, once the client has chosen a version.

Keep breaths that belong in the performance

Not every quiet moment is unwanted. A natural breath between clauses can help the voice sound human and steady. Listen to the phrase as a whole instead of deleting every gap the silence split finds.

Keep room tone when the edit needs it

Room tone is the low, steady background sound in a recording space. A small amount between phrases can make edits less noticeable. If you remove every trace of it, the background may switch abruptly between speech and digital silence.

How to fix uneven or misplaced splits

1. Undo a cut that interrupts speech. If a word or breath has been separated, undo that edit if available, or join the pieces by removing the unwanted boundary. Replay the phrase to make sure it flows naturally.
2. Listen for quiet words before splitting again. A soft opening word, an ending consonant, or a breath may be close to the background level. Zoom in on that section and make a manual cut in a clearly quiet gap instead of relying on the automatic boundary.
3. Check for steady background noise. A fan, computer, or distant traffic can make pauses sound less silent to the editor. If the split function creates too few sections, find the clear gaps by listening and make the remaining cuts by hand.
4. Review short pauses that create extra pieces. A brief pause for emphasis may be part of the speaker's delivery, not a boundary between takes. Keep the related words together, or undo that split and use a longer gap as the cut point.
5. Replay the joins at normal speed. Scrubbing closely helps you inspect a cut, but normal playback reveals whether the pause feels natural. Check a few seconds on both sides, not just the exact cut point.
6. Export a fresh copy after changes. Give the revised file a new name so it does not overwrite the original. Listen to the exported file from beginning to end before sending it to the client.

Why silence detection can miss a good cut

Silence detection looks for low audio levels, not sentence meaning. It cannot know whether a quiet sound is a breath, a soft consonant, a pause for emphasis, or the beginning of a new take. Background noise can also fill a gap that sounds silent to you.

For a clean voiceover, treat automatic splits as suggested edit points, not a finished edit. Check each boundary with the full phrase in mind. If the client needs exact timing or natural pauses, keep more room around the speech and trim carefully after listening.

Your recording and edited audio stay on your device while you use the Audio Editor. Nothing you open or edit there is uploaded.