Trim the silence off the ends. Keep the middle untouched.
You hit record, waited 3 seconds, then started talking. Or the voice memo recorded an extra 4 seconds of room tone at the end before you tapped stop. "Top and tail" trimming just snips that — without auto-editing pauses inside the recording. Drop the file, click trim, download the clean version.
drop your audio file
WAV, MP3, M4A, OPUS — any audio. One file at a time.
When to use this vs other silence tools
- This page (top & tail) — cuts only the start/end silence. Middle of the file unchanged. Use for voice memos with awkward startup, podcasts with pre-roll gap, samples needing clean attack/release.
- Silence remover — finds and cuts every silent gap. Use when you want to auto-compress a long recording: lecture, interview, sermon. Removes pauses.
- Audio cutter — manual: you pick the in/out points. Use when you want exact control or you need to grab a specific section.
How the detection works
We scan from the start of the file forward until we find the first sample whose absolute value exceeds the threshold (e.g., -45 dB = 0.005 in linear amplitude). We back off 200 ms from there — that's the trim point at the start. Then we scan from the end backward to find the last sample above threshold; we add 300 ms after that for the end trim point.
The samples between those two points become the output file. Nothing in the middle is altered.
Good defaults for common recordings
- iPhone voice memo — −45 dB threshold. The handset mic has very low noise floor.
- Lapel mic recording — −45 dB. Lavs are clean.
- Phone speakerphone recording — −35 dB. More room noise.
- Zoom call recording (computer audio) — −45 dB. Good signal.
- Field recording with handheld recorder — −35 to −45 dB depending on location.
- Music sample with clean tail — −55 dB to preserve very quiet decay.
FAQ
What if the file has no silence at the start or end?
We won't trim anything. The output will be the same as the input (just re-encoded if you chose a different format).
Will it cut a quiet intro fade?
Only if the fade goes below the threshold for long enough. If you intended the fade, it'll be trimmed — pick a higher threshold (-55 dB) to preserve quieter content, or use our audio cutter for exact control.
How much silence is "enough" to be trimmed?
Anything below the threshold continuously, from the edge. The first sample above threshold (going inward from either end) is the new boundary.
Will the file length be exact, or rounded?
Trim points are at the sample — about 0.02 ms precision at 44.1 kHz. For all practical purposes, exact.
What if my recording is all silence?
We'll output the file unchanged with a warning. The threshold scan found no audio above the line; we won't truncate to zero length.
Does anything upload?
No. Detection and trimming run in your browser.