Trim audio and shorten silence while keeping natural phrasing

Audio trimming keeps the interval between a chosen start and end. It is the appropriate first step when you only want a chorus, a voice memo excerpt or the useful part between long lead-in and lead-out silence. The musical phrases and speaking pauses inside that excerpt stay together.

Silence removal scans for quiet regions throughout the recording, removes the selected regions and joins what remains. Detection uses an amplitude threshold and a minimum duration. Your listening supplies the editorial decision: which pause helps the sentence breathe, and which one can be shortened? Use the detector to find quiet gaps efficiently, then review them in context. Noise beneath speech is a separate cleanup task.

Tunenoodle starts at −45 dB threshold, 300 ms minimum duration and 50 ms padding. Raising the threshold to a less negative value, such as −35 dB, classifies more low-level material as quiet. Lowering it to −55 dB is more conservative. The level is derived from short audio windows, so boundaries are approximate.

Minimum duration is the qualifying length of a quiet run before padding. It is not the amount of pause retained. Padding preserves material at both ends of each detected run. Increasing it protects nearby breaths and consonants but removes less time. Interface ranges are −70 to −20 dB, 0–2000 ms minimum duration and 0–500 ms padding.

For an ideal detected quiet run from 10.00 to 11.00 seconds, the default 300 ms minimum qualifies it. Padding of 50 ms at each side leaves 10.00–10.05 and 10.95–11.00; the middle 0.90 seconds is removed. The remaining pause is about 0.10 seconds, not 0.30 seconds. Actual detected boundaries depend on the scan windows.

For a conversational recording, a useful starting experiment is 500 ms minimum duration and 100 ms padding. Treat this as an experiment for your recording and adjust from what you hear. Listen to a complete sentence before accepting it. For music, first try trimming only the ends; internal rests often define the rhythm.

1. Preserve the original. The silence remover accepts locally decoded files up to 20 MiB, 5 minutes and two channels. Analysis and export run in the browser.
2. Import the file and inspect the detected start, end and duration values. Listen to original and trimmed versions at similar volume.
3. Use the keep control on any detected region containing a meaningful pause, a quiet word or a fading note. A rescan resets these individual keep choices, so recheck them after changing parameters.
4. Review the whole edited sentence or phrase, then export. Removing time shifts later timestamps. If you are working with video or parallel tracks, apply matching timeline edits so the parts stay synchronized.

If consonants or breaths disappear, lower the threshold, increase padding or keep that region. If the conversation feels rushed, increase minimum duration and preserve pauses that separate ideas. If noise keeps all gaps above the threshold, raise it in small steps and review the quiet words as well as the gaps. Use a separate noise-cleanup workflow if the noise continues beneath speech.

If a piano or reverb tail is interrupted, keep that detected region or use a manual end trim. When comparing versions, the same time on each playhead may refer to different words because earlier gaps have been removed. Re-locate the same phrase by listening. For parallel tracks, use the same cuts in the shared editing timeline to maintain synchronization.

Both the trimmer and silence remover export 16-bit PCM WAV or MP3 encoded at 192 kbps. WAV avoids another lossy encoding step and is useful for further editing; MP3 is usually smaller for sharing. Choose the clearest source available for either format. WAV preserves the available detail for the next edit, while an MP3 export makes a compact delivery copy.

Reopen the download and check the first word, the last decay and several joins. Local trimming and downloading differ from saving a copy to the cloud; exporting an existing cloud song is subject to the account’s download entitlement. Keep an original alongside the edited copy, especially if subtitles or other recordings depend on its original timing.